Advances in Manufacturing Technology – XXII Volume I Acknowledgements The organizational arrangements for the conference have been undertaken by staff within the School of Engineering and Design, Brunel University. The edit ors are especially grateful for the help given by M s Carole Carr, Ms Tracey Bailey and Ms Sarah Sun. The editors also wish to thank the many referees an d the following members of COMEH and of the ICMR2008 Conference Committee for their assistance with paper reviewing. Professor R W Baines (De Montfort University, UK) Professor C Brecher (Fraunhofer Institute – IPT Aa chen, Germany) Professor K Case (Loughborough University, UK) Professor K Chen (Tsinghua University, China) Professor K Cheng (Brunel University, UK) Dr A Crispin (Manchester Metropolitan University, U K) Professor G Chryssolouris (University of Patras, Gr eece) Professor M L Culpepper (Massachusetts Institute of Technology, USA) Dr J Efstathiou (University of Oxford, UK) Professor A Forbes (NPL, UK) Professor D Harrison (Brunel University, UK) Professor D K Harrison (Glasgow Caledonian Universi ty, UK) Dr J Howes (Institution of Engineering and Technolo gy, UK) Professor G Huang (Hong Kong University, Hong Kong) Professor T Ito (University of Tokushima, Japan) Professor M J Jackson (Purdue University, USA) Professor T Kuriyagawa (University of Tohoku, Japan ) Professor WB Lee (Hong Kong Polytechnic, Hong Kong) Dr H Makatsoris (Brunel University, UK) Professor T Mileham (Bath University, UK) Dr M Morgan (Liverpool John Moores University, UK) Professor A Y C Nee (National University of Singapo re, Singapore) Dr L Newnes (Bath University, UK) Professor T Perer a (Sheffield Hallam University, UK) Professor J Ni (Michigan University, USA) Professor D Pham (Cardiff University, UK) Dr Y Qin (Strathclyde University, Scotland) Dr M Saadat (Birmingham University, UK) Dr W Schäfer (Fraunhofer Institute – IPA Stuttgart , Germany) Professor P Shore (Cranfield University, UK) Professor D Stockton (De Montfort University, UK) Dr F P Wardle (UPM Ltd, UK) Advances in Manufacturing Technology – XXII Proceedings of the 6th International Conference on Manufacturing Researc h (ICMR2008) Brunel University, UK 9th - 11th September 2008 Edited by Professor Kai Cheng, Dr. Harris Makatsoris and Professor David Harrison Volume I About the Editors Professor Kai Cheng holds the chair professorship in Manufacturing Sys tems at Brunel University. His current research interests focus on micro manufactu ring, design of precision machines and instruments, and digital manufacturing and enterprise technologies. Professor Cheng has published over 160 papers in le arned international journals and referred conferences, au thored/edited 5 books and contributed 6 book chapte rs. Professor Cheng is a fellow of the IET and a member of the IMechE and Euspen. He is the head of the Advanced Manufacturing and Enterprise Engineering ( AMEE) Department which incldes 9 academics and over 30 research assistants and PhD students. The d epartment is currently working on a number of resea rch projects funded by the EPSRC, EU Programs, KTP Prog rams, DTI, and the industry. Professor Cheng is the European editor of the International Journal of Adv anced Manufacturing Technology and a member of the editorial board of the International Journal of Mac hine Tools and Manufacture. Dr Harris Makatsoris is the research coordinator for the Advanced Manuf acturing and Enterprise Engineering (AMEE) subject area and Lecturer in man ufacturing and engineering systems at Brunel University. He is the co-founder and co-director of the Brunel UIRC “The London Institute for Enterpri se Performance, Sustainability and Systems”. He is als o Honorary Research Fellow at the Chemical Engineer ing Department of Imperial College and a member of Brun el’s Centre for the Analysis of Risk and Optimisati on Modelling Applications (CARISMA). He leads an inter disciplinary research team comprising engineers, an d physicists undertaking research in bottom up comput ational nanotechnology, evolvable process design, manufacturing and enterprise systems and robotics. His research is currently funded by the EPSRC resea rch council. His research interests include artificial intelligence, multi-scaled materials modelling, com puter aided engineering, control, automation and robotics. He i s a Chartered Engineer and a Member of IMechE. He h as a first degree in Mechanical Engineering from Imperia l College London. He also holds a PhD in Computer Aided Systems Engineering from the Mechanical Engin eering department of Imperial College. Following completion of his PhD he worked as a Research Assoc iate for three years in the same department. During that time he led a research team in a £6m EU project rel ating to the development of a pioneering distribute d optimisation and control system for ASIC manufactur ing and employed evolutionary programming optimisation technology. He has eleven years overal l work experience in academic R&D and also in commercial software product development in the area of artificial intelligence, systems modelling, optimisation and control. He has established a univ ersity spin out Software Company in the UK in which he is still the Technical Director. He was also involved in the set up of a semiconductor wafer recycling co mpany based in Germany. In addition he is the non executi ve director in a European IT services company. He h as authored 35 papers in journal publications, peer-re viewed conferences and book chapters and one book. Professor David Harrison is currently Head of Design within the School of En gineering and Design, and his research interests are in sustainable design and pr inted electronics. Over the past decade he has lead a number of research projects inspired by the goals of envir onmentally sensitive design. These projects include the application of offset lithographic printing to the manufacture of electronics. New manufacturing pr ocesses arising from this project have been successfully pa tented and licensed. He has also worked on the development of the concept of "Active Disassembly", where by features are designed into products to p ermit them to disassemble at end of product life, facilit ating recycling. He is a Director of a spin out co mpany, Active Fasteners, set up to commercialise this work . Other recent projects supervised include work on ecological footprinting of products, eco innovation , and tools to calculate ecologically optimum produ ct lifetimes. Preface The Consortium of UK University Manufacturing Engin eering Heads (COMEH) The Consortium is an independent body and was estab lished at a meeting held at Loughborough University on 17 February 1978. Its main aim is to promote manufa cturing engineering education, training, and resear ch. To achieve this the Consortium maintains a close liais on with those Government Departments and other bodi es concerned with the initial/continuing education and training of professional engineers, while also res ponding to appropriate consultative and discussion document s and other initiatives. It organizes and supports national manufacturing engineering education research confer ences and symposia. COMEH is represented on the Engineering Professors' Council (EPC). The Consorti um consists of heads of those university department s or sections whose first priority is to manufacturing e ngineering and who have a direct responsibility for running honours degree courses in the field of manufacturin g engineering. Currently there are about seventy me mbers of COMEH. COMEH decided in 1984 that a national forum was nee ded in which the latest research work in the field of manufacturing engineering and manufacturing managem ent could be disseminated and discussed. The result of this initiative was that an annual series of these national conferences on manufacturing research (NCM R) was started, the first NCMR being held at the Universit y of Nottingham in 1985. The first ICMR (ICMR 2003) built upon the NCMR series of conferences and was h eld at the University of Strathclyde in 2003. The subsequent NCMR/ICMR conferences have been held as follows: 1986 Napier 1987 Nottingham 1988 Sheffield 1989 Huddersfield 1990 Strathclyde 1991 Hatfield 1992 University of Central England 1993 Bath 1994 Loughborough 1995 De Montfort 1996 Bath 1997 Glasgow Caledonian 1998 Derby 1999 Bath 2000 University of East London 2001 Cardiff 2002 Leeds Metropolitan ICMR 2003 Strathclyde ICMR 2004 Sheffield Hallam ICMR 2005 Cranfield ICMR 2006 Liverpool John Moores ICMR 2007 De Montfort ICMR 2008 Brunel i Table of Contents Preface Volume I Keynote Speeches Re-manufacturing practices at Rolls-Royce Plc Justin Barrows The increasing importance of ultra precision surfaces Paul Shore EU FP7 NMP programs and next calls Hans Hartmann Pedersen Manufacturing research and development agenda at TSB Neil Morgan Session - Advanced Manufacturing Technologies High energy milling of micro magnetic powder for fingerprint development 3 K. Nag, X. Liu, A. Scott and Y. K. Chen An evaluation of heat partition models in high speed machining of AISI/SAE 4140 steel 13 F. Akbar, P. T. Mativenga and M. A. Sheikh Dry grinding of soft steel with ultrasonic vibrations assistance 23 T. Tawakoli and B. Azarhoushang Experimental investigation of damages on machined surfaces of GFRP pipes 31 A. N. Sait, S.Aravindan and A. N. Haq A comparison between process 3D tolerance stack-up and tolerance chart 39 A. D. Prete, D. Mazzotta, G. Ramunni and A. Anglani A novel semibonded abrasive plate for finishing advanced ceramics 47 Y. Julong, L. Binghai and W. Zhiwei Design of angular position controller of fluid drive spindle for diamond turning 57 Y. Nakao and M. Ishikawa Mechanisms for grinding and polishing of silicon carbide with loose abrasive sub-aperture tools 65 H. Cheng, L. Ren, Y. Feng, Y. Yam, H. Tong and Y. Wang Technical advances in ultra-precision machining of freeform optics by function-oriented testing 73 C. Brecher, R. Schmitt, D. Köllmann, and M. Merz Optimization of surface finish and kerf taper in design of experiments based abrasive water jet cutting of aramid composites 81 T. U. Siddiqui and M. Shukla New methods for enhanced study of the electrochemical machining process 91 L. Slătineanu, M. Coteaţă, A. Drăghici, O. Dodun and I. Neaga ii Use of precision measurements of evolved geometrical deviations as a diagnostic tool for air-cooled diesel engine cylinder 101 S. H. R. Ali, H. H. Dadoura and M. K. Bedewy Augmented reality design methodology for computer numerical control machinery 113 W. A. Khan Optimisation of resin flow in a flexible mould infusion process for large pressure vessels 123 M. Gascons, N. Blanco and K. Matthys Experimental analysis of properties of materials for rapid prototyping 133 M. Šercer, P. Raos and A. Pilipović Session - Advanced Manufacturing Systems Advanced productization reference model 145 J. Kantola, A. Tuominen and H. Vanharanta Production scheduling in the corrugated industry: Challenges and solutions 153 B. Iba, E. Shehab and A. Swindells An industrial implementation of a methodology for the 161 optimisation of product design tolerances R. Eadie, J. Gao Industrial robotics in the lean enterprise – A case study 171 M. Hedelind and M. Jackson An assembly line information system study 181 K. Case, G. Bäckstrand, D. Högberg, P. Thorvald and L. J. D. Vin Analysis and optimisation of assembly lines feeding policies 189 A. C. Caputo and P. M. Pelagagge Decision support system for pallet utilization in the sheet feeder industry 199 E. Shehab, O. Celaya and A. Swindells Scheduling of non-repetitive lean manufacturing systems under uncertainty using intelligent agent simulation and decision support 207 T. C. Papadopoulou and A. Mousavi Safety culture change in a high-risk industry 217 T. Halima Application of the delay-time concept in a manufacturing industry 227 B. Jones, I. Jenkinson and J. Wang Model based design of economy of scope and scale production systems 235 Z. Cui and R. H. Weston Moving up the value chain of cocoa: The role of fair trade 243 L. D. Mensah, D. Julien and R. Greenough People development system to enhance problem solving capabilities in implementing lean process management: A case study in aerospace company 253 A. P. Puvanasvaran, M. H. M. A. Megat, S. H.Tang, M.Y. Rosnah, M. R. Muhamad, and A. M. S. Hamouda iii Model driven, change capable production systems 265 R. H. Weston, A. Rahimifard, J. Ajaefobi and Z. Cui An integrated methodology for strategic selection problems 273 S. M. Ali Khatami Firouzabadi An IT perspective of lean 279 Z. Badak, S. A. Abbas and C. Herron Applying modularity-in-production and lean production in a terry weaving machine assembly environment 289 S. Loke and A. Engelschenschilt Matching human resources to process-oriented roles in manufacturing companies 299 F. Liu and R. Weston Session - Extended Manufacturing Enterprises: Systems and Tools Multi agent system for negotiation in supply chain management 311 S. Saberi and C. Makatsoris The study of influrial factors for developing industrial clustering in Thailand 319 S. Komolavanij, M. Tsuji, Y. Ueki, C. Jeenanunta and V. Ammarapala Human systems modeling in support of manufacturing enterprise activities 327 S. N. Khalil and R. H. Weston The impact of industry group culture on the development of opportunism in 335 cross-sector alliances S. Grant Interrelations of development needs between innovation competence and innovation culture 345 A. Suominen, J. Jussila, P. Porkka, and H. Vanharanta WSN based intelligent cold chain management 353 W. Fu, Y. S. Chang, M. M. Aung, C. Makatsoris and C. H. Oh Linking supply chain management capability and manufacturing operations competence with organizational performance: A case study of Thai industries 361 N. Chiadamrong and N. Suppakitjarak Exploring the implementation of integrated and discrete IT systems – A framework for analysis 371 C. Burns and N. Hewitt-Dundas Characterisation of SME migration toward advanced technologies 383 R. A. Barton and A. J. Thomas Enterprise modelling in support of methods based engineering: Lean implementation in an SME 391 B. M. Wahid, C. Ding, J. O. Ajaefobi, K. Agyapong-Kodua, T. Masood, and R. H. Weston Low carbon manufacturing: Characterization, theoretical models and implementation 403 S. Tridech and K. Cheng iv Session - Micro/Nano and Precision Manufacturing Modular laser integration into machine tools 415 C. Brecher, F. Klocke, M. Emonts and J. Frank Laser process monitoring: A critical review 425 A. Stournaras, P. Stavropoulos, K. Salonitis and G. Chryssolouris A novel architecture for a reconfigurable micro machining cell 437 R. Al-Sharif, C. Makatsoris and S. Sadik Impact of heat transfer on femtosecond laser micromachining 447 H. Song, E. Miao and Y. L. Tu Evaluation of critical parameters in micro machining of hardened tool steel 457 A. Aramcharoen and P. T. Mativenga Grinding knowledge sharing through a knowledge warehouse development 465 A. Alabed and X. Chen Modelling and simulation of the dynamic cutting process and surface topography generation in nano-micro cutting 473 L. Zhou and K. Cheng On the design of a monitoring system for desktop micro-milling machines 483 P. Stavropoulos, A. Stournaras and G. Chryssolouris Investigation of feeding devices and development of design considerations for a new feeder for micro sheet forming 493 A. Razali, Y. Qin, C. Harrison and A. Brockett Complexity in engineering design and micro devices: A review of literature 505 K. Panikowska, A. Tiwari and J. R. Alcock Effect of pulse energy in micro electric discharge machining of micro holes 513 T. T. Öpöz, B. Ekmekci and A. Erden Micro-drilling holes on glass wafer for fabrication of MEMS 521 X. Luo, W. Chang, J. Sun, N. Ding and C. Mack Volume II Session - Product-Service Systems (PSS) Standardization within the service creation process in value generating networks 531 H. Meier, R. Krings and U. Kaiser Differences in forecasting approaches between product firms and product-service systems (PSS) 539 R. De Coster Product service systems in the automotive industry: An alternative business model for a sustainable system 549 C. Vezzoli and F. Ceschin v Challenges of product-service systems: A case study 557 E. Shehab, S. Evans, T. Baines, H. Lightfoot, A. Tiwari, M. Johnson and J. Peppard Supply network integration in complex product services 563 J. S. Srai and M. Gregory Design of generic modular reconfigurable platforms (GMRPs) for a product-oriented micro manufacturing systems 571 X. Sun and K. Cheng Design of modular service architectures for product service systems 581 H. Meier and K. Sadek Applying semantic web services to enterprise web 589 Y. Hu, Q. Yang, X. Sun and P. Wei VR-based approach to assembly and disassembly for large-scale complex products 597 P. J. Xia, Y. X. Yao, W. Y. Tang, Y. D. Lang and P. Chen Service attribute importance and strategic planning: An empirical study 611 V. Pezeshki and A. Mousavi Session - Design and Manufacturing Simulations Numerical simulation of reaction injection molding for refrigerator insulation 621 A. Stournaras, K. Salonitis, and G. Chryssolouris Modelling and simulation of grinding fluid nozzles 631 V. A. Baines-Jones, M. N. Morgan A. D. Batako and E. Brown Tool path optimization using graph algorithms 639 A. J. Crispin Development of a new learning technique for discrete event simulation by utilising previous software experience 647 A. Guerrero, J. Darlington, R. Weston, K. Case and R. Harrison Molecular dynamics simulations for nanomanufacturing processes: A critical review 655 P. Stavropoulos, K. Salonitis and G. Chryssolouris Finite element simulation of laser welding process of T-joint specimens 665 N. Siva Shanmugam, G. Buvanashekaran and K. Sankaranarayanasamy A sensitivity study of the effects of interface heat transfer coefficient on FE modeling of machining process for a wide range of cutting speeds 675 S. A. Iqbal, P. T. Mativenga and M. A. Sheikh Experimental studies and finite element simulation of A1-Fe composites during cold upsetting 683 T. Ramesh and R. Narayanasamy Design and development of MIAB welding module-parametric investigation and validation of electromagentic force and magnetic flux distribution using finite element analysis 693 S. Arungalai Vendan, S. Manoharan, G. Buvanashekaran and C. Nagamani Design and finite element mode analysis of noncircular gears 703 C. Lin, K. Cheng, D. Qin, C. Zhu, H. Qiu and X. Ran vi Session - E-Manufacture RFID-enabled information infrastructure for real-time digital manufacturing 713 Y. Zhang, G. Q. Huang, X. Chen, P. Jiang and F. J. Xu A software concept for planning in non-hierarchical production networks 723 S. Horbach and E. Müller Automatic identification of semantic relationships for manufacturing information management 731 W. Yu and Y. Liu An investigation on 3D data mining over the web and web-based CAD data management 739 L. Li, D. K. Harrison and J. S. Lynn Simultaneous scheduling of machines and automated guided vehicles using artificial immune system 749 A. Noorul Haq, A. Gnanavelbabu, J. Jerald, P. Asokan and K. Narendar Session - Manufacturing Supply Chains A comparison of supply chain effectiveness using the "quick scan" audit methodology 761 A. J. Thomas, P. Childerhouse, D. Towill and G. Phillips Business process value analysis using an analytical hierarchical process 771 S. Rashid, K. Agyapong-Kodua and R. H Weston Multi-criteria decision making in supplier selection using VIKOR and ELECTRE methods 779 P. Parthiban, Y. R. K. Chowdary, M. Punniyamoorthy and K. Ganesh Productivity enhancement in a manufacturing enterprise by improving management processes 793 S. Rashid, K. Agyapong-Kodua and R.H Weston The inclusion of shelf-life effects into IOBPCS inventory models 799 A. S. White and M. Censlive Session - Cost Engineering Improving the sales demand forecasting process 809 S. Miller, R. Khalil and D. Stockton Manufacturing cost modeling for technology assessment 817 Y. Xu, J. Wang, X. Tan, R. Curran, S. Raghunathan, J. Doherty and D. Gore Enhanced process costing methodology for dynamic manufacturing enterprises 825 K. Agyapong-Kodua, B. Wahid and R. H. Weston Last order costing (LOC) - an iterative methodology for the continuous improvement of manufacturing costs 833 A. Davies, E. John, A. J. Thomas and B. Worrall Lean tools and principles in the manufacturing cost modelling process 839 Y. Delgado Arvelo and D. J. Stockton vii Session - Computer Aided Engineering (CAE) Simulation support of lean layout considerations for new products: A case from large scale products 849 R. Darlington, B. Wahid, R. Weston, K. Case and R. Harrison Model development for turning process 859 I. Piotrowska, C. Brandt, H. R. Karimi and P. Maass Computer workstation evaluation approach by neural network (NN) 867 N. Phaoharuhansa, K. Krishnamra, S. Nanthavanij, and W. Kongprawechnon Optimal decisions in product modularity design using real options approach 873 Y. Wang, M. Dong and D. Yang Dielectric curing of web material – A continuous manufacturing process 883 A. M. Hasna Microcellular injection molding: Review and limitation of CAE application 899 P. Y. Chung, K. H. Lau and C. Y. Yip Develop a neural-network modelling tool for engine performance tested data modeling 909 M. H. Wu, W. Lin and S. Y. Duan Effects of the pipe-joints on acoustic emission wave propagation velocity 917 W. Wichaidit and Y. H. Joe Au Investigation of design and manufacture of cellular-type structures for injection moulding tools 923 M. Dotcheva, H. Millward and R. Bibb Process optimization and metal flow analysis of direct indirect extrusion of aluminum using FEM simulation 933 L. Niu, T. Sheppard and X. Velay Session - Product Life Cycle Management Development of a consumer end-of-life vehicle database system 943 A. Sweetman and Q. Wang Investigating factors affecting ERP implementation in made to order SME sector – An experimental account 951 A. Deep, P. Guttridge, S. Dani and N. Burns Decomposition of automotive manufacturing machines through a mechanism taxonomy within a product lifecycle management framework 961 J. F. Darlington, R. Darlington, R. Harrison, L. Lee, K. Case and R. H. Weston A comparison of consultancy-assisted and in-house new product development strategies within a small medical device company 969 P. Chapman, I. Corp, H. Millward and J. Stephens A knowledge-based collaborative process planning system for automotive mould manufacturing 977 K. L. Choy, Y. K. Leung*, C. K. Kwong and T. C. Poon Advanced Manufacturing Technologies 1 2 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT HIGH ENERG Y MILLING OF MICRO MAGNETIC POWDER FOR FINGERP RINT DEVELOPMENT K. Nag 1 , X. Liu 1 , A. Scott 2 , Y. K. Chen 3 1 . School of Computing, Engineering and Physical Scien ces, University of Central Lancashire, UK 2 . School of Forensic and Investigative Sciences, Univ ersity of Central Lancashire, UK 3 . School of Aerospace, Automotive and Design Engineer ing, University of Hertfordshire, UK Abstract Highly reflective magnetic powders have been available commercial ly for latent fingerprint development on dark background surfaces for a few years, however there are needs of a superior darker variety of the magnetic powder w hich would be ideally suitable for obtaining good contrast on light background surfaces. A novel dark magnetic powder is therefore suggested for the application in latent fingerprint development on light backgrounds. Based on a comprehensive analysis of the manufacturing techniques for the production of metallic powd ers and previous experiences, a series of dry milling trials were proposed us ing a high energy vibratory mill. In these milling trials, atomized iron powders of approp riate particle dimensions were chosen as the starting material. The starting atomized ir on powders are mixed with a specific process controlling agent in cylindrical plas tic pots, which are fixed on the vibratory mill. Stainless steel balls were used as the milling medium. The amount of starting iron powders and the milling time are designated so as to obtain flakes of different particle dimensions. After the designated mill ing process, the powders were carefully exposed to the air to avoid catching fire and separated from the steel balls. Thereafter, the samples of the powders were inspected us ing scan electron microscope and the result reveals that some good quality flakes with leafy characteristics and high aspect ratio were developed. The samples of the powders were also used to develop latent fingerprints on a common background at the university’s forensic investigation laboratory, and the result indicates that some flakes obtained c an be considered as good quality darker variety of magnetic powder for fingerprint development. Keywords: High Energy Milling, Magnetic Iron Flakes , Fingerprint Powder. 3 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT 1.0 Introduction Metallic flake powders have been widely used for fi ngerprint detection. In the UK there are several va rieties of fingerprint powders available from different forens ic suppliers, although very few of these powders ar e actually manufactured solely for the purpose of fin gerprint detection. Typical examples of flake powde rs which are used extensively for fingerprint detectio n are aluminum powder, brass powder and Magneta Fla ke (a commercial variety of magnetic flake). Highly refle ctive aluminum powder has been the most widely used powder for fingerprint detection in crime scenes in vestigation and is generally employed for bright fi ngerprint development on dark backgrounds (most effective on glass surface). However with the introduction of Magneta Flake in the last decade, high density magn etic flake (e.g. iron) when applied on crime scenes , has been able to overcome the problem of air-borne dust level typically associated with application of low density aluminum flake when applied with standard squirrel brush [1]. Also because of its low density, aluminum powder tends to take some time to settle during app lication. Furthermore, the magnetic flakes are appl ied on the crime scene with a magnetic applicator thereby leading to somewhat non destructive fingerprint det ection. Magneta Flake thus, produced a suitable alternative and as with aluminium flake it is found to be extr emely reflective and ideally suitable for dark background s. Recent home office evaluation of various fingerp rint powders suggest that Magneta Flake actually outperf orms aluminum on lot of surfaces like painted metal , gloss painted wood and others and closely matches t he performance of aluminum on glass surfaces [2]. W hile Magneta Flake has been successfully used for bright fingerprint development on darker backgrounds, the re is not a suitable powder obtained commercially which c ould be as effective as Magneta Flake on lighter backgrounds. Further, Home Office report suggests t hat one of the black magnetic powder varieties obta ined commercially is actually a mixture of two particles , one large magnetic carrier particles of iron (20- 200 µ m) and the other smaller non magnetic particles of ir on oxide (3-12 µ m) which actually develops the mark in adhering to the fingerprint residue. It is however, the least used powder in crime scene s investigations, and is not very effective on most surfaces as compared to that of aluminum powder and Magneta Flake. Previous research with different metal flakes suggests that iron flake with diameters 10-25 µ m wo uld allow print development to a quality considerably superior to t hat of other commercial black and magnetic black po wders, when applied by a magnetic applicator on light back grounds [3]. However, manufacturing of such powders had found little success over the years. The aim of the present study is therefore to develop a superi or quality of dark iron magnetic flake fingerprint powder suit ed for lighter backgrounds on a wide range of surfa ce textures. A set of initial experiments have therefo re been conducted to develop the darker variety of magnetic flakes and subsequently been analyzed for its suita bility as a fingerprint powder. 2.0 Factors Governing the Development of Dark Flake for Forensic Applications Although most of the commercially available metal f lakes including aluminum flakes are usually produce d by rotary ball milling in tonnage quantities , various high energy milling devices are often utili sed for rapid production of trial quantities of metal powders for experimentation. They differ mostly in terms of th eir design and modes of operation. However, the fundamentals o f change in particle morphology of ductile metal po wder during high energy milling is same for all the devi ces and can be attributed to a combination of phas es, i.e. micro-forging, fracture, agglomeration and de-agglomeration, all of them can take place simultaneously in a mill [4]. It is therefore important that periodic samples are drawn at differ ent stages of a milling experiment and analysed in order to identify the different mil ling phases. 4 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT 2.1 Role of Additives as A Process Controlling Agent (PCAs) Although a large number of solid or liquid additive s are used in practice in different milling applica tions, stearic acid (CH 3 (CH 2 ) 16 COOH) is the most common PCA used in metal flake p igments for both laboratory purposes and industrial applications. The most impo rtant function of stearic acid as an additive is es sentially to lubricate the powder particles, thereby helping the process of microforging of flakes. In absence of s tearic acid, frictional forces prevent particles sliding o ver each other and a considerably large number of p articles will be involved in the impact between the balls. W ith addition of stearic acid, it allows particles t o flow over one another resulting in fewer particles coming und er the entrapment zone. Thus fewer particles will r eceive the impact energy and will have a greater strain in crement, thereby initiating microforging. Further, studies have indicated that milling of aluminum under oxyge n without stearic acid has resulted in the powder remaining in the granular form and showed a slight increase in apparent density. With the usage of stearic acid there was a substantial fall in both in apparent de nsity and oxygen pressure, indicating that flakes h ave been formed with high surface area of aluminum reacting with oxygen. This has been explained by the mechani sm of interaction between stearic acid and the metal i n formation of the flakes. Essentially a metal stea rate is formed in which metal atom is bound in the surface and the stearate atoms are roughly oriented perpend icular to the surface. Further, layers of free stearic aci d may be formed above this layer which provides res istance to cold welding of the metal particles thereby inhibit ing agglomeration. Another important role of steari c acid in production of fingerprint powder is that the thin c oat of stearic acid remaining on the flake surfaces helps in the adhesion of flakes to a latent fingerprint depo sit. Previous studies have shown that removal of st earic acid coating by solution in a suitable solvent (usually soxhlet in hot acetone) seriously reduces the effec tiveness of flakes for fingerprint development [5]. It is to be mentioned that in previous studies with fingerprint powders, a variety of alternate organic coatings have been exp erimented with, particularly with substances which are secreted in sweat and/or sebum present in latent fi ngerprint residues (e.g. tripalmitin, tristearin, s qualene) [6] but none of them seemed to match the quality achiev ed by stearic acid. 2.2 Choice between Wet and Dry Milling Two types of milling environment can be defined bas ed on the formation of surface films on the metal powders, reactive and non reactive milling. In reac tion milling the powder surface reacts extensively in the fluid to produce surface films which inhibits agglo meration by welding. This would result in the mille d powder being extremely fine. In non reactive milling, the powder particles hardly react with the milling flui d, thereby bare metal surfaces are formed which enhances the w eldment of the powder particles. Metal powders mill ed in organic or inorganic fluids retain small amounts of the fluid dispersed throughout each particle. Thus , hydrocarbons containing hydrogen and carbon and car bohydrates containing hydrogen, carbon, and oxygen are likely to introduce carbon and/or oxygen into t he particle. In essence, in wet milling a milling l iquid can interact with the metal particle and influence in t he same way a gaseous medium would interact in dry grinding. Normally when the final product is dry, d ry milling is preferred as in the case of producing metal flakes for most industrial applications. Wet millin g of metal particles is considered only when flakes fail to form by conventional dry milling route. It is there fore suggested that dark fingerprint powder be prod uced by dry milling, wet milling with suitable liquid can o nly be considered if the dry milling route cannot y ield any positive results. 5 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT 2.3 Colour and Visual Quality of the Powder One of the important criteria of developing a finge rprint powder is to understand how we can obtain th e desired colour of the powder. The colour of the met al flake can be governed by two factors, the basic colour of the starting material used and the presence of surf ace films on the metal surface. The basis colour of the metal surfaces arises out of the interaction of the elect ric field of light waves with conduction electrons of the metal; the detailed phenomenon is not discussed here. Esse ntially it can be said that to obtain a dark finger print powder, the colour of the starting material should not be bright and preferably darker in appearance. However, during the milling operation the starting material may undergo variety of shape changes at different s tages of milling and would therefore exhibit different colou rs due to either specular or diffuse reflection of light. It is expected that a relatively equiaxial particle will reflect light in a diffuse manner where as flat fla kes will tend to be glossy due to the specular reflection. Furthe r the flakes should be deposited parallel to each o ther (leafing)[7], if not, the flakes will scatter the l ight by overall diffuse reflection and specular ref lection will take place over very small distances of the order of the flake diameter, which would result in the loose fl akes appear to sparkle. Further, although flat parallel flake surfaces would give specular reflection, the edges/perimeter of the flake will still be radiatin g at different angles (diffuse reflection). Thus wi th decreasing flake diameter the edge effect will tend to predomi nate and there will be more diffused reflection. It is therefore expected that to obtain dark fingerprint powder, the flake diameter has to be small, althoug h optimum size has to be determined based on the qual ity of developed fingerprints. Another important fa ctor which governs the colour of the final product is th e surface chemistry of the flakes i.e. either a pre sence of oxide layer, a layer of metal stearate or free stea ric acid; all of which are encountered in the milli ng process. Thickness of the oxide layer can influence the colo ur of the product particularly when the milling env ironment is non inert, i.e. in presence of air. The level of the metal stearate and/or free stearic acid could also influence the flake colour as interference effect could be pr oduced if the layers are thick enough. 3.0 Experimental Procedure The choice of an appropriate starting material has been determined based on previous studies [8], wher e it has been observed that very fine powder like iron carbo nyl when milled by standard dry grinding method, fo rms a solid mass by agglomeration and so wet milling in a suitable liquid had to be introduced for the forma tion of flakes. Since the study also suggested that small p article size of the starting powder is beneficial t o the production of flakes by high energy milling, in the present study, atomized iron powder manufactured b y Sigma Alldrich, of approx average particle size of 15 µ m has been selected as the starting material. A lso the powder selected is irregular in shape as flakes pro duced from irregular shaped powder tend to produce dark flakes as compared to spherical particles which pro duce highly reflective flakes like that of Magneta Flake. A set of experiments have been carried out with diffe rent amount of starting material in a prototype vib ratory mill at 3000 rpm, 50 Hz. The starting material alon g with 2-4 wt% of stearic acid as process controlli ng agent are charged in plastic containers and milled interm ittently for a length of time with the aid of 7mm s tainless steel balls. The milling time, the amount of starti ng material varied during the dry milling process t o obtain flakes of different particle dimensions. Iron flake powder of 1g during the milling was sampled and characterized, subsequently analyzed by scanning el ectron microscope. Remaining powder was utilised fo r developing latent fingerprint development. Since th e quality of the print can vary depending on the le vel of fingerprint residue, a standard procedure was there fore adopted to obtain a set of virtually identical 6 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT fingerprints. A single donor rubbed his hand to dis tribute sweat over his fingers before pressing all his fingers on a white piece of paper. The same process was rep eated for deposition of all the prints. The fingerp rints were then developed for each of the powder samples using a magnetic applicator and subsequently they w ere scanned directly in the university’s forensic depar tment with the aid of ‘Livescan’, a device used for AFIS (Automated Fingerprint Identification System) in cr ime scene investigation. 4.0 Results and Discussion 4.1 Milling Behavior of the Iron Powder and Feasibility of the Dry Milling Proce ss The milling of the powder has been carried out in a prototype vibratory mill which can generate great impact forces due to the rapid vibration of the motor thus resulting in faster and finer grinding. The starti ng atomized powder has been milled for a maximum of 9 hrs and i t has been observed that flakes of 25-30 µ m have be en obtained after milling for at least 4 hrs, dependin g on the amount of starting material used. Fig 1 sh ows the scanning electron microscopic image of the starting material which during the course of milling has un dergone a combination of milling phases, primarily microfor ging and fracture. The irregularity of the particle shape of the resultant flakes indicate that the starting irr egularly shaped particles are actually compressed i n a non uniform manner which has resulted in flakes with co mplex surface contours and jagged outlines. Fig. 1. Scanning Electron Microscopic image of the starting atomized iron powder manufactured by Sigma Alldrich During the experimental study it has been observed that the dry milling process in producing flakes is feasible only when the process is carried out with some care ful considerations. The powder flakes exhibited pyr ophoric behavior and tend to ignite spontaneously when bein g exposed to atmosphere. The milling containers, ma de of thin plastic did not heat up substantially during t he milling process and the powders only starts burn ing once the milling vials are opened and the flakes are bei ng exposed to atmosphere. The burning is in the for m of spatters, starting from few powder particles, then progressively spreads to other powder particles to form a lump at the end, unless they are interrupted in bet ween and separated. However, it has been observed t hat with controlled exposure of powder to the atmosphere by slow bleeding of air into the milling vials at the end of the milling process and carefully separating the powder from the balls, the powder tends not to burn in atmospheric conditions. It is therefore suggested t hat to negate the effect of oxidation of the powder particles, dry milling should be carried out in an inert atmos phere with the addition of 2-5% of oxygen to allow careful oxidation within narrow limits. This would enable t he formation of protective oxide films on the flake s thereby reducing the pyrophoric behaviour of the powder par ticles. 7 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT 4.2 Effect of Milling Time and the Amount of Starting Material Used Two sets of experiments were carried out with diffe rent amount of starting material; the first set of experiment with 5g of powder and the second experiment with am ount varying between 15-30g. Milling time has been varied between 4-9 hrs in order to achieve flakes o f different dimensions. Typically, flakes of averag e particle dimensions of 25-30 µ m and 10 µ m have been obtained as previous studies have suggested that such flake dimensions can be considered as optimum for latent fingerprint development. Fig. 2. Scanning Electron Microscopic image of dark iron flakes when starting material of quantity (a) 5g is milled for 4hrs (b) 5g is milled for 8 hrs and (c) 30g is mill ed for 8hrs and (d) graph showing the effect of mil ling time and the amount of starting material used (5g and 30g with s imilar ball to powder weight ratio) on the mean fla ke dimensions. It has been observed that with 5g of starting mater ial, dry milling for 4 hrs have resulted in flakes of average particle size of 25-30 µ m. Figure 2a shows a scan e lectron microscopic image of a typical sample of da rk powder from the 1 st set of experiments, after being milled for 4 hours . The particles as observed in Scan Electron Microscope (SEM), show good flaky characte ristic, but is not uniform throughout. It showed th e presence of bigger particles which have not been mi lled properly. However, more consistent flake sizes have been obtained when a sample of the resultant flakes has undergone sieving operation and +38 µ m of the fraction is being discarded. A different sample, u nder identical set of milling condition having unde rgone milling for 4 hours is being milled further for ano ther 15mins with the aid of glass beads to achieve more uniform flakes. However, it did not show any marked difference in the surface characteristics of the f lake. Further milling of powder flakes up to 8hrs has res ulted in achieving flake dimensions <10 µ m. Figure 2b shows a scan electron microscopic image of a typica l sample of dark powder from the 1 st set of experiments, having undergone milling for 8 hours. SEM analysis of the powder reveals consistent flakes have been obtained with very good leafy structures and with r elatively uniform distribution. (a) (b) (c) (d) 5g powder 30g powder 8 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT In the initial stage of the milling process, the particles are compressed via repeated impact of the milling medium which results in a deformation (flattening) of the particle. Thus good quality flakes have been obtain ed of required particle dimension after milling for 4hrs. Further milling of the flakes has resulted in increased fra cture of the particles due to the initiation and propagation of cracks across the particle. The cracks may occur du e to work hardening, ductile rupture as the flake becomes thi nner in some parts than in others, fatigue failure, fragmentation, initial defects and inclusions in particle, or a co mbination of these effects. This has resulted in de creasing the particle dimension to <10 µ m after 8 hrs of dry milling. During the milli ng process, it has been observed that the amount of starting material has pronounced effe ct on the milling pattern. While 25-30 µ m particle size has been obtained by milling 5g of powder for 4hrs, alm ost 8 hrs of milling was required for 30g of powder to achieve the same results. Figure 2c shows a scan el ectron microscopic image of a sample of dark powder from the 2 nd set of experiments, where 30 g of powder has under gone 8 hrs of milling. The sample after being mille d for 4 hrs did not sufficiently flatten the powders to form flakes although there has been a noticeable increase in diameter to thickness ratio. With further milling o f the powder, it shows the tendency of microforging and subsequent initiation of cracks until an optimum si ze 25-30 µ m flakes have been obtained after 8 hrs o f milling. Fig 2d shows the variation in mean powder size of two samples with respect to the milling tim e for 5g and 30g of starting material used. Other samples wi th 15-25 g of powder also took much longer than 4 h rs of milling to obtain the required flake dimensions. Th ese results clearly demonstrate that as the quantit y of the starting powder charged into the milling vials are decreased, the processes of microforging, fracture and agglomeration occurred more rapidly. In all the cas es, milling process was not carried out beyond 9 ho urs as it has been decided that total production time should not exceed 10 hours (including set up time, time fo r charging and discharging etc.) to make the process viable for rapid production of metal flakes on a la boratory scale. As a result, it was not possible to obtain f lakes of 10 µ m variety with 20-30 g of powder. It i s however, expected that longer milling time up to 16 hrs woul d be required to obtain 10 µ m variety of flakes un der the present experimental set up. The overall milling ti me can however be reduced by changing and optimisin g the process parameters like increasing the efficiency o f the vibratory mill, optimising the ball to powder ratio, size of the balls and quantity of the stearic acid used. The slight variation of the stearic acid level bet ween 2-5 wt% under present experimental conditions however, had negligible effect on the milling time as well as th e quality of flakes produced. 4.3 Comparative Analysis with Magneta Flake & Suitability for Latent Fingerpri nt Development Samples from two set of experiments have been visua lly inspected and digitally photographed to compare the coloration and appearance of the powder which would be an important factor when powders are to be appl ied on common backgrounds to test the suitability for l atent fingerprint development. It has been observed that flakes of particle dimension less than 10 µ m is muc h darker in appearance when compared to that of 25- 30 µ m flakes. This can be attributed to the fact that due to the smaller average particle size of sample, ed ge effect is more predominant which contributed to more diffused reflection of light, as explained earlier. A sampl e of commercially available Magneta Flake has also been analysed in SEM and a comparative analysis has been made with the dark powder produced, as briefed in F ig 3. Elementary analysis of the powders as well as the starting material, reveals that the dark powder has got higher oxygen content compared to that of Magn eta Flake, where it is almost negligible. This indicate s that the dark powder has undergone slight oxidati on during the course of dry milling as the starting material has found to be of pure iron without presence of an y oxides. It 9 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT is further corroborated from the fact that during t he experiments, it was difficult to open the covers of the plastic vials indicating that there is a pressure d rop inside as a result of the slight oxidation. Particle Magneta Flake Dark Powder Composition Iron flake Iron flake Process Wet Milling Dry Milling Description Flat structure, smooth surface Highly reflective Rough surface with jagged edges Dark appearance Distribution Uniform More or less uniform Grouping Thickness Moderate Thin Moderate Extremely thin Particle Diameter 10-15µm 25-35 µm, <10µm Visual Characteristics for a lump of powder Typical Scanning Electron Microscopic Image Fig. 3. Comparative analysis between dark flake pow der and bright magnetic flakes (Magneta Flake) A set of samples from both experiments have been ut ilised for developing latent fingerprints on a whit e background following a standard procedure to obtain identical fingerprints. Since the magnetic flakes are specifically useful for latent print development on porous surfaces such as wall, paper and polythene [9]; 80gsm A4 sized white papers have been considered fo r impression and development of latent fingerprints . A commercial variety of black magnetic powder has als o been included in the analysis to identify the qua lity of fingerprint developed with respect to that achieved by the dark flakes. Analysis of the developed prin ts reveal that good quality of prints has been obtained using both the 25-30 µ m and 10 µ m variety of the flakes. Fig 4. shows the scanned images of the latent fingerprint developed by the three types of powders considered. It is evident that the best quality of developed prints h as been obtained by 25-30 µ m variety of the flakes with good ridge details and reasonably good contrast. It has been observed that 10 µ m variety of the flakes also allowed reasonably good quality of print development with m uch darker appearance producing a better contrast a gainst a white background. However, this fine flake tends to paint over the surface of the paper, thereby som ewhat reducing the quality of the overall print developed . In either case, the print developed on white pape r background has found to be much superior to that ob tained by the commercially available magnetic powde r. Further analysis by the Automated Fingerprint Ident ification System (AFIS) in the university’s forensi c department also conform the findings. It is to be m entioned that the print quality did vary slightly f rom sample to sample indicating that best fingerprint quality can be achieved by optimising the flake characteris tics and 10 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 HIGH ENERGY MILLING OF MICRO MAGNETIC POWDER FO R FIN GERP RINT DEVELOPMENT particle dimensions. Further, it is also suggested that similar experiments can be conducted with othe r types of iron/steel powders as starting materials and find t heir potential as magnetic flakes for fingerprint d evelopment. Fig. 4. Quality of fingerprints developed under ide ntical set of conditions on a white paper using a m agnetic applicator by (a) commercially available magnetic fingerprint powder (b) 25-30 µm dark iron flakes and (c) 10 µm dark iron flakes 5.0 Conclusions High Energy Milling devices, particularly a vibrato ry mill can be used for rapid production of micro m agnetic metal flakes which has potential applications in fi ngerprint technology. The magnetic properties of ir on flake together with the flake characteristics achieved by dry milling can be utilised for developing a suita ble darker variety of magnetic flakes for latent fingerprint d evelopment on light backgrounds. Analysis of these flakes produced from a set of initial experiments reveal t hat print developed can easily surpass the quality achieved by commercially available dark magnetic powders on porous surfaces. The quality of the flakes can be f urther enhanced by optimising the dry milling parameters i ncluding the amount of starting material, milling t ime and other criteria like size of milling medium, amount of process controlling agents etc. Further, other v arieties of dark magnetic flakes can be produced from different starting material which can be suitable for develo ping latent fingerprints on a wide range of background s urfaces. References [1] J. D. James, C. A. Pounds, and B. Wilshire, “Magnet ic flake fingerprint technology,” Journal of Forensic Identification, Vol. 41, No. 4, 1991, pp 237-247. [2] Helen L. Bandey, Andrew P Gibson, “The powders proc ess, study 2: evaluation of fingerprint powders on smooth surfaces,” Fingerprint Development and Imaging Newsletter, Pub. No. 08/06, Feb 06. [3] James, J. D., Pounds, C. A., and Wilshire, B., “Fla ke Metal Powders for Revealing Latent Fingerprints, ” Journal of Forensic Sciences, JFSCA, Vol. 36, No. 5, Sept 1991, pp 1368-1375. [4] W. E. Kuhn, I. L. Friedman, W. Summers, A. Szegvari , “Milling of brittle and ductile materials”, ASM Handbook, Powder metallurgy, Vol 7, 1985. [5] J. D. James, and B. Wilshire, “New powder metallurg y techniques for fingerprint detection,” Key Engineering Materials (Switzerland ), Vol. 86-87, 1993, pp 11-16. [6] J. D. James, R. J. Gowland, and B. Wilshire, “Resea rch yields new uses for commercial powders,” Advanced Materials Technology/International, 1992, pp 94-99. [7] R. Besold, H. C. Neubing, and E. D. Lloyd, “Metal flakes: highly innovative powder products,” Powder Metallurgy, Vol. 32, No. 1, 1989, pp 25-28. [8] J. D. James, C. A. Pounds, and B. Wilshire, “Produc tion and characterisation of flake metal powders fo r fingerprint detection,” Powder Metallurgy, Vol. 34, No. 1, 1991, pp 39-43. [9] G.S. Sodhi, J. Kaur, “Powder method for detecting l atent fingerprints: a review,” Journal of Forensic Science International, Vol. 120, Issue 3, 2001, pp 172-176 (a) (b) (c) 11 12 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEED MACHINING OF AISI/SAE 4140 STEEL F. Akbar* 1 , P. T. Mativenga 2 , M. A. Sheikh 3 School of Mechanical, Aerospace and Civil Engineeri ng, University of Manchester, Manchester, M60 1QD, United Kingdom Tel: +44 (0 ) 16 1 20 6 38 3 0 (1) E-mail: faraz.akbar@postgrad.manchester.ac.uk , (2) E-mail: p.mativenga@manchester.ac.uk and (3) E- mail: mohammad.a.sheikh@manchester.ac.uk * The corresponding author Abstract Temperature and heat partition at the tool rake face are signif icant parameters which can influence tool wear mechanisms, tool life and hence machi ning quality. Therefore, reducing the amount of heat flowing into the cutting tool is of particular importance. In this paper, various analytical approaches for determining the percentage of the heat flowing into the cutting tool from the secondary deformation zone are evaluated. These analytical models were either derived solely from ther mal properties of the tool and workpiece materials or a combination of thermal and contact c onditions. Where experimental data was required, cutting tests were performe d for conventional to high speed machining of AISI/SAE 4140 high strength alloy steel at cu tting speeds ranging between 100 and 880 m/min. Cutting temperatures were measured experimentally using an infrared thermal imaging camera. This predicted heat partition was then benchmarked using heat partition values available in literature . The paper elucidates on important thermal properties of the tool and workpiece, cut ting process variables, and contact phenomena which should all be considered in predicting heat partition. Keywords: Heat Partition, High Speed Machining (HSM ) and Tool-Chip Contact Area. 1.0 Introduction High Speed Machining (HSM) at significantly increas ed cutting speeds and feedrates can lead to a marke d reduction in machining time. Therefore, HSM is now recognised as one of the key manufacturing technolo gies for higher productivity. Key advantages of HSM have been reported as high material removal rates, low cutting forces, the reduction in lead times and imp rovement in part precision and surface finish [1]. The 13 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL distinction between the conventional and high speed machining depends on the workpiece material being machined, type of cutting operation and the cutting tool used [2]. The main regions where heat is generated during the orthogonal cutting process are shown in Figure 1. Firstly, heat is generated in the primary deformation zone d ue to plastic work done at the shear plane. Secondl y, heat is generated in the secondary deformation zone due to work done in deforming the chip and in overcomin g the friction at the tool-chip interface zone. Finally, heat is generated in the tertiary deformation zone, i.e. at the tool-workpiece interface, due to work done to overc ome friction which occurs at the rubbing contact be tween the tool flank face and the newly machined surface of the workpiece. Fig. 1. Illustration of heat generation zones in or thogonal metal cutting There are very few studies which report numerical v alues for heat partition, considering workpiece and cutting tool combinations, these are summarised in the Tabl e I. Most of these, except [3], only cover uncoated tools for conventional cutting speeds. From these reporte d works on conventional machining, heat partition i nto the cutting tool ( R T) has been identified to range from 56% to 10% for machining of steels. Table I. Reported values of heat entering the c utting tool for specified cutting velocity range an d workpiece and cutting tool combination Authors and reference Workpiece material Cutting tool material Cutting velocity V c (m/min) Percentage of heat entering the cutting tool ( R T ) Loewen and Shaw [4] SAE B1113 Steel K2S carbide 30 to 182 40% down to 20 % Takeuchi et al. [5] Carbon steel (C = 0.55%) P15 carbide 100 10% to 30% Wright et al. [6] low carbon iron M34 high-speed steel 10-175 20% to 10% Lo Casto et al. [7] AISI 1040 Sintered carbide type P10 99 to 240 56% down to 24% Grzesik and Nieslony [8] AISI 1045 P20 uncoated carbide 50 to 210 50% down to 40% Grzesik and Nieslony [8] AISI 1045 Multilayer coate d (TiC/Al 2 O 3 /TiN) 50 to 210 35% down to 20% Abukhshim et al. [3] AISI 4140 Uncoated cemented carbide 200 to 600 and 600 to 1200 46% down to 15% and 15% to 64% This paper, presents a benchmarking of heat partiti on models available in literature. The reference he at partition data used which is summarised in Table I. Additionally, the tool-chip interface temperature in the secondary deformation zone is predicted. Research a pproaches to estimate heat partition into the cutti ng tool, Tool Chip Workpiece ( V c) Tertiary deformation zone Primary deformation zone Secondary deformation zone ( R T) q (1- R T) q 14 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL used in this study include a machining experiment a nd cutting mechanics analysis. AISI/SAE 4140 steel was turned using tungsten-based uncoated carbide cuttin g tools and the acquired cutting forces, chip thick ness, tool-chip contact area, and sticking and sliding le ngths were analysed for heat source characterisatio n. 2.0 Analytical Models for Prediction of Heat Partition Coefficient The fraction of heat which flows into the cutting t ool (which is assumed to be stationary relative to the heat source) is based on the determination of heat parti tion coefficient ( R T), which defines the percentage of the heat entering the cutting tool (Figure 1). The frac tion R ch defines the heat energy going into the moving chip. By further manipulating heat partition models ( R T = 1- R ch), the heat going into tool, R T, can be directly computed, as summarised in Table II. Table II. Summary of heat partition models for the determination of ( R T) Model Equation for the determination of ( R T) Model Equation for the determination of ( R T) Loewen and Shaw [4] pW ch T ach os pW ch TL N lqAlq N lq R λλ θθ λ 377.0 377.0 )( + −∆+ = Shaw [9] )]/()/(754.0[1 )]/()/(754.0[ )( TaWT TaWT TSH NA NA R λλ λλ + = Reznikov [10] ])/()2/3[(1 ])/()2/3[( )( TWWT TWWT TRR ααλλ ααλλ + = Tian and Kennedy [11] W T W T W T W T TTK Pe Pe Pe Pe R + + + + + = 1 11 1 1 )( λ λ λ λ Gecim and Winer [12] choWT T TGW Vrb R 807.0)( + = λ λ Kato and Fujii [13] ( ) ( ) / ( ) 1 ( ) / ( ) p T p W KF T p T p W C C R C C ρλ ρλ ρλ ρλ = + 2.1 Some Comments on the Existing Heat Partition Models and Their Limitation s 1) Some of the heat partition models are purely based on the thermal properties of the tool and the workp iece materials (for example, Reznikov [10] and Kato and Fujii [13] heat partition models). While this methodology has been made for conventional machinin g its relevance to HSM has not been validated. In HSM it has been reported that tool-chip contact are a increases and this widens the gap available for h eat transfer [3]. Therefore, the effect of contact area needs to be taken into account. 2) These models do not account for the influence of th e cutting speed and frictional heat flux on heat partition except Loewen and Shaw model [4]. Indeed, most of the studies on steel were performed only a t cutting speeds less than 100 m/min with some studie s at 300 m/min, which lies in the conventional cutt ing range where the effect of cutting speed may be negl ected. Heat flux should ideally be considered for a ll cases. 3) The main concern about Block’s procedures [14] is t he universal assumption (in all cases) of a uniform distribution of heat flux over the contact region, especially when considering the existence of both sticking and sliding zones which can simultaneously occur on the tool rake face. 15 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL 4) Most of these models, assume steady-state condition s when partitioning the heat generated in the prima ry and secondary zones. This leads to an underestimati on of heat partition into the cutting tool especial ly in the transient regime. 5) There is a lack of consensus in the methodology for accounting of heat partition and the effect of the remote primary heat source on the rake face tempera ture field. In HSM, it could be argued that the low interaction time diminishes the significance of the primary heat source on rake face temperatures. The estimated heat partition may be higher or lower than the measurements depending on how strong the factors are which discussed above. It is thus obvio us that the issue of the heat partition into the cu tting tool needs a more scientific study and, in particular, e xploration for the HSM case. This paper investigate s this aspect for the case of machining at conventional to high cutting speeds ranging from 100 to 880 m/min. 3.0 Prediction of the Tool-Chip Interface Temperature The computation of the mean tool-chip temperature, (equation (1)), is based on dimensional analysis developed by Cook [15]. The estimated tool-chip int erface temperatures were compared with the experime ntal data. This information was used to map the interfac e temperatures to cutting speeds. 1/3 0.4 ct p W V tu C θ ρ α   ∆ =    (1) where, ∆θt is the mean temperature rise at tool-chip interface , u is the total specific cutting energy, V c is the cutting velocity, ρ C p is volumetric specific heat for the workpiece, α W is the thermal diffusivity of the workpiece material and t is the undeformed chip thickness. Since the work m aterial properties α W and ρ C p depend on temperature, this equation is applied ite ratively, with new material properties calculated b ased on successive values of ∆θt, until convergence is achieved. Temperature depend ant thermal properties of the tool and the workpiece material are adopted from [3], [1 6]-[20]. 4.0 E xperimental Investigations 4.1 Experimental Details The cutting tests were performed on a Dean Smith an d Grace Lathe machine. A Kistler cutting force dynamometer model 9121 was used to measure the cutt ing forces. The inserts used were tungsten based uncoated cemented carbide, Sandvik gray that have g eometry designated as TCMW 16T304 5015 (ISO specification) [Triangular shape insert with 0.4 mm nose radius, 3.97 mm thickness and 7 o clearance angle]. The tool holder used was Sandvik STGCR/L 2020K 16. The pre-bored workpiece used was AISI/SAE 4140 high tensile alloy steel with an external diameter of 200 mm and 2.5 mm tube thickness. The cutting te sts were performed at eight different cutting speeds of 100, 197, 314, 395, 565, 628, 785, and 880 m/min. These cutting speeds were set by the available RPM on the conventional lathe. New cutting edge was used for every 16 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL cutting speed. The feedrate ( f) and depth of cut ( ap) were kept constant at 0.1 mm/rev and 2.5 mm respe ctively. The length of cut was limited to 5 mm. 4.2 Temperature Measurements In the current work, temperatures were measured usi ng an infrared thermal imager FLIR ThermaCAM SC3000. This system allows extensive analysis of hi ghly dynamic objects and events typically found in metal machining research applications. The infrared therm al imaging camera was mounted near the machine turr et and placed directly over the tool rake face (at 30 cm stand off distance) during cutting tests. The cu tting tool temperatures were measured at the “flying line” on the tool rake face. This line was appropriately sel ected at prominent observable locations on the rake face of the cutting tool. When recalled on the stored image the ‘flying line’ gave the temperature value at the req uired tool-chip interface. 5.0 Results and Discussion 5.1 Tool-Chip Interface Temperature Cutting tool temperatures along the tool-chip inter face were measured from the IR images. The estimate d experimental tool-chip interface temperature is pre sented in Figure 2 and shows a good agreement with the model developed by Cook [15] within the conventiona l speed region. But in the HSM region increasing deviation from the experimentally measured temperat ure is observed. It must be noted that Cook’s equat ion uses the cutting velocity instead of the chip veloc ity, this leads to an under-estimation of the veloc ity effect in the high cutting speed region. Furthermore, it util ises the undeformed chip thickness and hence the he at source is considered to be purely in the shear zone and th e model appears to neglect the secondary deformatio n zone frictional heating. However, the results are in goo d general agreement and only represent a small perc entage deviation. To quantify the discrepancy between the experimental and predicted results, the percentage error (e) was used. The chart in Figure 3 compare the predict ed average tool-chip interface temperatures to the measured results for all cutting tool inserts teste d and eight different cutting speeds applied. The r esults are nevertheless in good general agreement. 400 500 600 700 800 900 1000 1100 1200 100 197 314 395 565 628 785 880 Cutting velocity (m/min) In te rfa ce te m pe ra tu re ( o C) Predicted Measured -8 -6 -4 -2 0 2 4 6 8 10 12 14 100 197 314 3 95 565 628 78 5 880 Cutting velocity (m/min) Pe rc en ta ge er ro r (e) Fig. 2. Variation of the tool-chip interface temperature with the cutting speed Fig. 3. Percentage error of calculated versus experimentally measured tool-chip interface temperature 17 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL 5.2 Cutting Forces The cutting and feed forces versus cutting speed re lationship for the various cutting tests are shown in Figure 4. It is observed, that the cutting force decreases when the cutting speed is increased from low to me dium 100 to 395 m/min, marginally increases with the cutting speed up to 565 m/min, and subsequently decreases in the high speed cutting region. The feed force also decr eases gradually when the cutting speed is increased in all the cutting tests. These results generally agree wi th the accepted effect of cutting speed in cutting forces. 5.3 Chip Compression Ratio and Shear Angle Figure 5 shows the variations in the chip compressi on ratio λh and shear angle φ with cutting speed. It is observed, that the chip compression ratio decreases while the shear angle increases when the cutting s peed is increased from 100 to 565 m/min. When cutting incre ased up to 800 m/min, the chip compression ratio increases because of an increase in the deformed ch ip thickness, whereas the shear angle decreases. Consistently, the chip thickness increases along th e cutting edge but the shear angle decreases. The i ncrease in the shear angle reduces the area of the shear plane and consequently reduces the cutting force. 0 1 00 2 00 3 00 4 00 5 00 6 00 7 00 0 20 0 4 00 6 00 80 0 Cutting velocity (m/min) M ac hi n in g fo rc es (N ) Feed force Cutting force 14 16 18 20 22 24 26 28 30 0 200 400 600 800 Cutting velocity (m/min) Sh ea r an gl e (d eg . ) 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 Ch ip co m pr es si o n ra tio shear angle chip compression ratio Fig. 4. Variation of the machining forces with the cutting speed Fig. 5. Variations in the chip compression ratio an d shear angle with the cutting speed 5.4 Tool-Chip Contact Area The nature of contact between the chip and tool rak e face has a major influence on the mechanics of machining, heat generation, and heat partition into the cutting tool. The tool-chip contact area of th e worn carbide inserts were obtained at different cutting speeds from a Polyvar optical microscope equipped w ith image-processing software. Figure 6 presents the ca lculated tool-chip contact area obtained from the o ptical microscope images. From Figure 6, it is observed th at the contact area is significantly affected by th e cutting speed. There is a decrease in the contact area when the cutting speed is increased from 100 to 395 m/m in. Beyond 395 m/min, the contact area increases with t he cutting speed. This trend is in good agreement w ith the reduction in contact area for lower cutting speeds and also agrees with more recent findings [3] in HS M. 18 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 200 400 600 800 Cutting velocity (m/min) To o l-c hi p co n ta ct ar ea (m m 2 ) Fig. 6. Variation of the tool-chip contact area wit h the cutting speed 5.5 Heat Partition into the Cutting Tool From Figure 7, it can be seen that predicted heat p artition coefficients by Kato and Fujii ( R KF) and Reznikov ( R R ), increase with the cutting speed with the maximum heat going into the tool being 63% and 72% respectively. These two models depend purely on the thermal properties of the tool and the workpiece materials, and do not take into account the contact process parameters. The Tian and Kennedy model ( R TK), which takes into account the influence of Peclet nu mbers for both the tool and the workpiece, provides values of the heat partition coefficient closer to Rezniko v ( R R ). On the other hand, Loewen and Shaw model ( R L ) shows that the heat partition coefficient decreases with an increase in the cutting speed. It is obser ved that at cutting speed of 100 m/min the heat partition coeff icient is 59.6% dropping to 11% at the maximum cutt ing speed of 880 m/min. In this case, the heat partitio n coefficient is modelled as a function of chip vel ocity, depth of cut, tool-chip contact length and most important ly, the frictional heat flux. Shaw’s model ( R SH) predicts heat partition to be in between 13.6% down to 6% which r epresents the lowest amount of heat going into the tool of all the models. The values of the heat partition in to the tool by Gecim and Winer ( R GW ) range from 33% down to 15%. From Figure 7, Gecim and Winer ( R GW ) and Loewen and Shaw ( R L ) appear to be the best models for estimating the heat partition values for a wide ran ge of cutting speeds by Abukhshim et al. [3] (Table I). It appears from this comparison that the additional in formation contained in the more accurate models are the variables of frictional heat flux, contact area, ch ip velocity, shear plane temperature, area shape fa ctor and the triple product of ( C p λ ρ). 19 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 200 400 600 800 Cutting velocity (m/min) H ea t p ar tit io n in to th e cu tt in g to o l Reznikov Kato and Fujii Shaw Loew en and Shaw Gecim and Winer Tian and Kennedy Fig. 7. Variation of heat partition with the cuttin g speed 6.0 Conclusions • The results show that the contact phenomenon in add ition to thermal properties influences the heat partition coefficient. In conventional machining, c ontact area decreases with increasing cutting speed , but in high speed region, increasing the cutting sp eed increases the contact area. • Cook’s model provides a reasonable estimate of the tool-chip interface temperature within the conventional machining region but slightly underest imates the tool-chip interface temperature in HSM region. This could be attributable to the use o f the cutting velocity instead of the chip velocity and neglecting the heat flux. The measured cutting temperature increases with the cutting speed. • The analytical methods by Kato and Fujii, and, Rezn ikov which are based solely on the thermal properties of the tool and the workpiece, appear to overestimate the percentage of heat that goes into the tool at most cutting speeds except the lowest. Kato and Fujii’s model is similar to Reznikov’s model but incorporates a scaling factor. Tian and K ennedy despite additionally considering the Peclet number follows the same trend as Kato and Fujii, an d, Reznikov. For low cutting speeds the model developed by the Loewen and Shaw, and, Gecim and Wi ner that contain contact phenomena and chip velocity in their formulations, appear to be closer in agreement for estimating the heat partition. However, Loewen and Shaw model neglects the increas e of heat going into the tool in the HSM region. Shaw follows similar trend but with lowerin g magnitude. • Existing analytical heat partition models do not pr ovide accurate quantitative estimates of the fracti on of heat from the secondary deformation zone flowing into the cutting tool. 20 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN EVAULATION OF HEAT PARTITION MODELS IN HIGH SPEE D MACHINING OF AISI/SAE 414 0 STEEL • Inferring from the best models, it can be concluded that the area shape factor, mean temperature rise due to plastic deformation in the primary deformati on zone (shear plane temperature), frictional heat flux and contact area in addition to the usual ther mal properties of the tool and the workpiece materials should all be used in developing a more a ccurate heat partition model. References [1] Fallbohmer, P., Rodrguez, C. A., Ozel, T. and A ltan, T. "High-speed machining of cast iron and all oy steels for die and mold manufacturing", Journal of Materials Processing Technology, 98, 104-115, 2000. [2] Schulz, H. and Moriwaki, T. "High-speed machini ng", Annals of the CIRP , 41 (2), 637-643, 1992. [3] Abukhshim, N. A., Mativenga, P. T. and Sheikh, M. A. "Investigation of heat partition in high spee d turning of high strength alloy steel", International Journal of Machine Tools and Manufacture, 45 1687-1695, 2005. [4] Loewen, E. G. and Shaw, M. C. "On the analysis of cutting tool temperatures", Transactions of ASME, 71, 217-231, 1954. [5] Takeuchi, Y., Sakamoto, M. and Sata, T. "Improv ement in the working accuracy of an nc lathe by com pensating for thermal expansion", Precision Engineering, 4 (1), 19-24, 1982. [6] Wright, P. K., McCormick, S. P. and Miller, T. R. "Effect of rake face design on cutting tool temp erature distributions", ASME Journal of Engineering for Industry, 102 (2), 123-128, 1980. [7] Lo Casto S., Lo Valvo E. and Micari, F. "Measur ement of temperature distribution within tool in me tal cutting. Experimental tests and numerical analysis", Journal of Mechanical Working Technology , 20, 35-46, 1989. [8] Grzesik, W. and Nieslony, P. "A computational a pproach to evaluate temperature and heat partition in machining with multilayer coated tools ", International Journal of Machine Tools & Manufactur e, 43, 1311-1317, 2003. [9] Shaw, M. C. "Metal cutting principles", Clarend on Press, Oxford, 1989. [10] Reznikov, A. N. "Thermophysical aspects of met al cutting processes", Mashinostroenie, Moscow, 198 1. [11] Tian, X. and Kennedy, F. E. "Maximum and avera ge flash temperatures in sliding contact", Transactions of the ASME Journal of Tribology, 116, 1994. [12] Gecim B. and Winer, W. O. "Transient temperatu res in the vicinity of an asperity contact", Transactions of the ASME Journal of Tribology, 107, 333-342, 1985. [13] Kato, T. and Fujii, H. "Energy partition in co nventional surface grinding", Transactions of the ASME Journal of Manufacturing Science and Engineering, 121, 393-398, 1999. [14] Blok, H. "Theoretical study of temperature ris e at surfaces of actual contact under oiliness lubr icating conditions", Proceedings of General Discussion on Lubrication an d Lubricants, Institute of Mechanical Engineers Lon don, 222- 235, 1938. [15] Cook, N. "Tool wear and tool life", Journal of Engineering for Industry, 95, 931-938, 1973. [16] Woolman, J. and Mottram, R. A. "The mechanical and physical properties of the british standard EN steels". 1, Pergamon press, London, 1964. [17] Kim, H. and Oh, S. I. "Evaluation of heat tran sfer coefficient during heat treatment by inverse a nalysis", Journal of Materials Processing Technology, 112, 157-165, 2001. [18] Gale, W. F. and Totemeier, T. C. "Smithells me tals reference book", Elsevier, London, 2004. [19] Childs, T., Maekawa, K., Obikawa, T. and Yaman e, Y. "Metal machining theory and applications", Ar nold, a member of Hodder and Headline group, London, 2000. [20] Yen, Y.-C., Jain, A., Chigurupati, P., Wu, W.- T. and Altan, T. "Computer simulation of orthogonal cutting using a tool with multiple coatings", Machining Science and Technology, 8 (2), 305–326, 2004. 21 22 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS DRY GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VIBRATIONS Taghi Tawakoli, Bahman Azarhoushang* Institute of grinding and precision technology (KSF ) , Furtwangen University, Germany Abstract Compared to other machining processes, grinding involves high spe cific energy. A major part of this energy is transformed into heat which has a detrimental effect on surface integrity and grinding wheel wear. In conventional dry gr inding, as there are no cutting fluids to transfer the heat from the contact zone , minimizing the grinding energy and grinding forces are the matters of importance. To make a step forward to pure dry grinding a new technique, called ultrasonic assiste d grinding has been developed. The advantages of ultrasonic assisted grinding were pr oved mostly for the brittle material. Our investigations show the improvement on t he surface roughness, considerable reduction of the grinding forces and thermal dam age in case of using ultrasonic assisted dry grinding (UADG) comparing to conventional dry grinding (CDG) for a soft material of 42CrMo4. A decrease of up to 60-70% of nor mal grinding forces and up to 30-50% of tangential grinding forces has been achieved. Keywords: Dry grinding, Ultrasonic assisted dry gri nding, Grinding forces, Surface Roughness, Cutting fluids. 1.0 Introduction The cutting fluids are mainly used in metal removal processes due to their effect on transmitting gene rated heat in the contact zone, reduction of friction in the t ool-workpiece contact zone and chip transportation from the cutting area. On the other hand cutting fluids have serious disadvantages, such as health hazards and the explosiveness of oil vapor, environmental pollution , wear of the elements of the machine tool and incr easing manufacturing cost. In order to decrease the negati ve environmental impact of the cutting fluids and r educing manufacturing costs, new machining techniques such as dry machining [1][2] are used. During grinding m any of the super abrasive grits which are in contact wi th the workpiece do not perform real cutting, but i nstead generate heat by rubbing and plowing the workpiece surface in the contact zone. The high heat generati on associated with a high negative rake angle and with a great contact length in grinding processes, can greatly increase the temperature in the contact zone. Witho ut sufficient cooling and lubrication, this can cau se thermal damage on the workpiece surface [3]. That is why cu tting fluid is necessary in most grinding applicati ons, and the methods of minimum grinding fluid or dry grindi ng have not yet been fully successful in industrial 23 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS applications [4][5]. Generally in conventional dry grinding (CDG), as there is no cutting fluid to tra nsfer the heat from the contact zone, problems frequently occ ur in terms of thermal damage on the workpiece surf ace, increasing the grinding energy and grinding forces, wear of grinding wheel, low material removal rate (regarding relatively low depth of cuts) as well as poor surface integrity compared to conventional gr inding with cutting fluids. A recent and promising techniq ue to overcome these technological constraints is k nown as ultrasonic assisted dry grinding (UADG). The princ iple of this technique is to superimpose high frequ ency (16–40 kHz) and low peak-to-peak (pk-pk) vibration amplitude (2–30 µ m) in the feed or crossfeed direct ion to the tool or the workpiece. UADG is a hybrid process of CDG and ultrasonic oscillation. It is applicabl e to both ductile and brittle materials. By using ultrasonic assisted machining significant improvements in thru st force, burr size, material removal rate, tool wear, heat g eneration, noise reduction and surface finish have been reported. Zhang et al. [6] have both theoretically and experimentally concluded that there exists an o ptimal vibration condition such that the thrust force and torque are minimized. Takeyama and Kato [7] found t hat the mean thrust force in drilling can be greatly reduce d under ultrasonic vibrations. Drilling chips are t hinner and can be removed more easily from the drilled hole. B urr formation at the entrance and the exit sides is greatly reduced with the low cutting forces. Thus, the over all drilling quality is improved with the employmen t of UAD. Azarhoushang and Akbari [8] have achieved sign ificant improvements in the circularity, cylindrici ty, surface roughness and hole oversize by applying ult rasonic vibration to the tool with out using any cu tting fluids. Prabhakar [9] has experimentally demonstrat ed that the material removal rate obtained from ult rasonic assisted grinding is nearly 6-10 times higher than that from a conventional grinding process under sim ilar conditions. Uhlmann [10] found that for ceramic mat erials, ultrasonic assisted grinding can be applied as an efficient production technology and the ultrasonic assisted creep feed grinding provides enormously re duced normal forces at slightly increased wheel wear and surface roughness. Tawakoli et al [11] demonstrated that in ultrasonic assisted dressing of CBN grinding wheels , considerable reduction in grinding forces and dre sser wear is achievable. In this investigation, a UADG system has been desig ned, fabricated and tested. Improvements in the R z and R a (parameters of surface roughness) of the ground sur faces, reduction of the grinding forces and thermal damages on the ground surface due to superimposing of ultrasonic vibration in the dry grinding of 42Cr Mo4 have been achieved. The effect of vibration amplitu de, feed speed and depth of cut on surface roughnes s and the grinding forces have been investigated. 2.0 Experimental Setup and Procedures Fig. 1a illustrates schematically the experimental set-up. The workpiece holder consists of a piezoele ctric transducer, a booster, a horn and a special fixture . The ultrasonic power supply converts 50 Hz electr ical supply to high-frequency electrical impulses. These high frequency electrical impulses are fed to a piezoelectric transducer and transformed into mecha nical vibrations of ultrasonic frequency (23 kHz), due to the piezoelectric effect. The vibration amplitude i s then amplified by the booster and the horn and tr ansmitted to the workpiece attached to the horn. The resultan t vibration of the workpiece fixed in the tool hold er reaches 10 µ m (i.e. 20 µ m peak to peak) at a frequency of a bout 23 kHz. Vibration is applied to the workpiece in the feed direction of the grinding wheel. The amplitude of the ultrasonic vibration can be adjusted by cha nging the setting on the power supply. The experimental set-u p used to study UADG is shown in Fig. 1b. 24 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS (a) (b) Fig. 1. (a) Scheme of the experimental set-up. (b) Experimental set-up for ultrasonic assisted dry gr inding. The experimental equipment consists of the followin g:  Machine tool: Elb Micro-Cut AC8 CNC universal surfa ce grinding machine  Ultrasonic Vibration Generator: Mastersonic MMM gen erator-MSG.1200.IX  Eddy current displacement measurement system: Micro epsilon eddyNCDT 3300, to measure the amplitude of vibration.  Surface roughness tester: Hommel-Werke, model T-800 0  Dynamometer: Kistler piezoelectric dynamometer mode l 9255B 25 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS 3.0 Experiments The settings of main machining parameters for the p resent study are summarized in Table 1. Table 1. Major machining parameters Grinding wheel Vitrified bond Al 2 O 3 grinding wheel, Grain Size 120 Workpiece 42CrMo4, Hardness 85 HRB, Dimensions 60* 55*30 mm*mm*mm Grinding conditions Feed speed v ft= 500-1000-1500-2000 mm/min; Cutting speed v c = 60 m/s; Depth of cut a e = 0.010- 0.030 mm; No Coolant (Dry grinding) Grinding process Dry surface grinding Dressing conditions Wheel speed v cd = 60m/s, Depth of dressing a ed = 50 µ m, Overlapping ratio U d = 2, Total depth of dressing a ed -total= 100 µ m Dressing tool Diamond single point dresser width b d =2 mm Direction of ultrasonic vibration Feed direction Ultrasonic vibration conditions Frequency f=23 KHz, Amplitude A=10µ m The tests were carried out for both UADG and CDG wi th the same instrument. However, during the CDG the ultrasonic generator was switched off. Every workpi ece was divided into three different sections (Fig2 ). 4.0 Experimental Results and Discussion Almost all of CDGs were unsuccessful due to the the rmal damage on the ground workpiece surface. As the re were no cutting fluids to transfer the high heat fr om the contact zone this result had been expected. Fig. 2 shows photographs of the ground surfaces. It is app arent that ultrasonically assisted ground surfaces have experienced much less thermal damage compared to co nventional ground surfaces. (UADG A=10 µm F=23 KHz) (CDG) Fig. 2. The ground surfaces, v c =60 m/s a e =20µm a) v ft =1000 mm/min b) v ft =1500 mm/min c) v ft =2000 mm/min. 26 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS The effect of vibration amplitude, feed speed and d epth of cut on surface roughness and grinding force s were studied. In order to achieve reliable data each tes t was repeated 3 times. In all the figures, lines w ere formed by calculating the least-squares fit through the da ta points for a second-order polynomial equation. F ig. 3 shows the relationship between vibration amplitude and normal grinding force. The amplitude zero in th is figure represents results of conventional dry grind ing. The experimental results show significant impr ovement for UADG compared to CDG in different vibration amp litudes. Apparently, the reason for these improveme nts is the change of the nature of the cutting process, which is transformed into a process with a multipl e-impact interaction between the abrasive grits and the form ed chip. Fig. 3. Specific Grinding forces vs. Vibration Ampl itude (a e =20µm, f=23 kHz). Figs. 4–7 compare the grinding forces and surface r oughness produced by UADG with CDG under different depth of cuts and feed speeds. Experiments were car ried out at v c =60 m/s , f=23 kHz, A=10 µ m. Based on the results from previous stages, it is believed that U ADG performs enhanced under these conditions. These conditions are not essentially the optimal ones. Fo r depths of cuts more than 10 µ m in CDG thermal dam ages of the ground surfaces were observed. This phenomen on is shown with a fire symbol in the figures 4 and 5. It should be noted that the scatter in the measured su rface roughness and grinding forces obtained throug h UADG is much less compared to CDG. It means that us ing UADG increases the repeatability of the process . The maximum oscillating velocities (up to 87 m/min) and accelerations (up to 208,840 m/s 2 ) are generated at the amplitude of 10 µ m and a frequency value of 23 kHz. The larger the vibration amplitude, the greate r the material removal rate per active grain and the high er the kinetic energy with which the grits strike t he work surface. Due to the high frequency interaction of a ctive grains on the workpiece, the cutting process in UADG becomes discontinuous and ultrasonic impact action occurs, thus causing the material to begin to rollo ver more easily as well as more micro cracking propagation i n the cutting zone which both make an effective int eraction between grits and workpiece surface. Therefore the grinding forces and frictional effects are decrease d, so that less plastic deformation occurs. 27 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS Fig. 4. Specific Grinding forces vs. Depth of Cut, v ft =1000 mm/min (UADG: A=10µm, f=23 kHz). Fig. 5. Grinding normal force vs. Feed Speed, a e =20 µm (UADG: A=10µm, f=23 kHz). It has already been proven that deformation process es for ultrasonic assisted machining are restricted in the vicinity of the cutting edge along the surface of t he workpiece and are not observed underneath the cu tter, in contrast to the conventional machining process [12] . Plastic deformation of the machined surface in ca se of using ultrasonic oscillation is less than that in c onventional machining. In addition the coefficient of friction in grinding decreases with an increase in sliding spee d between the grit and the material. As the sliding speed in UADG due to ultrasonic vibration is higher than sli ding speed in CDG, the coefficient of friction redu ces. This suggests that in UADG a fewer number of strong bond s between the grit and the material are formed. Aut hors assume that by oscillation of the workpiece in feed direction, the rubbing and plowing regimes which c ause the major part of plastic deformation are reduced so th at the grinding specific energy is also reduced and the thermal damage on the ground surface is significant ly decreased. 28 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS Fig. 6. R a and R z vs. Depth of Cut, v ft =1000 mm/min (UADG: A=10µm, f=23 kHz). Fig. 7. R a and R z vs. Feed Speed, a e =20 µm (UADG: A=10µm, f=23 kHz). Reduction in plowing and rubbing regimes is also le ad to reduction of the distance between peaks and v alleys and consequently decreasing R z . Due to feed ultrasonic oscillation (sinusoidal mo vement of the workpiece in feed direction) the possibility of the interaction between the grit and the workpiece surface in each contact length will be increased. It is thought that the gr it will have more chance to cut the peak of the sur face and therefore the R z parameter of the surface roughness will be improve d. 5.0 Conclusion • Comparative experiments of the grinding forces demo nstrated up to 70% reduction in normal grinding force and up to 50% in tangential grinding forces for the workpieces machined with superimposed ultrasonic vibration. Most of CDGs wer e unsuccessful due to the thermal damage on 29 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D R Y GRINDING OF SOFT STEEL WITH USE OF ULTRASONIC VI BRATIONS the ground workpiece surface. The reason for this p henomenon was due to the absence of cutting fluids in the process and consequently the generati on of high heat in the contact zone. These improvements are subjected to the change of the nat ure of the cutting process in UAD, which is transformed into a process with a multiple-impact i nteraction between the tool and the formed chip resulting in interrupted cutting and reducing the g rinding forces, frictional effect and plastic deformation zone. • It was also found that using UADG leads to signific ant improvements on the R z and R a parameter. It is assumed that the improvement in these parameters is due to the fact that the grit in UADG has a higher chance to cut the peak of the surface due to the feed ultrasonic oscillation and increasing the possibility of the interaction of the grit and the workpiece surface in each contact length. References [1] F. Klocke, G. Eisenblaetter, “Dry cutting” , CIRP Ann. Manufact. Technol. Volume 46 (2) (1997) 519–526. [2] P.S. Sreejith, B.K.A. Ngoi, “Dry machining: ma chining of the future”, J. Mater. Process. Technol. 101 (1) (2000) 287–291. [3] C. Heinzel, “Methoden zur Untersuchung und Opti mierung der Kühlschmierung beim Schleifen“, Dissert ation Universität Bremen, 2003. [4] T. Tawakoli and M. Rabiey, “Trockenschleifen, G renzen und Möglichkeiten”, 6. Seminar “Moderne Schleiftechnologie und Feinstbearbeitung” in Stuttg art, Hrsg. T. Tawakoli, 17.05.2006. [5] T. Tawakoli and M. Rabiey, “An Innovative Conce pt for Dry grinding with Resin and Vitrified Bond C BN Wheel”, International Grinding Conference ISSAT 2007/ISME . [6] L.B. Zhang, L.J. Wang, X.Y. Liu, H.W. Zhao, X. Wang, H.Y. Luo, “Mechanical model for predicting th rust and torque in vibration drilling fibre-reinforced compo site materials”, International Journal of Machine Tools & Manufacture (41) (2001) 641–657. [7] H. Takeyama, S. Kato, “Burrless drilling by mea ns of ultrasonic vibration”, Annals of CIRP 40 (1) (1991) 83–86. [8] B. Azarhoushang, J. Akbari, “Ultrasonic-assiste d drilling of Inconel 738-LC”, International Journal of Machine Tools and Manufacture, Volume 47, Issues 7-8, June 2007, Pages 1027-1033 [9] P.D. Prabhakar, P.M. Ferreira, M. Haselkorn, “ An experimental investigation of material removal r ate in rotary ultrasonic machining”, Transactions of the North American Manufacturing Re search Institution of SME Vol. 20,(1992) pp. 211-218. [10] E. Uhlmann, “Surface Formation in Creep Feed G rinding of advanced Ceramics with and without Ultra sonic Assistance”, Annals of CIRP , Vol.47/1/1998. [11] T. Tawakoli, E. Westkaemper, A. Rasifard, “Ult rasonic Assisted Dressing of vitrified CBN Grinding Wheel”, 4 0th CIRP International Seminar on manufacturing Systems , Liverpool, UK, 2007. [12] V.I. Babitsky, A.V. Mitrofanov, V.V. Silbersch midt, “Ultrasonically assisted turning of aviation materials: simulations and experimental study”, Journal of Ultrasonics (42) (2004) 81–86. 30 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES EXPERIMENTAL INVESTIGATION OF DAMAGES ON MACHINED SURFACES OF GFRP PIPES A.Naveen Sait 1 , S.Aravindan 2 , A.Noorul Haq * 3 1. Research Scholar, Department of Production Engineer ing, National Institute of Technology, Tiruchirapalli – 620 01 5 . 2 . Assistant Professor, Department of Mechanical Engin eering, Indian Institute of Technology, New Delhi – 110 01 6 . 3 . Professor, Department of Production Engineering, Na tional Institute of Technology, Tiruchirapalli – 620 01 5 . * Corresponding author: anhaq@nitt.edu Abstract Composite materials play a vital role in various applications in in dustries such as aerospace, automobile and machine tool industry. The main reason for the significant application of composite materials is due to their excellent properties such as high specific strength, high specific stiffness, high damping, l ow thermal expansion and good dimensional stability. Glass Fiber Reinforced Plastic (GF RP) is an advanced composite material, which is considered to be an alternative t o heavy exotic materials. It is widely used in a variety of applications which include s aircraft, robots and machine tools. Most of these applications require high quality m achined surfaces with high dimensional accuracy and good surface integrity. Machining param eters, which are not carefully adjusted to the task of cutting GFRP’s may l ead to severe damage of work material as well as machine tool. Machining operations of cur ed GFRP pipes are mainly finishing operations and hence the quality achieved is of great interest. Such GFRP pipes are finding application in transportation of corrosi ve fluid industries. The quality of cutting results is very often carried out by vi sual inspection method. This work mainly focuses on the damages of the machined surfac es of hand lay up GFRP pipes. Experiments were conducted based on Taguchi’s technique and the results are also tabulated and presented in this paper. Apar t from visual inspection the machined surfaces are also subjected to SEM analysis for microstruct ural studies. Keywords: GFRP , Machining, Taguchi, Surface Roughne ss, Damages 31 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES 1.0 Introduction Composite materials play a vital role in various ap plications in industries such as aerospace, automob ile, machine tool industry [1]. The main reason for the significant application of composite materials is d ue to their excellent properties such as high specific strength , high specific stiffness, high damping, low therma l expansion and good dimensional stability [2]. Glass Fibre Reinforced Plastics (GFRP) have gained great importance among the composites since glass fibre r einforced polyster emerged in the early thirties. E nlarging the field of technical applications mainly for ligh t weight components, which are mechanically highly loaded, finally resulted in the development of more efficie nt reinforcing materials [3]. Demand for machining GFRP materials has always existed [4] and thus the study on machining plays great importance in composite i ndustry. Machining parameters, which are not carefully adjus ted to the task of cutting GFRP’s may also lead to severe damage of machine tools. Glass fibres, which are hi ghly abrasive by nature may cause premature roundin g of cutting edges. Distinct differences in hardness bet ween fibre and matrix combined with the high cuttin g resistance fibres result in edge chipping. Since ma chining operations for cured laminates are mainly f inishing operations the quality achieved is of some interest [5]. De-lamination, fiber pull-out, fibre-fragment ation, burning and fuzzing are some of the damages caused by machining on GFRP as reported by Wang & Zhang [6]. For the machining of GFRP the content of the notion “quality” has to be enlarged as compared to metal working. The material damage caused by the machinin g process made it evident that beyond the evaluatio n of surface shape and roughness a description of materi al affection has to take place. At present, commonl y accepted standards of measurement techniques and ch aracteristic indices do not exist. Moreover, the classification of cutting results is very often car ried out by visual inspection. All these difficulti es of quality description and measurement made it necessary to us e GFRP of well known and equal composition for all of the tests and to carry out measurement on the same instruments, applying the same method. For this rea son no other cutting data, even though available, have bee n used for the comparison of cutting techniques. Ma ny authors have published cutting results of GFRP, but since in many cases the material damage and surfac e quality are not completely described and also they are not always comparable. This paper, therefore, i s intended to give a general survey of the damages on machined surface by visual inspection method and S EM analysis. 2.0 Experimental Details 2.1 Materials and Processes GFRP pipes were made using the resin composition of Isophthalic (50%) and Vinylester (50%). The volume fraction of the materials is 65:35 (Resin: Glass). The fiber orientation angle of the specimen used f or the tests is 90 0 . The hand lay up pipe composite specimens were of 75mm length, 30mm and 55mm of inner and outer diameters respectively. A CNC lathe (FANUC) with 7. 5KW spindle power and maximum speed of 4500rpm was used to perform the machining operation. The fo rce measurement was carried out by using a Kistler dynamometer. The data acquisition was carried out b y appropriate software called Dynaware kistler. 32 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES Coated Carbide tool inserts (K 20 grade) were used for machining. The cutting tool i nserts used for the machining are of readily available Kennametal make. The geometry of the cutting tool insert is as foll ows: rake angle -7 0 (negative), 7 0 clearance angle, 80 0 edge major tool cutting, 0 0 cutting edge inclination angle and nose radius of 0.8mm . Tool wear was measured using Passing and Reflection type Tool Maker’s microscope having a least count of 0.5micron. Flank wear was m easured by the width of wearland on the flank below the cutting edge. The crater wear was measured by the d epth of cup in the rake face. The surface roughness was evaluated using a surface roughness measuring instr ument of Kosaka Lab, Japan. The cut off length of the instrument is 0.80 mm. A digital camera of Pentax ( optio 50) make with a focal length of 5.4mm to 16.2 mm and 5.0 mega pixels of clarity was used to obtain t he image of the machined surface. The machined surf aces are studied through images obtained from digital ca mera and SEM. 2.2 Taguchi Method Robust design is an engineering methodology for obt aining product and process conditions, which are minimally sensitive to the various causes of variat ion to produce high quality products with low devel opment and manufacturing costs [7]. Taguchi’s parameter de sign is an important tool for robust design. It off ers a simple and systematic approach to optimize design f or performance, quality and cost. Taguchi methods w hich combine the experiment design theory and the qualit y loss function have been applied to the robust des ign of products and process and have solved even complex p roblems in manufacturing. Taguchi method uses a special design of orthogonal arrays to study the en tire parameter space with a small number of experim ents. The experimental results are then transformed in to a Signal-to-Noise (S/N) ratio. Taguchi recommends the use of S/N ratio to measure the quality characteris tics deviating from the desired value. The S/N rati o for each level of process parameters is computed based on th e S/N analysis. Regardless of the category of the q uality characteristic, a greater S/N ratio corresponds to better quality characteristics [8]. 2.3 Plan of Experiments The methodology of Taguchi for three factors at thr ee levels was used for the implementation of the pl an of experiments. The orthogonal array L 18 is selected which has 18 rows corresponding to the number of tests with the required columns. The plan of experiments compr ises of eighteen tests where cutting velocity (V), feed rate (f) and depth of cut (d) are considered as dif ferent factors and their corresponding levels are c hosen as shown in Table I. The composite pipe specimens were machined by using L 18 orthogonal array separately with the same machining parameters for each of the eight een test conditions. The quality responses to be o bserved are machining force, flank wear, crater wear and su rface roughness. The experimentally collected data are then optimized using S/N ratio values calculated for eac h test conditions separately. The machined surfaces are also viewed and images are obtained with a high zoom dig ital camera to check the quality of the machined surfaces. The machined surfaces are also subjected to SEM analysis for confirmation of quality. 3.0 Results and Discussion 33 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES In order to describe the cutting results with respe ct to surface quality and material damage, two diff erent measured values are used as follows, i. Averaged roughness ii. Width of the damaged zone For the interpretation of these values some importa nt remarks should be kept in mind. First of all the measurement of roughness in FRP is less dependable than in metals, Because protruding fibre tips may l ead to incorrect results or at least to large variations o f the reading. Additional errors may result from ho oking of the fibres to the stylus. Therefore only glass and carb on fibre composites are accessable to roughness measurement. In this case, since it is a glass fibr e reinforced plastic the measurement on roughness i s possible on machined surfaces. The result of the roughness m easurement is to a great extent dependent on the st ylus path with respect to the fibre direction. Since the main direction of fibres may change from layer to layer, care has to be taken to keep the stylus path either in o ne layer or to get a good average of all different layers. For this reason maximum or single values of roughness l ike peak to valley roughness (Rt) are less suitable than integral values like averaged roughness (Ra). For t he same reason the roughness has been measured atle ast several times and then again been averaged. The eva luation of materials damage by microscopic inspecti on, of course, bears the uncertainty, whether or not all m aterials defects – especially those beneath the sur face – have been detected. In addition to this, damage first oc curs as single spots of delamination rather than as complete margin, which is typically for a high intensity of damage. From the structural – mechanic’s point of v iew an integrated value, containing the damaged area or, a n averaged width might be a more precise descriptio n of the damage situation. At present no sufficient data abo ut the engineering strength of damaged FRP are know n, which could decide whether the maximum damage, the averaged damage or the size and shape of damage should be measured. The criteria of visible appeara nce may in many cases be as important as the perfor mance of the work piece. Hence, in this work the damages are first observed under visual inspection method b y considering width of the damage and later confirmed through SEM images. Also the machinability in this work was evaluated b y the parameters such as surface roughness (R a ), machining forces and tool wear. The results obtained through experiments are presented in Table I. Taguchi’s Des ign of Experiments techniques were used for conducting exp eriments. Also these techniques are effectively use d for optimization of parameters. By considering the cutt ing velocity, feed rate and depth of cut and their interactions, the number of experimental trials req uired was determined as 18 and the experiments were conducted with different cutting inserts of the sam e specification to obtain more data. Machining of GFRP is continued with the same insert up to a maximum material removal of 30 cm 3 . For such constant volume of material removal, the tool wear, machining force and surface finish are measured un der different machining conditions. The Taguchi’s appro ach to experiment design consists of the following steps. The first step in Taguchi method is to determine th e quality characteristic which is to be optimized. The output or response variable which influence effectively on the quality of the product is known as quality characteristic. In this study, the tool wear, machi ning force and surface roughness are the quality characteristics. In the second step, the control pa rameters or test parameters which have significant effects on the quality characteristic are identified with the required number of levels. In the third step, the a ppropriate orthogonal array for the control parameters is sele cted after calculating the minimum number of experi ments required to be conducted by considering the interac tive effects. In the Taguchi method of optimization , the signal-to-noise ratio is used as the quality charac teristic of choice. 34 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES Smaller the better characteristic : ∑−= 2 n 110log yN S (1) where, y - Average of observed values, n-Number of obser vations In this study, “the smaller the better” characteris tic is applied to determine the S/N ratio for tool wear, machining force and surface roughness, since all th ese parameters are to be minimized. Table I: Machining parameters and its responses Optimal combinations of parameters are determined b ased on assumed weightage of 1: 2: 3: 4 for crater wear, flank wear, machining force and surface roughness r espectively. The weightage of parameters was assumed on the basis of physical significance of each paramete r during machining. Surface roughness plays an impo rtant role in many areas and is a factor of greater impor tance in the evaluation of machining accuracy [12], and hence it is given maximum weightage. Machining forc e plays the next prominent role after surface rough ness [2], and therefore the next best weightage was assu med to it. Apart from surface roughness and machini ng force, tool wear also contributes significantly in determining the optimum machining characteristics. Mostly flank wear is considered, since it largely affects the stability of the cutting wedge and consequently the dimensional tolerance of the machined work surface [9]. And hence the weightage for flank wear is assu med as the third best, while the weightage for crater w ear was assumed to be the least. The S/N ratios o f combined objective with flank wear, crater wear, machining f orce and surface roughness are calculated and prese nted in Table I. 3.1 Visual Inspection Method Test Condition Number Speed (m/min) Feed rate (mm/rev) Depth of Cut (mm) Flank Wear (mm) Crater Wear (mm) R a (micron) F m (N) S/N ratio of 100% Combined Objective 1 100 0.05 0.5 0.023 0.018 3.95 17.50 28.20 2 100 0.1 1 0.018 0.048 8.21 47.50 27.51 3 100 0.2 2 0.028 0.018 6.22 17.50 30.00 4 150 0.05 0.5 0.017 0.018 4.25 17.70 27.25 5 150 0.1 1 0.048 0.015 8.52 15.00 28.83 6 150 0.2 2 0.025 0.015 5.03 15.00 33.38 7 200 0.05 1 0.018 0.008 6.07 07.50 30.82 8 200 0.1 2 0.018 0.013 6.07 12.50 30.30 9 200 0.2 0.5 0.023 0.012 6.34 11.50 29.60 10 100 0.05 2 0.025 0.013 4.73 12.80 27.86 11 100 0.1 0.5 0.020 0.030 6.14 30.00 28.38 12 100 0.2 1 0.020 0.023 7.51 22.50 26.81 13 150 0.05 1 0.025 0.025 3.81 25.00 31.38 1 4 15 0 0.1 2 0.0 1 6 0.0 1 3 4.0 3 12 . 5 0 33 . 8 9 15 150 0.2 0.5 0.015 0.008 3.73 07.50 32.88 16 200 0.05 2 0.023 0.008 4.41 08.00 27.51 17 200 0.1 0.5 0.028 0.018 8.12 17.50 30.30 18 200 0.2 1 0.023 0.012 5.08 12.00 28.20 35 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES As discussed earlier the machined surfaces are zoom ed and observed using a 5.0 mega pixel digital came ra. The surface of the GFRP pipe before machining is sh own in figure 1. The roughness of the unmachined surface was found to be around 80 microns. Figures 2 and 3 are the images corresponding to test condit ion numbers 5 and 2 respectively. The measured average roughnesses of these surfaces are found to be 8.52 microns and 8.21 microns respectively. These two te st conditions produced the maximum roughness in ter ms of numerical value. And it is clearly confirmed by the images presented in figures 2 and 3. Not only h ave the two test conditions produced poor finish but also t hrough visual inspection method maximum damages are observed. If the cutting parameters of these two te st conditions are analyzed, it is observed that in these two cases minimum feed rate was not there. Hence it is clearly inferred that feed rate is the most signifi cant parameter which influences the quality of the machi ned surface. Therefore the feed rate should be mini mum to get good quality of machined surfaces on machining hand lay up GFRP pipes. Fig. 1. Unmachined Surface Fig . 2. Machined surface of TCN 5 Fig. 3. Mac hined surface of TCN 2 Figures 4 and 5 are the images corresponding to tes t condition numbers 13 and 15 respectively. The mea sured average roughnesses of these surfaces are found to be 3.81 and 3.73 microns respectively. These two te st conditions produced the minimum roughness in terms of numerical value. And it is clearly confirmed by the images presented in figures 4 and 5. The damages on the machined surfaces of figures 4 and 5 are found to be very minimum as compared to the surfaces presented in figures 2 and 3. If the cutting parameters of th ese two test conditions are analyzed, it is observed that i n most of these cases the feed rate was found to be minimum. Hence again this study reiterates that the feed rat e is the most significant parameter which influence s the quality of the machined surface. Fig. 4. Machined surface of TCN 13 Fig. 5 . Machined surface of TCN 15 3.2 SEM Analysis of Machined Surfaces 36 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES In order to confirm the results obtained from visua l inspection method, the machined surfaces are subj ected to SEM analysis. Figure 6 (a) & (b) represents the mic rostructure obtained on lower and higher magnificat ions respectively of machined surface for test condition number 5 (poor cutting conditions). The surface ro ughness of these surfaces was found to be 8.52 microns. The increased surface roughness is due to the poor sel ection of machining parameters. And also the damage on the su rface is also found to be severe. Figure 7 (a) & ( b) represents the microstructure obtained on lower and higher magnifications respectively of machined sur face for test condition number 15 (good cutting conditio ns). The surface roughness of these surfaces was fo und to be 3.73 microns. The minimum surface roughness is d ue to the proper selection of machining parameters. And also the damage on the surface is also found to be minimum. Figure 8 (a) & (b) represents the microstr ucture obtained on lower and higher magnifications respect ively of machined surface for test condition number 14 (optimal cutting conditions). The surface roughness of these surfaces was found to be 4.03 microns. Te st condition number 14 is considered to be the optimal cutting condition since the S/N ratio of the combi ned objective is maximum as presented in Table I. Under some specific cutting conditions the surface of ma chined specimens are damage free as shown in figures 7(a) and 7(b). But under some other specific cutting conditions, debonding and fiber breakage takes plac e easily as shown in figures 6(a) and 6(b). More n umber of porous sites and damaged zones are observed in t he case of hand laid up composite tubes. The averag e surface roughness of the machined hand lay up compo site tubes is comparatively of larger order due to inherent pores during manufacture and induced damag es during machining. From all these studies, it is confirmed that the results obtained are identical i n all the three investigation methods namely, Taguc hi’s optimization technique, Visual Inspection method an d SEM analysis. (a) (b ) Fig. 6. Machined surface of Hand Lay up pipes - poo r surface finish (TCN 5) (a) (b) Fig. 7. Machined surface of Hand Lay up pipes - bes t surface finish (TCN 15) 37 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL INVESTIGATION OF DAMAGE ON MACHINED SU RFACES OF GFRP PIPES (a) (b) Fig. 8. Machined surface of Hand Lay up pipes – opt imal test condition (TCN 14) 4.0 Conclusion In this work, the quality of the machined surfaces of GFRP pipes made by hand lay up process is thorou ghly analyzed. From this study, it is observed that the feed rate is the significant machining parameter wh ich affects most the quality of the machined surfaces. The visual inspection method which was carried out in this work is found to be a useful tool to check the dama ges of the machined surface. It also gives identica l result as that of results obtained from any numerical opti mization technique or by SEM analysis. Debonding an d fiber breakage often takes place in the case of con ventional cutting conditions. More number of porous sites and damaged zones are observed on the machined surf aces of hand lay up composite tubes. To get damage free surfaces, optimized machining parameters have to be used. The SEM analysis of the machined surfac e confirms that these optimized parameters not only r educes the tool wear but, also reduces the damages and failures on the machined surface. References [1] Santhanakrishnan G, Krishnamurthy R, Malhotra S.K., 1998, Machinability Characteristics of fibre reinf orced plastics composites, Journal of Mechanical working Technology, Vol (17), pp.195-204. [2] Sang-Ook An, Eun-Sang Lee, Sang-Lai Noh, 1997, A st udy on the cutting characteristics of glass fiber r einforced plastic with respect to tool materials and geometri es, Journal of Materials Processing Technology Vol(68), pp.60-67. [3] Konig W, Grap P, 1989, Quality definition and asses sment in drilling of fiber reinforced thermoset, Annals of CIRP , Vol.(38), pp.119-124. [4] Konig W, Cronjager L, Spur G, Tonshoff HK, Vigneau M, and Zdeblick WJ, 1990, Machining of new material s, Annals of CRIP, Vol (39), pp.673-681. [5] Konig W, Grab P, 1985, Machining of fibre reinforce d plastics. Annals of CIRP , Vol (34), pp.537-548. [6] Wang X.M, Zhang L.C, 2003, An experimental investig ation into the orthogonal cutting of unidirectional fibre reinforced plastics , International Journal of Machine Tools & Manufact ure Vol (43), pp.1015-1022. [7] Phadke M.S., 1998, Quality Engineering using Robust Design., Prentice-Hall, Englewood Cliffs, NJ. [8] Douglas C. Montgomery, 2001, Design and analysis of Experiments. New York: John Wiley & sons, Inc. [9] Zoya Z.A, Krishnamurthy R, 2000, The performance of CBN tools in the machining of titanium alloys, Journal of Material Processing Technology, Vol (100), pp. 80-86. 38 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART A COMPARISON BETWEEN PROCESS 3D TOLERANCE STACK-UP AND TOLERANCE CHART A. Del Prete, D. Mazzotta, G. Ramunni & A. Anglani Department of Engineering Innovation, Faculty of En gineering, University of Salento, Italy Abstract Tolerance allocation and manufacturing operations selection in pr ocess planning are two essential factors governing the product cost in manufacturi ng. For this reasons, the development of more efficient approaches has become an in escapable necessity in order to solve the tolerance allocation and the manufacturing oper ations selection problems during process planning. In the DFM/CE (Design for Manufacturing/Concurrent Engineering) context, a dimension of t he product is often subjected to modifications, once a dimension of the product i s changed the impact on the manufacturing side has to be considered. The process p lan has to be changed without delay, the process designer has to take rapidly a corre ct decision. In order to solve this practical DFM/CE problem the tolerance chart met hod is used. When manufacturing a product with routed operations over a series of machining cuts, process engineers face with a tolerance stack-up problem. T o handle the tolerance stack-up problem in manufacturing, the most effective way is to use a tolerance chart. The tolerance chart is a graphic tool which ensures an accurat e development of the mean working dimensions and the required tolerances by the manufacturing process. The most essential task of tolerance charting is the toleranc e-chain identification. Nowadays, several computer aided systems are available, they are able to carry out a statistical analysis of dimensional variations, in these software tools tolerance and attributes are interactively extracted from CAD models, featur e attributes are varied within the specified tolerance range, and user-defined st atistical distributions are used in simulation runs to determine: the contributors, the ext ent of contribution, and statistical distribution of the analyzed part dimensions. The present work suggests a new approach using the Computer Aided Tolerance tools (CAT) implementing a tolerance chain in this environment and finally comparing the obtained results with two traditional tolerance chart methods: Surface Chain Model, an d Digraphic Method. Keywords: Tolerance charting, Process planning, Tol erance allocation, 3D Stack-up. 39 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART 1.0 Introduction During a metal cutting process planning, process en gineers should be able to assign the right toleranc e to every machining cut operation, this is a tolerance stack- up problem. To handle the tolerance stack-up proble m in manufacturing the most effective way is to use a to lerance chart thanks to which it is possible to ens ure an accurate development of the mean working dimensions and the tolerance required by the manufacturing process. In the past the tolerance chart analysis w as carried out manually [1]-[2]-[3], but since 1980 s some researchers have developed computer aided tolerance -charting tools in order to automatically proceed t o the tolerances allocation [4]-[5]-[6]. Subsequently oth er methods have been proposed and actually, the pro cess tolerance chain is carried out through the usage of manual methods (i.e. Digraphic Method [6] e Surfac e Chain Model [7]). Other authors have developed appropriat e methods in order to manage the correct sequence o f manufacturing operations and their respective toler ances [8]-[9]-[10]-[11]-[12] but in any case they a re not able to manage review requests coming from the manu facturing processes analysis (DFM/CE approach). The authors’ aim in the present work is therefore to de termine the tolerance for each sequenced manufactur ing operation. This philosophy leads to minimize the ma nufacturing cost for each geometric feature. In ord er to reach these objectives the authors extend the usage of a traditional computer aided tolerance tool (Vi sVSA ® ) to solve a practical DFM/CE problem [7]. The obtain ed results, thanks to the usage of a CAT tool, have been compared with the ones obtained through the applica tion of manual methods. 2.0 Proposed Methodology and Test Case Setup A test case has been considered as reference for th e application of the proposed methodology with its original tolerance chart [7] (Fig. 1). If tolerance allocation phase is performed by tradi tional “One Dimension” (1D) approach like Digraphic Method or Surface Chain Model two key aspects are t aken into consideration: tolerance control and tole rance planning. Tolerance control is performed by a conti nuous control on the real process in order to consi der its real capabilities (usually defined SPC - Statistical Process Capabilities ) while, tolerance planning is carried out thanks to the adoption of proper: tolerance cha rts and tolerance chain calculations. Tolerance Chart needs, as shown in Fig. 1, the foll owing input data: Design dimensions – “Blueprint Dimensions” ( B i and ± bi); Operations ( Opi ) and their sequence; Reference datum planes select ion; Process Capability ( C p) of each machining operation (± x i); Stocks Removal ( Y i). From a graphical point of view, in the tolerance chart the selected datum plane is indicat ed with a black ball while, the machined surface by the Opi is indicated by a black arrow. To each machined sur face is assigned a letter, while to each machining operation is assigned a progressive number starting from 0 (l ast operation in the sequence) and increasing it fo r every other cut applied to the machined surface. 40 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART Fig. 1. Tolerance chart for the considered referenc e case [7] Tolerance Chain gives as output data: Mean of worki ng dimensions ( X i); Stock removal tolerances (± yi); Mean and tolerance of Resultant Dimension ( R i and ± ri). In this treatment is reported uniquely the toler ance chart and the tolerance chain results without expla nation about the usage of the tolerance chain metho ds. As previously said the considered methods are Digraphi c Method and Surface Chain Model, both methods provide the same results summarized in Fig. 1, and afterwards compared with achieved results of this w ork. In the studied reference case the “1D” method used to evaluate the tolerance chain is replaced with a “3D” tolerance chain calculation carried out directly w ithin a computer aided tolerance environment (Fig. 2). Fig. 2. Proposed methodology: “3D” approach to all ocate the tolerances during process planning In the present work the authors have applied the pr oposed methodology to the test case shown in Fig. 1 to demonstrate the effectiveness of the obtained resul ts and the improved flexibility. The operative step s adopted (Fig. 3) in order to setup the numerical model are: 1 0 1 0 2 1 0 2 1 0 2 A B C D A B C D Tolerance Control Process Control - SPC Tolerance Planning Tolerance Chart + CAT 1. 2. 41 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART 1. 3D Model creation within CAD environment (UG-NX ® 4.0.4.2); 2. Model import within CAT environment; 3. Numerical model setup (modeling and characterizatio n of features plane); 4. Measurements definition; 5. Simulation setting and execution; 6. Results analysis and comparison with Blueprint Dime nsions. Fig. 3. Proposed operative steps In order to let available the CAD information withi n the CAT environment a transfer format is needed ( JT file). This operative step is performed within the CAD environment with a dedicated export template. W hen the geometric model has been built the next steps w ere: a) to create a file into the CAT environment a ble to contain: the information arising from the geometric features, GD&T (Geometric Dimensioning and Tolerancing) information, assembly sequence, measur ements, results; b) to import the design model with in the CAT tool. The authors have implemented a model havi ng proper features plane in order to simulate datum and machined surfaces during the virtual cutting operat ions ( Opi ), obtaining the setup reported in the Fig. 4 to perform the “3D” tolerance chain. CAT environment Stack up Analysis 3D Model Comparison with Blueprint Dimensions (R i vs. B i , ± r i vs.± b i ) JT file 42 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART Fig. 4. CAT model In particular, taking as reference the Fig. 4 layou t it can be said that: green features planes repres ent the rough surfaces (A2-B3-C3-D2), red features planes represe nt the machined surfaces by OPi (A1-B2-B1-C2-C1-D1) and finally the blue features planes are the finish ed surfaces (A0-B0-C0-D0). For each feature plane i t has been created: i) a profile tolerance of surface having as tolerance zone (Gaussian distribution) the Cp (Process Capability - x i from Fig. 1). The defined profile tolerance of surf ace has as first reference the datum A and as second reference the machined surface Opi ; ii) a perpendicularity constraint (perpendicular toler ance) having a tolerance zone equal to zero and as reference the datum A; iii) a datum assigned to the machined surfaces in order to guarantee the process chain (Fig. 5). Definition Description GD&T Symbol Tol.[mm] 1°DATUM 2°DATUM Datum D2 Unprocessed Surface d 0 C A1 Processed surface by S 1_OP 10 d 0.3 A C b 0 A D C2 Processed surface by S 2_OP 10 d 0.3 A D b 0 A D1 Processed surface by S 3_OP 20 d 0.3 A D b 0 A E B2 Processed surface by S 4_ OP 20 d 0.2 A D b 0 A A0 Finished surface by S 5_OP 30 d 0.2 A E b 0 A F C1 Processed surface by 6 _OP 30 d 0.2 A F b 0 A D0 Finished surface by S 7_OP 40 d 0.1 A F b 0 A B1 Processed surface by S 8_OP 40 d 0.1 A F b 0 A C0 Finished surface by S 9_ OP 50 d 0.1 A F b 0 A G B0 Finished surface by S 10_OP 60 d 0.05 A G b 0 A Fig. 5. Features plane definition 43 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART 3.0 3D Tolerance Stack-up A measurement setup has been carried out within the CAT software (Fig. 6) in order to check: the worki ng dimension ( X i), the stock removal ( Y i), the resultant dimension ( R i) and the respective tolerance zones (± x i, ± yi, ± ri). The numerical simulation has been launched havin g the setup reported below: Number of runs: 1000; Extreme Simulation, Monte Carlo Simulation in order to calculate the contribution of each tolerance [1 4], Geometric Tolerance (in this way the simulation use s the assigned geometric tolerances). Extreme Simul ation is based on the Worst Case Model and usually it is referred to the “Method of Extremes” [14]-[15] that is the simplest and most conservative of the traditional a pproaches. In this approach, the tolerance at the i nterface is simply the sum of the individual tolerances. The fo llowing equation (1) calculates the expected gap va riation. ∑ = = n i iiwc tat 1 (1) Where: twc = maximum expected variation (equal bilateral) usi ng the Worst Case Model; ai = sensitivity factor that defines the direction and magnitude for the ith dimension. In a one dimensional stack-up, this val ue is usually +1 or –1; ti = equal bilateral tolerance of the ith component in the stack-up. Fig. 6. Measurement setup Once simulation has been performed the obtained res ults (Fig. 7) can be analyzed thanks to proper post - processing tables and they can be compared with Blu eprint Dimensions. 44 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART Fig. 7 – Working Dimension, Stock removal and Resul tant Dimension obtained by “3D” tolerance stack-up within CAT Fig. 8. Comparison between Blueprint dimension, "1D " resultant dimension and “3D” resultant dimension Comparing the obtained results with the values repo rted in the Fig. 1 it is possible to state that the proposed methodology can be used as a valid alternative meth od to the manual one (Fig. 8), moreover the propose d methodology, using a “3D” tolerance stack-up approa ch, is more flexible and efficient. When the result ant dimension is calculated ( R i, ± ri) and it is compared with a Blueprint dimension ( B i, ± bi), if there is a difference between R i and B i the process designer has to modify Y i. Another possible scenario is related to possible differences between ri and bi; in this case, the process designer has to modify x i, and/or the number and/or the operation sequence. Often, after these c hanges, it is necessary to make a new calculation f or the tolerance chain with a subsequent time consuming ca lculation phase. In total according with a DFM/CE approach, the authors, with the proposed methodolog y, are able to modify rapidly, the tolerance stack- up changing for instance the x i, or the operation sequence, etc. 4.0 Conclusions This paper presented an efficient approach to alloc ate the tolerance during process planning phase. Th e proposed approach considers the chance to use a Com puter Aided Tolerance software in substitution to a manual method (i.e. Digraphic Method, Surface Chain Model). The obtained results lead to assert: the g lobal correlation between the obtained values and the effectiveness arising from the usag e of the proposed methodology. Possible further development can be: • Apply the approach to an industrial and more diffic ult case. • Implement this methodology in a parametric CAD/CAM in order to perform automatically the tolerance chain starting from the CAM operation seq uence. • Take into consideration, in the simulation, directl y the real process capability data for each operati on (statistical method – Monte Carlo simulation). Acknowledgements 45 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A COMPARISON BETWEEN PRO CESS 3D TOLERANCE STACK-UP A ND TOLERANCE CHART This research has been supported by AVIO Group. In particular the authors express their thanks to Cafa ro F. for his support on the developed activities. References [1] O. R. Wade, in: T.J. Drozda, C. Wick, Tolerance control, Tools and Manufacturing Engineer s Handbook, Vol. 1, Machining, ASME, New York, 1983, pp. 1-60. [2] C. J. Marks, Tolerance charts control production ma chining, Am. Machinist 97 (1953) 114-116. [3] J. K. Matter, Tolerance charts forecast accuracy, A m. Machinist 4 (1947) 114-118. [4] S. A. Irani, R.O. Mittal, E.A. Lehtihet, Tolerance chart optimization , International Journal of Production Research , 27 (1989) 1531-1552. [5] R. S. Ahluwali, P. Ji, Process planning in concurre nt engineering environment, The International Industrial Engineering Conference Proceedings , San Francisco, May 20–23, 1990, pp. 535–540. [6] P. Ji, J.Y.H. Fuh, R.S. Ahluwalia, A digraphic appr oach for dimensional chain identification in design and manufacturing, ASME Trans. J. Manuf. Sci. Eng . 118 (1996) 539–544. [7] J. Xue, P. Ji, Identifying tolerance chains with a surface-chain model in tolerance charting, Journal of Materials Processing Technology, 26 November 2001, pp 93-99. [8] P. Ji, A tree approach for tolerance charting, International Journal of Production Research 31 (1993) 1023-1033. [9] B. K. A. Ngoi, O.C. Teck, A complete tolerance char ting system, International Journal of Production Research 31 (1993) 453-469. [10] B. K. A. Ngoi, C. S. Tan, Graphical approach to tol erance charting – a maze chart method, International Journal of Advanced Manufacturing Technology 13 (1997) 282-289. [11] P. Ji, Determining dimensions for process planning: a backward derivation approach, International Journal of Advanced Manufacturing Technology 11 (1996) 52-58. [12] P. Ji, An algebraic approach for dimensional chain identification in process planning, International Journal of Production Research 37 (1999) 99-110. [13] ASME Y14.5 M – 2004 “Dimensioning and Tolerancing”. [14] Student Training Manual of VisVSA ® Version #6.0. [15] Dimensioning and Tolerancing Handbook - Paul J. Dra ke,Jr. 46 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADVANCED CERAMICS Yuan Julong 1,2 , Lv Binghai 1 , Wang Zhiwei 2 1 . National Engineering Research Center for High Eff iciency Grinding, Hunan University, Changsha, 410 0 8 2 , China 2 . Ultraprecision Machining Research Center, MOE Key L aboratory of Mechanical Manufacture and Automation, Zhejiang University of Technology, Hangzhou,310 0 1 4 , China Abstract To reduce or eliminate the surface damage caused by lager parti cles invading into the machining zone, and improve the processing efficiency in the ultra-precision abrasive machining for advanced ceramics, the concept of semibonded abras ive machining, which employs a newly developed semibonded abrasive plate (SBA P) as machining tool is put forward in this paper. The manufacturing process of the SBAP is introduced. A high polymer SSB, which can be solidified as d rying and be redissolved by water, is used as the semibond material, a SBAP made of 1000# Si C abrasive is presented as an example. Experiments for the studies of sur face damage in semibonded and loose abrasive lapping are carried out with the polished damage-free silicon wafer and 180# B 4 C as the large particles. The surface quality and material removal rate of workpieces in semibonded and loose abrasive lappi ng are discussed with SEM photos, AFM images, surface roughness and the weight of removed workmaterial. Experimental results show that the SBAP lappin g leads to little surface damage, and the material removal rate in SBAP lapping is much hi gher than loose abrasive machining. In the case of large particles invasion, the surface damage of loose abrasive machining gets worse and the material removal rate incre ases greatly, while those of SBAP lapping are little affected. This result can be attributed to the ‘trap’ effect of SBAP to large particles. The feasibility of concept of semibonded abrasive machining is proved. Keywords: Semibonded abrasive machining, semibonded abrasive plat, ‘trap’ effect, surface damage 47 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS 1.0 Introduction Advanced ceramics, such as silicon, sapphire, quart z, and silicon nitride, are key materials for optic al, electronic, magnetic, and structure components, whi ch set stringent requirements on surface quality an d precision. Abrasive machining process is the essent ial method for finishing the advance ceramics. Abra sive machining, as the retaining state of the abrasive i s concerned, can be classified into loose abrasive machining, such as lapping and polishing, and fixed abrasive m achining, such as grinding. Lapping/polishing is an effective mean to obtain super smooth and low damag e surface [1]. The disadvantages of loose abrasive machining are that the material removal rate is low , and the surface quality is sensitive to the size consistency of the abrasive grains. Once, larger particles, whi ch are the larger ones mixed in normal abrasive gra ins, grain agglomerate, chips fallen off from workpieces, or l arge ones from the environment, invade into machini ng zone, surface damage such as scratch and deep crack will be caused [2]-[5]. Increasing reworks result in low processing efficiency and high cost. Although the material removal rate in fixed abrasiv e machining is much higher, and the surface roughne ss is up to several nanometers, the surface damage is ine vitable. Take Electrolytic in-process dressing (ELI D) grinding for example, in the work of Ohmori.H [6], although the surface roughness Ra of ground silicon wafer is up to 2.8nm, the depth of surface damage layer i s 1 µm. Problems such as workpiece ‘chipping’ during grinding process, manufacturing of super-fine grind ing wheel, and self-dressing of grinding wheel are not solved completely [7]. For these reasons, fixed abr asive machining has not been taken as the finishing process for the high surface quality required advanced cera mics. High surface integrity and processing efficiency ar e expected to be reached simultaneously and it is h ardly achieved by the existing abrasive machining process . In this paper, a novel abrasive machining employi ng the semibonded abrasive plate (SBAP) is put forward. It is conceived that the requirement of high processi ng efficiency and surface quality in the semibonded ab rasive machining is met by the ‘trap’ effect of SBA P, which can reduce or eliminate the surface damage su ch as deep crack and scratch caused of large partic les. The conception of the semibonded machining and manu facturing process of SBAP is introduced. The effectiveness and ‘trap’ effect of SBAP is investig ated with experiments. 2.0 Conception of the Semibonded Abrasive Machining During the ideal abrasive machining, grain particle s are in same size, and loads on grains are even. O nce larger particles emerge, the workpiece is supported with t he several larger ones as shown in Fig.1. The incre asing loads on these particles result in surface damages. It is essential to improve the consistency of grai n and clean the environment. But, these works cost much and cou ldn’t resolve the problem completely. In order to avoid the surface damage from the ‘inva sion’ of larger particles, a novel abrasive tool-SB AP is developed. As shown in Fig. 2, SBAP, same as the co mmon grinding tools, consists of abrasive grains, b ond material and pores. What makes different between co mmon grinding tool and SBAP is that the adhesion between grains and bond material in SBAP is relativ e weak. When a larger particle emerges on the SBAP 48 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS surface, grains surrounding the larger particle can be forced to move for adjustment. A ‘trap’ is form ed on the surface of SBAP, and the larger particle is contain ed in SBAP. Fig.3 illustrates the ‘trapping’ proces s of larger particle. Since the adjustment of grains on SBAP su rface is ‘plastic’, load on the larger particle wil l not increase obviously that prevents surface from damag e such as deep crack and scratch. ` load tool (such as pad) (b) surface damage caused by larger particle during abrasive machining workpiece tool (such as pad) grain load (a) ideal abrasive machining Surface damage larger particles workpiece Fig. 1. Illustration of surface damage caused by la rger particle during abrasive machining abrasive grain bond material pore Fig. 2. Compositions of SBAP abrasive grain on SAGP surface 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 workpiece larger particle abrasive grain on SAGP surface abrasive grain on SAGP surface workpiece larger particle workpiece larger particle 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Fig. 3. ‘Trapping’ process of larger particle on SB AP surface Since retention force to retain the abrasive grains in the bond is intervenient between the fixed abra sive and the loose abrasive, the novel abrasive tool is named se mibonded abrasive plate (SBAP). The ability of SBAP to contain the larger particles, here, is called ‘trap ’ effect. According to this conception, the SBAP sh ould possess the following primary characteristics: 1) porosity to provide enough space to contain the larger parti cle, and 2) relative low retention force to ensure the movement of gains to form a ‘trap’. On the other hand, rete ntion force on grains should not too weak to stand agains t collapse caused by the grinding force. 3.0 Manufacturing of SBAP The bond material plays a key role in SBAP. A high polymer SSB, which can be solidified by drying, is developed to be used as bond agent. It can be redis solved in water forming a kind of glue-like liquid, which can be taken as the semibond material. SBAP is manu factured as a fixed abrasive tool. As shown in Fig. 4, the surface layer of SBAP is wetted with water to form a semibonded layer before it is used to machine the workpiece. The adhesion between wetted SSB and abra sive grain can be adjusted by controlling the amoun t of water sprayed on the SBAP surface. 49 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS Fig.5 shows the manufacturing process of SBAP. SSB, abrasive and other additives are mixed together homogeneously in hot water. Then, the mixture is po ured into a mould as shown in Fig.6, and baked unde r constant temperature 60°C for 72 hours. Fig.6 shows a SBAP, whose volume ratio of abrasive, pore and b ond material is 5:3:2. Fig.8 is the SEM micrograph of t he SBAP. abrasive grain redissovled SSB pore solidified SSB Fig. 4. Illustration of semibonded layer on SBAP s urface Demoulding Drying and solidifying Deforming Stirring and casting Proportioning Surface dressing S B AP Fig. 5. Manufacturing process of the plates Fig. 6. Mixture of SiC abrasive and SSB in a mould Fig. 7. A manufactured SBAP Fig. 8. SEM micrograph of SBAP 50 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS 4.0 E xperiment To demonstrate the effectiveness of SBAP, a series of experiments are carried out. Larger particles ar e added onto SBAP surface during lapping process, and the s ilicon wafer surface lapped by SBAP is observed to analyze the ‘trap’ effect of SBAP. A silicon wafer lapped by loose abrasive is taken as a comparing sa mple. 4.1 Experimental Setup Experiments are carried on a Nanopoli-100 ultra-pre cision polishing machine. As shown in Fig.9, a piec e of polished wafer is fixed on the workpiece carrier, a nd loaded against SBAP. Large particles are stored in the filler. As the vibrator works, large particles fall onto the SBAP surface, and will be transported int o the lapping zone. The area on the wafer surface lapped with larger particles and the area lapped without l arger particles are observed respectively. In another exp eriments, silicon wafer is lapped by loose abrasive . 1 2 3 4 5 6 1-SBAP, 2-workpiece carrier, 3-clamp of workpiece, 4-filler of larger particles, 5-vibrator, 6-cross b eam Fig. 9. Schematic illustration of SBAP lapping sys tem employed in this study 4.2 Experimental Conditions The experimental conditions in this study are liste d in Table 1. Fine finishing silicon wafer with sur face roughness Ra 0.744nm and Rmax 11.54nm is employed a s workpiece. SBAP used in this study is made of 1000# SiC abrasive. The volume ratio of abrasive, p ore and bond material is 5: 3: 2. During the SBAP l apping experiment, SBAP surface is wetted by de-ionized wa ter. In the loose abrasive lapping experiment, 20% wt 1000# SiC water based slurry is used. Processing pa rameters, such as load and rotational speed of plat e, in the two experiments are same. To compare the effect of large particle on the surface quality in the two experiments, 180# B 4 C abrasives are taken as the larger particle, and a dded into machining zone respectively. During SBAP lapping, the filling rate of larger par ticles is 0.2g/s. In the lapping process, 2%wt B 4 C abrasives are added directly in slurry. Surfaces lapped with or without larger particles ar e observed with SEM (Hitachi S-4700, resolution: 1.5nm/15KV, 2.1nm/1KV) and AFM (Vecco Nanpscope3a E nviroscope, scanning scope: 90 µm×90 µm×5 µm, horizontal resolution 0.2nm, vertical resolution 0. 03nm). Surface roughness is measured with Pethomete r S2 51 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS (Maer Co., accuracy: 0.1nm, Filter: ISO 2CR, cut-of f 0.8 mm and Evaluation length 7 consecutive cut-of f). Material removal rate is measured by a precision el ectronic balance (accuracy: 0.1mg). Table 1: Experimental conditions Items SBAP Lapping Loose Abrasive Lapping Workpiece Ø2 0mm Silicon Wafer, Ra 0.744nm, Rmax 11.54nm Lapping Machine Nanopoli-100 Abrasive 1000 # SiC Added Larger Particle 180# B 4 C Load 15kpa Rotational Speed Of Plate 50rpm Processing Time 10min Abrasive Concentration In Lapping Slurry -- 20% Wt Lapping Plate -- Iron Plate Volume Ratio Of Abrasive, Pore And Bond Material 5: 3: 2 -- 4.3 Results and Discussion Fig.10 and Fig.11 show the SEM and AFM images of si licon wafers lapped with loose abrasive respectivel y. Fig. 12 and Fig. 13 show the SEM and AFM images of silicon wafers lapped by SBAP respectively. It can be judged from the Fig.10 that abrasive grains rolling against the wafer surface is the dominant material removal mechanism during the loose abrasive lapping process . As shown in Fig. 10(b), the surface quality is aggravated by the rolling actions of the larger par ticles added into lapping slurry. Fig.11 shows the effect of the added larger particle on the depth of scratch i n loose abrasive lapping. The maximum depth of the scratch on the wafer surface increased obviously from 208.4 nm to 568.6nm, as the larger particle is added into slurry. The results show that it is difficult to prevent th e surface form the damages caused by the invasion o f lager particles in loose abrasive lapping. (a) Surface lapped without larger particle (b) Surface lapped with added larger parti cle Fig.10. SEM images of lapped surface of silicon wa fer with loose abrasive 52 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS (a) Surface lapped without larger particle (b) Surface lapped with added larger parti cle Maximum depth of the scratch is 208.4nm Maximum depth of the scratch is 568.6nm Fig. 11. AFM images of lapped surface of silicon w afer with loose abrasive (a) Surface lapped without larger particle (b) Surface lapped with added larger particle Fig. 12. SEM images of lapped surface of silicon w afer with SBAP (a) Surface lapped without larger particle (b) Surface lapped with added larger particle (Maximum depth of the scratch is 28.9nm) (Maximum depth of the scratch is 31.7nm) Fig. 13. AFM images of lapped surface of silicon w afer with SBAP It can be judged from Fig.12 that abrasive grains g roove across the wafer surface during SBAP Lapping. It indicates that the retention force of semibond mate rial on SBAP surface is strong enough to retain the grains. It can be seen in Fig.13 that grooves on the surface l apped by SBAP do not become deeper obviously by the added larger particles. The maximum depth of groove in SBAP lapping is about 30nm, and is much lower t han that in loose abrasive lapping. It indicates that t he surface damages, such as deep crack and pit, can be eliminated or reduced by SBAP obviously. The reason is that larger particles are trapped in SBAP surfa ce, and 53 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS the negative effect on surface quality of large par ticles is counteracted by ‘trap’ effect of SBAP as discussed in the section 2. Fig.14 shows the roughness of wafer surface lapped by loose abrasive and SBAP. It can be seen that the surface roughness obtained by loose abrasive lappin g is worse than that obtained by SBAP lapping, and the surface roughness Ra is worsened from 0.7 µm to 3.18 µm by the added larger particles in loose abrasive lapping. On the contrary, the surface roughness Ra changes from 0.02 µm to 0.03 µm as the larger particles added, it indicates that the larger particles have no remarkable effect on the surface roughness in SB AP lapping. loose abrasive SBAP 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Su rfa ce ro u gjn e ss R a (µm ) without larger particles with added larger particles 0.7 3.18 0.02 0.03 loose abrasive SBAP 0 5 10 1 5 2 0 2 5 m a te ria l r e m o a va l r a te (m g/ m in ) without larger particles with added larger particles 6.1 9.2 22.1 2 3.4 Fig.14. Surface roughness in loose abrasive and S BAP lapping Fig.15. MRR in loose abrasive and SBAP lapping Fig. 15 shows the material removal rater (MRR) in t he two processes. As the lager particles added, the MRR increases from 6.1mg/min to 9.2 mg/min in loose abr asive lapping, and MRR increases from 22.1 mg/min t o 23.4 mg/min in SBAP lapping. MRR in loose abrasive lapping is much lower than that in SBAP lapping. T he main reason is that the number of abrasive grains t aking part in action in SBAP lapping is much more t han that in loose abrasive lapping. For the stronger mechani cal action of larger particles, MRR in loose abrasi ve lapping increases as lager particle added. Since th e larger particles are trapped in SBAP, the MRR in SBAP lapping is not affected by the added larger particl es obviously. As the above experimental results show, the effecti veness of ‘trap’ effect of SBAP is demonstrated. Al l of these characteristics of SBAP own to its special st ructure. SBAP has a porosity structure. Grains on t he SBAP surface are semibonded, and can be forced to move f or adjustment under certain load. It is much easy f or the larger particles indenting into the SBAP surface th an they into abrasive lapping plate. It means that the load on the larger particle in SBAP lapping is much lower t han that in loose abrasive lapping. As a result, th e surface damage caused by the larger particle in SBAP lappin g is much less than that in loose abrasive lapping. 5.0 Conclusion • A novel semibonded abrasive plate (SBAP) for finish ing advanced ceramics is developed in this study. Retention force to retain abrasives in the S BAP is intervenient between fixed abrasive and loose abrasive. When a larger particle invades into the machining zone during SBAP lapping process, 54 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 A NOVEL SEMIBONDED ABRASIVE PLATE FOR FINISHING ADV ANCED CERAMICS grains surrounding the larger particle are forced t o move for adjustment, and a ‘trap’ containing the larger particle is formed on the surface layer of S BAP. • A high polymer SSB, which can be solidified as dryi ng and be redissolved by water, is developed to be used as semibond agent. The ‘trap’ effect of SBA P is demonstrated by a series of experiment. The experiment results indicate that surface damage cau sed by larger particles can be eliminated or reduced by the ‘trap’ effect of SBAP. • Since the SBAP is newly developed abrasive tool, fu rther studies, such as design method of SBAP, dressing of SBAP, lapping parameter optimization, a nd testing method for mechanical properties of SBAP should be carried out. Acknowledgement The authors wish to acknowledge the assistance and support of Natural Science Foundation of China (50535040, 50705028). References [1] Yasunaga N, Obara A, Tarumi N. Study of mechanochem ical effect on wear and its application to surface finishing. Res. Electrotech. Lab , 1977, 776: 50-134 [2] Ogita Y, Kobayashi K, Daio H. Photoconductivity cha racterization of silicon wafer mirror-polishing sub surface damage related to gate oxide integrity. Journal of Crystal Growth , 2000, 10(1-3): 36-39 [3] Zhong L, Zhang J, Holland K, et al. A static model for scratches generated during aluminum chemical-m echanical polishing process: Orbital technology, Jpn, J. Appl. Phys., Part 1, 1999, 38: 1932-1938 [4] Basim G, Adler J, Mahajan U, et al. Effect of parti cle size of chemical mechanical polishing slurries for enhanced polishing with minimal defects. J. Electrochem. Soc , 2000, 147(9): 3523-3528 [5] KASAI T, HORIO K, KARAKI-DOY T, et al. Improvement conventional polishing conditions for obtain super smooth surface of compound semiconductor wafers[J], Ann. CIRP , 1990, 39 (1): 297-301 [6] Ohmori H, Nakagawa T. Analysis of mirror surface ge neration of hard and brittle materials by ELID (Ele ctrolytic In- Process Dressing) grinding with superfine grain met allic bond wheels. Ann. CIRP , 1995. 44(1): 287-290 [7] H. Ohmori, N. Itoh, T. Kasai, T. Karaki-Doy, K. Hor io, T. Uno, M. Ishii, Metal-resis bond grindstone a nd method for manufacturing the same, US Patent 620 3 5 8 9 , 2001 55 56 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING DESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIVE SPINDLE FOR DIAMOND TURNING Yohichi Nakao, Masanori Ishikawa Mechanical Engineering, Kanagawa University Abstract This paper describes the angular position controller for a fl uid drive spindle. The fluid drive spindle is designed for the precision machin e tool for producing various precise and complicated surfaces. The fluid drive spindle has the same operational principle with the water drive spindle[1]. Then rotational spee d control system was designed for fluid drive motor that has simplified structu re of the fluid drive spindle. In the present paper, the angular position control system i s designed based on the rotational speed control system. We describes that trans fer function of the angular position control system is the second ordered system. T hus desired response of the angular position control system can be obtained by specifying the natural frequency as well as the damping ratio of the transfer function. Performance of the designed angular position control system is studied via simulations and experiments. The results show that good step response is obtained. Steady state error of the step response is 0.18 degree that is equivalent to the resolution of the rotary encoder used in our experiment. In ad dition, it is verified that the designed controller is capable of compensating the i nfluence of the external constant load torque on the angular positioning accuracy. Keywords: Fluid drive spindle, Water drive spindle, Angular position control, Rotational speed control, Diamond cutting, Ultra-precision machine t ool. 1.0 Introduction The water drive spindle [1] was developed as a spin dle for the ultra-precision machine tool. Features of the water drive spindle are the following: the wate r flow power is utilized for spinning the spindle r otor, and the water static pressure is utilized for suppo rting the rotor. Bend flow channels are formed in t he spindle rotor so that the power of the water flow c an be converted into the driving power for spinning the rotor. Performances of the water drive spindle were studied through experimental and simulation studie s [1]. It is then verified that the spindle rotates 1 0,000 rpm by supplying the water flow of 20 l/min. Rotational speed of the water drive spindle can be controlled by the flow rate that is supplied to the 57 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING spindle [2]. The water drive spindle was also appli ed to diamond cutting experiments. Then fine surfac es with optical quality were successfully obtained [3] . The water drive spindle was designed for diamond cu tting of axisymmetric parts. The surfaces can be produced by simple rotation of the spindle for spin ning workpiece, in addition to rectilinear motion o f the diamond cutting tool. However, the rotational speed decreases as the external load torque due to the cutting forces increases. Accordingly, feedback con troller is needed to design so that spindle speed c an be maintained regardless of cutting forces acts on the spindle. So, rotational speed controller design fo r the water drive spindle was considered by using the flu id drive motor [4] that has similar driving princip le to the water drive spindle. Precision machining operations by the water drive s pindle can be extended if the angular position of t he spindle can be controlled. However, the angular pos ition control of the water drive spindle currently cannot be carried out due to the limitation of the possible rotational direction of the spindle. Thus, a fluid drive spindle is proposed in the present paper so t hat bi-directional motion can be obtained. In the p resent paper, the angular position control design for flui d drive spindle/motor is considered. The angular po sition control system is designed based on the rotational speed control system together with disturbance obse rver [4]. Performances of the designed angular position control system are examined through simulations and experimental studies. The results show that good st ep response of the angular position control system is obtained. Steady state error of the step response w as 0.18 degree that is equivalent to the resolution of the rotary encoder used in the experiments. 2.0 Water Drive Spindle ω Water hydrostatic journal bearing Water hydrostatic thrust bearing Pressurized water Water drive motor Cross section of motor Water hydrostatic journal bearing Fig. 1. Structure of water drive spindle The structure of the water drive spindle is illustr ated in Fig. 1. The spindle rotor is supported by t he water hydrostatic bearings in the radial and axial direct ions. As shown in Fig. 1, the water drive spindle h as bend flow channels, named the exit channels, which are formed at two cross sections of the rotor. The bend channels are designed for converting the angul ar momentum of the water flow into torque for spinning the rotor. In addition to use the water fl ow for lubricant and driving fluid, the water flow can be used as the coolant to prevent thermal deformation of the spindle. Performances of the water drive spi ndle have been experimentally and theoretically studied [3]. Experimental results have verified that runout of the water drive spindle was sufficiently small so t hat it meets the requirement for the diamond cuttin g. 58 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING Then the water drive spindle was tested for diamond cutting. Result of the cutting tests indicates tha t the surface roughness of workpiece is several 10 nanome ters in R a [3]. θ Micro end milling cutter/ grinder Water driven spindle / rotary table ω (a) Machining of asymmetric formed surface θ Water driven spindle Single point diamond tool Workpiece (b) Machining of free form surfaces Fig. 2. Spindle layout and control modes for produc ing various precise parts Fig. 3. Structure of fluid drive spindle 3.0 S pindle for Diamond Cutting/Grinding Figures 2(a) and (b) illustrate possible two contro l modes of the spindle in the diamond cutting/grind ing applications. For instance, Fig. 2(a) illustrates t hat asymmetric formed surfaces can be machined by diamond turning if the displacement of the cutting tool in the infeed direction as well as the angular position of the spindle can be synchronously contro lled. In this case, the rotational speed of the spi ndle Water hydrostatic thrust bearing Water hydrostatic journal bearing Controller Rotary encoder Flow control valve CW CCW (a) Spindle system Exit channels (b) Exit channel (CW rotation) (c) Exit channel (CCW rotation) 59 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING determines the cutting speed. Thus the spindle is u sually operated in higher the rotational speed. Machining of precise free form surfaces or complex surfaces with micro milling/grinding tools is also carried out, as illustrated in Fig. 2(b). In this c ase, it is needed to control the angular position o f the work spindle. In contrast, rotational speed control is n eeded for the spindle that rotating milling/grindin g tools. In this machining process, the cutting/grinding spe ed can be adjusted by the rotational speed of the milling/grinding tools and its size. Thus the work spindle is usually operated in low rotational speed . From these considerations, it is verified that the angul ar position control of the spindle in the various s peed range is required in the precision machining of par ts with asymmetric formed surfaces. D/A Servo Amp. Labyrinth sealAB Servo Valve U/D Rotary encoder Rotor Ball bearing ω A-cross section (CW) ω B-cross section (CCW)B supply p ort A supply port Fig. 4. Structure of fluid drive motor and experime ntal setup Objective of the present paper is to develop fluid drive spindle that is capable of controlling not on ly the rotational speed but also the angular position in t he various range of the speed. Advantage of the spi ndle is that the spindle can be used for various objectives of the precision machining, especially for produci ng small precise parts. In the present paper, design o f the angular position controller is considered by means of the fluid drive motor that was designed for stud ying rotational speed control of the water drive sp indle and the fluid drive spindle. 4.0 Structure of Fluid Drive Spindle Figure 3 illustrates basic structure of the fluid d rive spindle that is designed for conducting variou s machining processes as illustrated in Fig. 2. If th e spindle is intended to be operated in the range o f relatively low rotational speed, low viscosity oil can be used instead of water for driving the spindl e. This is a case if the spindle will be used as rotary tab le as illustrated in Fig. 2, (b). So the spindle wi ll be driven by water or low viscosity oil depending on the appl ications. Thus the spindle is called as the fluid d rive spindle. Driving principle of the spindle is basica lly the same to that of the water drive spindle. Na mely, the fluid drive spindle uses bend channels that are formed at two cross sections of the spindle rotor, which enables the spindle to convert fluid flow power to the torque for spinning the rotor. Unlike the water drive spindle, two independent exit channels are designed in the left and right hand sides of the rotor as s hown in Fig. 3. The independent exit channels, having op posite directions in the bend direction, enable bi- directional rotations of the rotor by switching the supply ports A and B that are connected with the corresponding exit channels A and B, respectively. Thus the angular position control can be carried ou t by designing an appropriate controller. 60 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING Before developing the fluid drive spindle, a fluid drive motor as illustrated in Fig. 4 was designed f or studying the rotational speed controller and angula r position controller. The main difference of the f luid drive motor with the fluid drive spindle is that th e rotor of the fluid drive motor is supported by ba ll bearings, instead of the hydrostatic bearings. In t he present paper, the angular position controller i s designed for the fluid drive motor. In the present study, low viscosity oil, 5×10 -3 Pa ･s, is chosen for driving the fluid drive motor. As shown in Fig. 4, the rotation of the fluid drive motor is measured by a rotary encoder that is conn ected at the end of the rotor. In the experiments, a serv o valve with single stage; the rated flow rate is 2 0 l/min; was used to control flow rate as well as to switch the control ports A and B. Oil temperature control facility was equipped to the hydraulic pump. Oil te mperature was therefore normally controlled to 25 degree Celsius in the serious of the experiments. Fig. 5. Block diagram of rotational speed control s ystem with disturbance observer ˆ lT ωˆ K2 1 1 c sT+ + + kt ˆ dθ －－ －+ Ks Disturbance observer + + K1 1 s ˆθu^ Fig. 6. Block diagram of angular position control s ystem 5.0 Angular Position Control System In our previous study, the feedback control system for the fluid drive motor was designed in order to regulate the rotational speed of the motor regardle ss of the external load torque [4]. In particular, the disturbance observer, which was designed for suppre ssing influence of the external load torque, was introduced. Then the compensator based on the distu rbance observer was included in the feedback contro l system. Block diagram of the feedback control syste m is depicted in Fig. 5. It was verified [4] that t he designed rotational speed controller effectively co mpensates the influence of the external load torque on the motor. Simulation and experimental studies veri fied that desired step response of the fluid drive motor can be obtained by means of the controller. 61 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING The designed rotational speed control system togeth er with the disturbance observer can be used as an inner loop for the angular position control system. Then, the angular position control system is intro duced in this section. Relationship between the input u and angular position θ is depicted in Fig. 6. The angular position feedback loop is then designed and added t o the control system. Assuming that the influence o f the external load torque acting on the motor can be canceled by the disturbance observer as well as th e associated compensator, the transfer function of th e angular position feedback control system is then given by Eq. (1). ( ) 2 2 1 1 2 p n n G s s s ζ ω ω = + + (1) In Eq. (1), ωn and ζ can be represented by Eqs.(2) and (3), respectivel y. 1 s n c K K T ω = (2) ( )2 1 1 11 2 s c s K K T K K ζ = + (3) Here, Tc: the time constant of the spindle, Ks: the total gain of the controlled system. These re lationships can be derived based on the mathematical model [4]. In addition, K1 and K2 are the controller gains to be determined so that desired response can be obtained . So, characteristics of the angular position contr ol system are determined by specifying appropriate ωn and ζ. Then K1 and K2 can be determined by Eqs. (4) and (5). 1 2 c n s T K Kω = (4) 1 2 2 1 sn KK K ζ ω = − (5) 6.0 Performances of Angular Position Control System Design of the angular position feedback control sys tem is carried out by determining the natural frequ ency ωn and damping ratio ζ in Eqs. (4) and (5), respectively. Simulations and experiments were conducted by specifying ωn =50 rad/s and ζ =0.7. Performances of the designed control system were fi rst examined through simulations. In a series of th e simulations, the controlled system composed of the fluid drive motor as well as the servo valve was modeled based on the nonlinear equations [4]. Figur e 7 represents step response of the angular positio n control system that was obtained by simulation. For comparison, the step response of G p(s) is also depicted in Fig. 7. Rise time of the step response in the simulation becomes slightly larger than that of the response of G p(s). This is due to the influences of the nonlinearit ies in the control system. However, if smaller step input was applied into the system, the difference in the rise time can be reduced, as sho wn in Fig. 8. In any case, however, steady state error in the simulation was less than 0.02 degree. Thus, it is verified that the designed controller is capable of controlling the angular position of the fluid driv e motor. 62 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING Performances of the angular position control system were studied through experiments as well. The experimental results are indicated with the simulat ions in Figs. 7 and 8, respectively. In the experim ents, step responses were measured for both rotational di rections. It is verified that the same responses we re observed regardless of the rotational directions. Step responses of the angular position control syst em were also investigated when external load torque was applied to the fluid drive motor. In the experi ments, the constant external torque was applied to the Time sec. 0.0 0.2 0.4 0.6 0.8 1.0 Rot ati on al an gle deg . 0 2 0 4 0 6 0 8 0 1 0 0 ExperimentModel Simulation Time sec. 0.0 0.2 0.4 0.6 0.8 1.0 Rot ati on al an gle deg 0 1 0 2 0 3 0 4 0 Experiment Simulation Model (Eq. (1)) Time sec. 0.0 0.2 0.4 0.6 0.8 Rot ati on al an gle deg 12 14 16 18 20 22 Without disturbance observer under load condition With disturbance observer under no load condition With disturbance observer under load condition Fig. 7. Step response of angular position control s ystem (step size : 90 degree) Fig. 8. Step response of angular position control s ystem (step size : 30 degree) Fig. 9. Influence of external load torque on step r esponse of angular position control system 63 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ESIGN OF ANGULAR POSITION CONTROLLER OF FLUID DRIV E SPINDLE FO R DIAMOND TURNING motor by hanging a weight from the rotor. The step responses of the motor are depicted in Fig. 9. The weight mass of 66.23 g was applied to the rotor in the experiment. As shown in Fig. 9, if the compensa tion by the disturbance observer was not used, the stead y state error in the angular position of the motor becomes significantly remarkable. The steady state error reached -3 degree. In contrast, effectiveness of the disturbance observer is demonstrated because no steady state error can be observed. From the simulation and experimental studies, it is verified that the angular position control of the fluid dri ve motor can be carried out by the designed angular position controller. 7.0 Summary In the present paper, the angular position control for the fluid drive spindle was considered. Perform ances of the angular position controller were examined us ing the fluid drive motor. It was described that va rious precise, free form and complex surfaces can be prod uced by controlling the angular position of the flu id drive spindle. Based on the mathematical model of t he controlled system composed of the fluid drive motor and flow control valve, the angular position control system was designed. It was shown that the closed loop transfer function can be represented by the second ordered system. Consequently, the angul ar position controller design was carried out by speci fying the natural frequency and the damping ratio o f the transfer function. Performances of the angular position control system designed in the present study were investigated through simulations and experiments. The results sh owed that desired step response of the angular position control system was obtained. Steady state error of the step response was 0.18 degree that is equivalent to the resolution of the rotary encoder used in the experiment. In addition, it was verifie d that the designed controller is capable of compensating the influence of the external constant load torque on the angular positioning accuracy. The designed angu lar position controller will be applied to the flui d drive spindle for producing various precise parts. Acknowledgement This research work is financially supported by the Grant-in-Aid for Scientific Research (C) of Japan Society for the Promotion of Science. References [1] Y. Nakao, M. Mimura and F. Kobayashi, Water Ene rgy Drive Spindle Supported by Water Hydrostatic Be aring for Ultra-Precision Machine Tool, Proc. of ASPE 200 3 Annual Meeting, pp. 199-202, (Portland, 2003-10). [2] Y. Nakao, Rotational Speed Control of Water Dri ve Spindle for Diamond Cutting, Proc. International Conference on Manufacturing Research 2007, pp. 54-58, (Leicest er, 2007-9). [3] Y. Nakao and Y. Sagesaka, Diamond Turning using Water Drive Spindle, Proc. of 11 th International Conference on Precision Engineering, pp. 157-161, (Tokyo, 2006 -8). [4] Y. Nakao and M. Ishikawa, Design of Fluid Drive Spindle and its Rotational Speed Control, Proc. of ASME 2008 International Mechanical Congress and Expositi on, CD-ROM, (Seattle, 2007-11). 64 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS MECHANISMS FOR GRINDING AND POLISHING OF SILICON CARBIDE WITH LOOSE ABRASIVE SUB-APERTURE TOOLS H. Cheng 1,2 , L. Ren 1 , Y. Feng 1 , Y. Yam 2 , H. Tong 2 , Y. Wang 1 1. Dept. of Optical Engineering, Beijing Institute of Technology, Beijing, China 2. Dept. of Mechanical and Automation Engineering, the Chinese University of Hong Kong, Shatin, New Territories, Hong Kong Abstract Reaction-bonded silicon carbide component is fabricated by th e reaction bonding of silicon carbide grains with silicon using liquid silicon i nfiltration, whose desirable properties include low density, high hardness and low ther mal coefficient of expansion which make it particularly suitable to meet stringent requ irements such as mirrors in aerospace or high-energy applications. However, the high hardness an d complex crystal-phase structure imply the manufacturing of high prec ision mirrors very difficult. The present study focuses on the processes of computer-controlled small tool loose abrasive manufacturing for silicon carbide material. Here, t he characteristics of material removal rate, surface roughness and surface morphology in gr inding and polishing of silicon carbide using different sub-aperture pads are reported, and how these factors are affected by process parameters such as the machining force, the relative rotation speed, abrasive type and size, etc. are also examin ed in detail, and a summary of findings will be given in the concluding section. Keywords: Silicon carbide, grinding, smoothing, mat erial removal. 1.0 Introduction Controlling such process parameters as the machinin g force, dwelling-time, relative rotation speed and abrasive grit size, etc. in computer-controlled sub -aperture tool optical fabrication can lead to a ch anging in material removal rate, in work-piece surface roughn ess and shapes. Moreover, small pad grinding/polish ing is proven to be an effective strategy for high-accurac y machining of complex shaped optical components ma de of brittle materials (e.g., glass) [1]. Quantitative r esearches for iterative error-correction of optical lens or mirrors have been carried out with a sub-aperture pad, whic h is a conventional precisely shaped rigid pad made of pitch or polyurethane [2],or alternatively, a defor mable tool [3] applies different loose abrasives. T he transfers 65 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS pressure through an abrasive slurry to the entire s urface material of the components. Material is then removed by chemical and mechanical interactions between the abrasive and the components. Silicon carbide materials characteristic as low den sity, high hardness and low thermal coefficient are provoking the interests of researchers, which are v ery good for optical components worked under poor conditions, i.e., with high-energy lasers and aeros pace mirrors [4]. However, the high hardness and co mplex crystal-phase structure of silicon carbide and the labor-intensive nature of mirror fabrication made i t difficult for us to make high-precision components with low c osts. It has been shown that the high-speed lapping using bound pellet-shaped diamond abrasives fixed on a la pping disc and made into a special lapping tool can perform a high efficiency machining for rapid remov al of surface error and initial smoothing [5]. Howe ver, after that, obvious scraping marks are often left o n the work-piece surface, which make a further grin ding process unavoidable in order to remove residual def ects. Therefore, the present study reports on the charact eristics of material removal rate, surface roughnes s and surface morphology in grinding and polishing of sil icon carbide using different sub-aperture pads, and how these factors are affected by process parameters su ch as the machining force, the relative rotation sp eed, abrasive type and size, etc. are also examined in d etail. 2.0 Computer-controlled Loose Abrasive Manufacturing Methods To improve the capability of loose abrasive manufac turing on large aspheric mirrors, a function-expand ed optical manufacturing facility is presented with wh ich it is possible to fabricate high-quality optica l components. The system is based on sub-aperture sca nning techniques, and combines the faculties of gri nding and polishing, has the features of conventional loo se abrasive machining and the characteristics of an automated tool tracking orbits in a planar model. T his has proven to be a reasonable approach and the small pad can fit close to the local curvature of the wor k-piece surface and at the same time, computer cont rol makes grinding and polishing processes repeatable and eff icient [6]. Figs. 1(a) and (b) show an overall structure photog raph and movement schematic view of the CNC machini ng system developed in house, respectively. The system arranged as gantry structure, four pillars, two cr ossbeams and a base form the main frame of the machine. The system is designed to be controlled on four axes. A s the heart of this apparatus, a sub-aperture pad measuri ng 50 millimeters in diameter is screwed on an end face of a principle axis, and is rotationally driven by a mot or via a coupling. In order to follow the work-piec e surface shape and keep better fitting with it as the pad mo ves along the surface, a ball hinge is screwed on a n end of the motor principle axis to combine the upside end of the pad as shown in Fig. 1(c). A tool feed mecha nism consisting of two crossed linear guides, two ball s crews, and two serve motors provide back-forward an d left- right feed of the tool-table. A work-piece holder i s mounted on a Z-axis rotary stage. The manufacturing process is used in an iterative m anner to produce the desired surface. First, the su rface must be tested and providing either digital residua l shape-error data or an interferogram which repres ents the surface errors distribution. The digital surface er ror data are obtained directly by using an interfer ometer or 66 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS surface profiler. Next, the measured surface errors are compared to the errors predicted prior to the last figuring operation. The parameters for the next pro cessing cycle are chosen and a NC code matrix is co mputer generated. According to the prior data, a computer algorithm predicts the expected figure for the next process will be performed, and after the pre-determinist ma nufacturing time the work-piece is tested again and the cycle repeated. X Y Z (a) (b) (c) Ball hinge Motor Work-piece Pad Fig.1. . Photograph (a) and movement representation (b) and schematic illustration of sub-aperture pad machining (c) of the new CNC machine 3.0 E xperimental Details The machining process of a silicon carbide work-pie ce was divided into two stages. Firstly, grinding p rocess was performed to remove the scraping marks on the s urface after rapid lapping and initial smoothing efficiently. Three pads made of cast aluminum alloy , cast iron and silicon carbide were adopted to gri nd the work-piece respectively by using different loose ab rasive particles, say, selenium carbide, boron carb ide and diamond powder. After that, fine error removal and finishing was performed using traditional precisely shaped pads made of pitch and polyurethane and fine diamon d particles. Empirical results were presented to es tablish the validity of the process in terms of efficiency and surface quality. Material removal rates were an alyzed and measured. How these factors are affected by process parameters such as the relative pressure, rotation speed, abrasives grit size, etc. were also examined. 3.1 First Stage of Grinding Silicon carbide specimen was used as the work-piece . A series of experiments were conducted using diff erent pads, loose abrasives, pad rotational speeds, and r elative pressures to study their influences on the grinding performances. Detailed grinding parameters used wer e listed in Table I. 67 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS Firstly, considering the high hardness of silicon c arbide materials, three different pads, i.e., cast aluminum alloy, cast iron and silicon carbide were adopted t o perform the grinding process by using selenium ca rbide particles, boron carbide particles size as W14 (siz e@7~14µ m) and diamond ones size as W7 (size@3.5~7µ m ) respectively. The relative pressure is 3kgf and rot ational speed is 200rpm. The maximum material remov al rate versus the above parameters plotted in Fig. 2(a). O bviously, diamond particles can produce the rapides t material removal, and the cast aluminum alloy pad a nd the cast iron pad have nearly the same removal r ate. When the diamond particle sizes are increased from W7, W10 (size@5~10µ m), W28 (size@20~28µ m) to W40 (size@28~40µ m), the max material removal rates increase greatly as shown in Fig. 2(b). The effect verifies that the diamond abrasives have a good rem oval ability which is applicable for the initial an d fine grinding of the silicon carbide mirrors. (a) Loose Abrasives (b) W7 Cas t iron pad Cast aluminum alloy pad Silicon carbide pad W10 W28 W40 0.0 0.2 0.4 0.6 0.8 1.0 C ast aluminum alloy pad Cast iron pad Silicon carbide pad Boron Carbide Selenium Carbide Diamond Powder Diamond Particle Max imu m Rem ov al Rat e (µ m/m in) Max imu m Rem ov al Rat e (µ m/m in) Fig. 2. Maximum material removal rates of different pads and abrasives. As expected, the silicon carbide pad has the highes t removal rate for its high hardness whereas the pa d is hard to fabricate. The cast aluminum alloy pad is easier to be made and fit well with the work-piece surfac e. Table I. Parameters of Grinding Experiments Parameters Value Pads cast aluminum alloy, cast iron, silicon carbide Size of Diamond Particles W7,W10,W28,W40 Size of Selenium Carbide Particles W14 Size of Boron Carbide Particles W14 Relative Pressure (in kgf) 1,2,3,4,5 68 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS Considering the important influence of the relative pressure and the rotational speed on the material removal, experiments are made using the cast aluminum alloy pad and diamond abrasives size as W7, W10, W28 and W40 under different tool speeds and the relative pr essures. It can be seen from Fig. 3 when the relati ve pressure keeps constant 3kgf and the pad rotational speeds are increased from 50rpm to 300rpm, the rem oval rates increase steadily. But the removal rates decr ease greatly when the rotational speeds are exceede d 300rpm. That means the effect of the rotational speed is si gnificant during grinding, loose abrasive particles would be threw out of the working area when a too high rotat ional speed is used, too less abrasives to perform effectively grinding. Therefore an obviously decrea sed trend can be observed from the removal rate cur ves as shown in Fig. 3. It can also be seen from Fig. 4 th at the removal rate keeps a nearly linear relations hip to the relative pressure when the pad rotational speed kee ps constant 200rpm, and increasing the relative pre ssure between the pad and the work-piece from 1kgf to 5kg f. ) 0 50 100 150 200 250 Tool Rotational Speed (rpm) W7 W10 W28 W40 Max imu m Rem ov al Rat e (µm/ min ) Fig. 3. Maximum material removal rates of different tool rotational speeds. Max imu m Rem ov al Rat e (µm/ min ) . ) Relative Pressure (kgf) W7 W10 W28 W40 0 1 2 3 4 Fig. 4. Maximum material removal rates of different relative pressures. The above experimental results provide a guidance t o choose diamond particles as loose abrasives to gr ind the silicon carbide work-piece. According to experience s, take both the removal rate and the grinding accu racy into consideration, diamond particles size as W40 i s useful to initial grind the surface with a figure error larger than 100µ m (root-mean-square). W28 abrasive is more proper for grinding the surface with an error betw een 30µ m and 100µ m. The AFM result in Fig. 5(a) shows t hat obviously accidented structures residual on the ground surface. Observing the local micro-structure shown in Fig. 5(b), there is a plastic deformation phenomenon can be found and the materials are pushe d to one side, which is different from the brittle removal mechanism. Based on a more reliable explanation, th e high pressure and speed, and then a high temperat ure produced would make a temporal strongly scraping ac tion on the working area, and results in a plastic deformation with large residual stress. Therefore, fine grinding process with a low pressure and speed is necessary. 69 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS (a) (b) Fig. 5. The AFM Photographs of the ground area. 3.2 Second Stage of Polishing In order to carry out proper polishing process to a chieve a higher level surface quality, detailed exp eriments have been made focusing on error-smoothing. The pad s made of pitch and polyurethane, and diamond particles were used to perform the polishing experi ments. When using diamond abrasive sizes as W0.5 (<0.5µm) and under the constant rotational speed 200r pm and relative pressure 3kgf, it is found in Fig. 6(a) the max material removal rate decreases and the sur face roughness increases as the pad hardness increa ses, say from 55# to 91# pitch and to another kind of pad po lyurethane. Therefore, pitch pads with relative low hardness are apt to achieve a high-quality optical surface. Using pad made of 55# pitch to polish, as expect, the removal rate and the surface roughness increase along with the abrasive size increases shown in Fi g. 6(b). From Fig. 6(c) and (d), the material removal and th e surface roughness were found to be sensitive to b oth the speed of the tool and the relative pressure. When t he speed is lower than 250rpm, both the removal rat e and the roughness are keeping steadily, and great chang ing occurred as the rotational speed higher than 25 0rpm. The same case also appears as the pressure larger t han 3kgf. In the above discussion of polishing proc esses, it is important to determine the reasonable range of t he said parameters. Referencing to the above processing experiments, a smoothing optical surface with nanometer accuracy c an be fabricated. From the Veeco interferometer measuring results shown in Fig. 7(a), the surface roughness can achieve to 9.11nm, and the AFM result in Fig. 7(b) shows that the accidented structures have been remo ved and obviously flatting areas produced on the polish ed surface. Max imu m Rem ov al Rat e (µm/ min ) 55 # Pitch 64# Pitch 73# Pitch 82# Pitch 91# Pitch (a) Pad Hardness Removal rate Sur fac e Rou ghn es s (nm)Surface Roughness 70 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS (c) Tool Rotational Speed (rpm) Relative Pressure(kgf) (b) Diamond Particle Size Removal Rate Surface Roughness W0.5 W1.0 W1.5 W2.5 W7 (d) Max imu m Rem ov al Rat e (µ m/m in) Sur fac e Rou ghn es s (nm) Sur fac e Rou ghn es s (nm) Sur fac e Rou ghn es s (nm) Max imu m Rem ov al Rat e (µ m/m in) Max imu m Rem ov al Rat e (µ m/m in) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Removal Rate Surface Roughness Removal Rate Surface Roughness 100 150 200 250 300 Fig. 6. material removal rate and surface roughness along with the changing of (a) the pad hardness; ( b) the diamond particle size; (c) the rotational speed; and (d) th e relative pressure (a) (b) Fig. 7. . Veeco measured roughness result (a) and t he AFM measured microstructure (b) of the polished surface. 71 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MECHANISMS FOR GRINDING AND POLISHING OF SILICON CA RBIDE WITH LOO SE ABRASIVE SUB -APERTURE TOOLS 4.0 Conclusion This paper investigates the mechanisms of loose abr asive machining of silicon carbide components. Experiments of the grinding and polishing for the w ork-piece are performed by virtue of a computer-con trolled small tool manufacturing facility. The changes of t he material removal rate and surface roughness with the varying the relative pressure, rotation speed, abra sive type and size are revealed. According to the e xperiments, different sized diamond particles are effective loo se abrasive for both grinding and polishing of sili con carbide material. Further experimental results and analysis are also helpful for matching process parameters t o achieve a high quality surface, which confirms the validity of the proposed loose abrasive machining approach for grinding and polishing the silicon carbide componen ts. Acknowledgement The authors wish to acknowledge the financial suppo rt of the Natural Science Foundation of China (Gran t No.60644003), the Beijing Nova Program of China (Gr ant No.2006B24) and the Excellent Young Scholars Research Fund of BIT (Grant No.2006Y0101). References [1] Jones RA, Rupp WJ (1991) Rapid optical fabrication with computer-controlled optical surfacing. Optical Engineering 30(1): 1962–1969 [2] Paula G (1997) Automating lens manufacturing. Mecha nical Engineering 119(3): 88–91 [3] Smith BK, Burge JH, Martin HM (1997) Fabrication of large secondary mirrors for astronomical telescope s. Proc of SPIE 3134:51–61 [4] Tobin E (1995) Design, fabrication and test of mete r-class Reaction Bonded SiC mirror blank. Proc of S PIE 2453:12–21 [5] Tam HY, Cheng HB, Wang YW (2007) Removal rate and s urface roughness in the lapping and polishing of RB -SiC optical components. Journal of Materials Processing Technology 192-193: 276 – 280 [6] Cheng HB, Feng ZJ, Cheng K, Wang YW (2005) Design o f a six-axis high precision machine tool and its ap plication in machining aspherical optical mirrors. Internatio nal Journal of Machine Tools & Manufacture 45(9): 1 085 – 1094 72 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-ORIENTED TESTING Christian Brecher, Robert Schmitt, Danny Köllmann, Michael Merz Fraunhofer Institute for Production Technology IPT, Aachen, Germany Abstract Freeform geometries enable new systems in geometrical optics by combining the functionality of different optical components and, thus, reduc ing the required number of components in the system. The main problem in using that kind of freeform shaped optics is the deterministic manufacturing of these geometr ies and their evaluation in measurement. While ultra-precision diamond turning assisted with Fast Tool Servo Systems is shown to be a suitable process to manufacture fre eform geometries for the application of reflective and refractive optical components [1], till recently it was difficult to assess the quality of the manufactured work pi eces in means of form accuracy. While it was known that Fast Tool Servo Systems can posi tion very accurately in the submicron range [2], the measurement of the produced work pieces was another issue. In the first case just the linear posit ioning of the mechanical axis has to be measured – in the latter case, a 3D measurement of a sur face is required. The fringe deflectometry measurement technique shown in this paper fills this gap and gives feedback for implementing corrections in the manufactu ring process. Surface deviations are measured and their reasons can be analyzed. With this information the causes for systematic manufacturing errors can be eliminated or compensated for. The paper exemplarily shows that approach in manufacturing, measuring and correction the production of a freeform mirror for illumination optics. Keywords: ultra-precision, diamond machining, Fast Tool Servo System, freeform optics, fringe deflectometry measurement. 1.0 Introduction The manufacturing of optical components with comple x geometries like freeform surfaces is getting more and more important. Mass production of these optics is enabled by replication methods like injection moldi ng. For the machining of the required mould inserts with fr eeform geometries, ultra-precision technologies lik e Fast Tool Servo turning can be applied to achieve the ap propriate optical surface quality in the range of 1 0 nm Ra and a form accuracy below 1 µ m. Furthermore, the ma chined surfaces need to be measured to guarantee th e 73 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING manufacturing quality. Especially, the measuring of the form accuracy remains challenging regarding th e requested accuracy. NURBS (Non Uniform Rational B-S plines) are widely accepted in industry to describe freeform geometries. The set-point generation for F ast Tool Servo Systems directly based on that mathematical representation has been significantly improving manufacturing accuracy and was shown in t he past. The form testing of aspheric and freeform surfaces, as a part of the high-precision measuring technolo gy, has a high economic impact for the modern industry. The n ew measurement principle of phase measuring deflectometry enables fast, full surface and contac tless measurement of optical surfaces. Compared to conventional optical measuring methods, like interf erometry which is limited to plane, spherical und s lightly aspherical surfaces, the deflectometric testing of almost any geometry is a great advantage. The analy sis of the originated data provides the established performanc e indicators for optical components, like astigmati sm, coma, peak-to-valley as well as a target-actual com parison. In addition, the determination of local sl ope and curvature offers essential information regarding th e optical function. For future developments it is i ntended to carry out a target-actual comparison based on these measurands. Within the scope of this research, reflective demon strator geometries with optical test functions were manufactured to analyze the ability of the deflecto metry measurement system. Furthermore, different techniques to compensate for systematic machining e rrors in form accuracy were tested and compared. Th e freeform measurement system was used to verify the improvement regarding form accuracy. The direct measurement of the geometry unveils manufacturing e rrors which are not visible directly. Sources of su ch errors are for instance effects of systematic error s of clamping devices and of manufacturing with uncontinuous cut. In the process chain of ultra-pre cision machining of freeform geometries in optical quality the approached measurement technique is so far the missing quality feedback to the production system. On the basis of the results shown in this paper, it is env isaged to analyze the manufacturing of optical free form surfaces regarding the occurring disturbances and t o develop applicable performance indicators in orde r to achieve a reliable compensation of systematic error s during the machining. 2.0 Process Chain for Manufacturing Freeform Geometries The process chain for the manufacturing of complex optical components starts with the optical design. The main functions of the design process are the calcul ation of the optical path and the analysis of the o ptical components concerning applicability, producibility and manufacturing tolerances. Besides the homogenei ty of the used materials, the achievable surface roughnes s and form tolerances of optical surfaces are respo nsible for the quality of the whole optical system. The finish ing process for the freeform mirror is Fast Tool as sisted diamond turning. The turning process, which is norm ally only applicable for rotationally-symmetric geometries, is enhanced by a highly dynamic axis (F ast Tool Servo System) moving the diamond tool. The applied aerostatically beared Fast Tool Servo Syste m is able to move the tool in an operational range which is limited by the mechanical stops of the axis to a st roke of 10 mm. The dynamic of the axis is mainly li mited by the maximal electrical power. The uniphase linear m otor is able to drive to approximately 50 µ m at 300 Hz. The basis form of the work piece with 1 mm non-rota tionally symmetric geometry parts is premachined us ing a micro milling process, since the diamond turning process is only able to apply a chip thickness of u p to 20 µ m in a roughing process. The setpoint generatio n in the control system of the milling machine is t rajectory 74 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING oriented. The freeform geometry provided by the CAD (Computer Aided Design) system is split into linea rly interpolated points on the trajectory. The resultin g manufacturing errors are acceptable for the prema chining step. However, the ultra-precision machining with t he Fast Tool Servo System demands for a geometry representation with less interpolation errors, if a sub micron form accuracy is aspired. Furthermore, discontinuities in the setpoints, which are inevita ble with only linear interpolation, lead to non-acc eptable dynamic positioning errors and higher positioning n oise of the axis in the transition zones. To achiev e the demanded manufacturing accuracy even so, the interf ace from the CAD to the Fast Tool Servo System was revised. Freeform geometries in CAD usually are specified in NURBS (Non-Uniform Rational B-Splines). This mathematical description can be used for analytical surfaces which can even incorporate edges as well as for real freeform geometries. For manufacturing purpose s, restrictions regarding the shape of the surface need to be introduced. Especially using the ultra-precision turning production process, the surface slope need s to be in the range of the diamond tool’s cutting edge. The a pplied direct processing of the NURBS data in the F ast Tool Servo System implements access to the complete original surface. The data processing of the Fast Tool Servo System and the basis machine are synchronized by the measured actual axes’ position of the basis machine. In addition to the increased positioning p recision, the provision of a complete surface allow s for advanced pre-control, since the surface can be deri vated analytically without numerical differentiatio n errors. The surface geometry of the first produced prototyp e is measured to analyze and correct systematic err ors in the production chain. 3.0 Measuring Freeform Geometries The characterization of the prototype was carried o ut using a deflectometry measurement system. The measurement device is used in the production enviro nment with the goal of implementing a quality contr ol feedback, improving the production process of optic al surfaces with freeform geometries. Fig. 1. SpecGAGE 3D – Measurement device Fig. 2. Measurement principle [3] Fig. 1 shows the measurement device. The new princi ple of phase measuring deflectometry, shown in Fig. 2, enables non-contact and full surface measurement of optical components. Sinusoidal fringe patterns are projected onto a screen. Their reflections on the s urface of the sample are recorded by a camera. Usin g well- 75 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING known phase-shift algorithms, the system measures t he reflection angle in each pixel of the camera and thus calculates the slope of the surface. The topography is then reconstructed by integrating the local slo pe data and the required information of the local curvature via differentiation. The used measuring system has lat eral and vertical measurement ranges of 80 x 80 mm 2 and ±18° respectively and a lateral resolution of 80 µ m. The range of its applications, due to the measurability of almost any geometry, extends to lenses and mirr ors, like lenses for mobile phones, eyeglass lenses and head- up displays, as well as mold inserts, polished surf aces, wafers and solar cells. 4.0 Data Acquisition and Analysis First of all, a suitable orientation of the prototy pe relative to the camera was determined. To accomp lish high contrast, the angle of elevation was reduced until the reflectivity was sufficient. Afterwards, the me asuring setup was adjusted, so that the visible surface are a under test was at maximum and the occlusion was minimized. For this position, a distance between ca mera and surface was calculated with the objective to accomplish the required lateral resolution and sens itivity for the biggest part of the surface [4]. Af terwards, a comprehensive 3D acquisition of the surface (Fig. 3 ) was carried out. Additionally, the target surface was converted into the measurement data format in order to achieve a target actual comparison. -6.545 mm -5.418 mm 6 545 5 4 1 8 Fig. 3. Actual surface Fig. 4. Matched data wit h cross section The determination of the form deviations via error map or cross sections requires the matching of the target and the actual surface. This was done via a coarse registration on the basis of three selected points and a subsequent fine registration which minimizes the di stances between the actual and the target surface. 3.2 µm Fig. 5. Cross section of matched target and actual surface 76 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING As a result, form deviations of 3 µ m at the center and 20 µ m in areas with steeper slopes were detecte d, as Fig. 5 shows. Influences on production quality orig inate from machining errors from different sources which can be categorized in systematical and stochastic e rrors. Systematic errors of the machine system are geometrical and static alignment errors of the mach ine axes. Especially the highly dynamic Fast Tool S ervo System has dynamical errors, since the working rang e is up to 1 kHz. Long production times combined wi th the thermal behavior of the machine system and the work piece itself lead to thermal errors, which are especially hard to compensate. Stochastic errors, w hich cannot be compensated for, and only minimized or prevented, are effects of the properties of the mac hined material like impurities and chip removal characteristics, but most importantly, environmenta l influences like temperature and humidity should b e held constant. The dependence of the form error on the a ctual slope can be explained by geometrical errors of the diamond tool’s radius shaped cutting edge. This err or mainly affects the shape of the surface in regio ns with high slopes and highly affects the optical function of the surface which firstly depends on the slope in the case of geometric optics. Since this kind of geometrical error is systematic. It can be compensated by incl uding the radius into the calculations of the tool path. 5.0 Compensation of Systematic Machining Errors Fig. 6 on the left schematically shows the form err or without a correction of the tool’s cutting edge geometry, assuming a tool radius of R=0.2 mm and a surface sl ope of the contour of 20°. The correction value dep ends on the local slope of the surface perpendicular to the direction of the tool path. The correction valu e can be extrapolated from the original point. The local slo pe is represented by the normal vector. In the case of a surface encoded in NURBS, the normal vector can be easily evaluated [5]. With these basic conditions, the correction value is ),,(,)cos( )cos(1 ϕρη η η κ NfwithR =−⋅= (1) The form error without compensation always results in the removal of too much material. Thus, the compensation value is always positive, moving the t ool away from the work piece. In the example of Fig . 6, using a tool radius compensation the form error can be significantly reduced from 10 µ m to 1 µ m. 0.6 0.7 0.8 0.2 0.22 0.24 machining w/o correction FT S ax is [mm ] infeed axis [mm] ideal geometry actual geometry 0.6 0.7 0.8 0.2 0.22 0.24 machining with correction FT S ax is [mm ] infeed axis [mm] tool path cutting edge Fig. 6. Form error resulting from the tool geometry regarding first order correction 77 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING 6.0 Measurement of Prototypes Manufactured with Radius Compensation After the renewed manufacturing of the freeform geo metry, another measurement was carried out. As a re sult, the form deviations of the manufactured prototype w ith radius compensation were significantly smaller with <1.6 µ m peak-to-valley for the whole surface. Anoth er advantage is the improved optical quality of the surface which is visible in the measurement data due to the absence of areas with invalid data. -0.46 5 mm 0.577 mm 0 465 0 57 7 Fig. 7. Actual surface after radius compensation F ig. 8. Matched data with cross section Fig. 9. Cross section of matched target and actual surface after renewed manufacturing In addition to provide an analysis of the topograph y (Fig. 7), the measurement system also measures th e slope and curvature of the surface. In contrast to the un distinguishable topography, the slope and curvature supply a more direct visualization of the beam shaping funct ion of the surface. 1,5 µm 78 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING Fig. 10. Slope Fig. 11. Curvature In the future, it is intended to carry out target v s. actual comparison on the basis of slope and curv ature. Currently, the feedback of these correction values is insufficient, due to the fact that the majority of manufacturing machines still use height information as input data. The slope and the curvature informa tion have the advantage of emphasizing the local surface geometry variation included in high spatial freque ncies parts, while downsizing low spatial frequency infor mation which is expressed in the position data and conceals the local variations in matters of scale. with radius correction w/o radius correction Detailed view of illuminance distribution Fig. 12. Illuminance distribution of demonstrator o ptical element Fig. 12 shows the measurement of the illuminance di stribution itself. The illuminance was measured wit h a CCD chip by directly exposing it to the projection without the object lens of the camera. In the left figure, a global view of the function, representing the IPT l ogo, is shown. In the right figure, the enhancement of the radius correction is shown. The illuminance in the fence parts of the logo is distorted and unevenly d istributed, depending on the cutting direction, which is direct ly affected by the tool radius without compensation . Applying the radius compensation, these geometrical errors could be corrected. 79 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 TECHNICAL ADVANCES IN ULTRA-PRECISION MACHINING OF FREEF O RM OPTICS BY FUNCTION-O RIENTED TESTING 7.0 Conclusion The Fraunhofer IPT provides the complete production chain for the manufacturing of optical components with freeform geometries. Both the direct manufacturing of mirrors and lenses as well as the manufacturing of molds including the replication of lenses with inje ction die molding of plastics and compression moldi ng of glass are covered. The CAD data from the optical de sign can be directly and optimally employed in the production process. The quality feedback for the pr oduction chain can be supplied with the fringe deflectometry measurement technique in the case of freeform geometries. The measurement system enables the detection of effects caused by systematic manufactu ring errors like geometrical errors resulting from the actual shape of the diamond tool’s cutting edge and their elimination in the production process step. The functionality of optical components is strongly influenced by the local slope of the geometry. Sin ce the deflectometry measurement technique directly measur es the slope, in the future, this surface informati on can possibly be included into the interpolated surface data and enhances its usability for optical surface measurement. Acknowledgement This research work is part of the cooperative proje ct “ProNanoMess” (reference number: 02PG2703), fund ed by the Federal Ministry of Education and Research ( BMBF), and of the SFB/TR4 “Process Chains for the Replication of Complex Optical Elements“, funded by the German Science Foundation DFG. The 3-D tailore d freeform mirror was designed by OEC AG, Germany. References [1] Brecher, C.; Weck, M.; Winterschladen, M.; Wetter, O.: Manufacturing of Free-Form Surfaces using a Fas t Tool Servo (FTS) and an online Trajectory Generator. In: Proceedings of the ASPE Spring Topical Meeting on Control of Precision Systems, Cambridge, Massachusetts (USA), 19 – 20 April, 2004 [2] Brecher, C.; Merz, M.; Wenzel, C.: Optimisation of Servo Control for Highly Dynamic Axes for Ultra Pre cision Freeform Machining. In: PCIM – International Confer ence for Power Electronics and Intelligent Motion, Nuremberg, 22 – 24 May, 2007 [3] 3D Shape GmbH: Measurement Principle – http://www.3d-shape.de/produkte/pmd_e.php [28.03.2008] [4] Kammel, S.: Deflektometrische Untersuchung spiegeln d reflektierender Freiformflächen. Universitätsverl ag Karlsruhe 2005, p. 38 [5] Brecher, C.; Lange, S.; Merz, M.; Niehaus, F.; Wint erschladen, M.: Off-Axis Machining of NURBS Free-Fo rm Surfaces by Fast Tool Servo assisted turning. In: A nnals of the German Academic Society for Production Engineering (WGP), Vol. XIII, Issue 1, 2006, p. 189 -192 80 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DESIGN OF EXPERIMENTS BASED ABRASIVE W ATER JET CUTTING OF ARAMID COMPOSITES Tauseef Uddin Siddiqui 1, Mukul Shukla 2 1. Research Scholar, Mechanical Engineering Department , MNNIT, Allahabad, India 2. Assistant Professor, Mechanical Engineering Departm ent, MNNIT, Allahabad, India Abstract Abrasive water jet cutting (AWJC) is particularly useful for cutting difficult-to-cut materials like composites, rocks, super alloys, glass and ceramics . However, the produced surface quality is poor particularly towards the jet’s exit s ide, requiring post processing operations. These finishing operations cause furthe r delamination in case of polymer composite laminates. Therefore, optimum selection of process parameters is a major issue in good quality AWJC. In the present research work, a standard L27 (313) orthogonal array based on Taguchi design of experiments is used for carrying out AWJC of aerospace grade aramid composites. The utility concept has been applied for multi-objective optimization of surface roughness and kerf taper. The experimental results and analysis of variance indicate that qual ity level has the most significant influence on multiple kerf quality characteristics. Keywords: Abrasive water jet cutting; Surface rough ness; Kerf taper; Aramid composites; Utility concept. 1.0 Introduction Abrasive water jet machining (AWJM) is one of the recent non-traditional machining process widely used in the industry for machining of difficult-to-cut materials such as composites, Ti alloys, concrete, glass and ceramics. It is a contact less machining process with the distinct advantages of no thermal distortion, high machining versatility and small cutting forces [1]. In AWJM, material removal depends upon the erosion caused by abrasive particles on the work surface. A small stream of abrasive particles is entrained in the water jet so that water jet’s momentum is partly transferred to the abrasive particles. Water as a carrier fluid is used to accelerate abrasive particles to produce a highly coherent AWJ, which is focused on the work piece surface through a nozzle [2]. The schematic of AWJC process is shown in Figure 1. The different material removal 81 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S mechanisms and characteristics of machined surface in AWJM have been presented and discussed by several investigators [2]–[5]. Fig. 1. Schematic diagram of AWJC process Composite materials are increasingly used in the aerospace industry, due to their superior properties like high specific strength, high specific modulus of elasticity, high corrosion resistance and low density. Unlike glass and carbon fiber reinforced plastic (FRP), the machining of aramid FRP laminates is comparatively more difficult due to their higher toughness [6]. But due to their increased use in high performance anti-ballistic and aircraft structural components, an efficient method is necessary for cutting aramid composites. Studies by Konig [7] on cutting of FRPs revealed that the cut surface quality was dependent upon the process parameters such as water jet pressure, abrasive flow rate, nozzle diameter, standoff distance and material thickness. Singh et al. [8] experimentally studied the effect of different process parameters on AWJ cut surface finish for different materials (aluminium, steel, glass and rubber). It was found that better surface finish was obtained on the top part of the cut surface at lower water jet pressure and by increasing abrasive flow rate and decreasing traverse speed. Wang and Guo [9] developed a semi-empirical model for the predictive depth of jet penetration for AWJC in order to achieve through cuts and to eliminate delamination effect in polymer matrix composites. It was found that the depth of penetration decreases with an increase in jet traverse rate and 82 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S increases with water jet pressure and abrasive flow rate however the rate of increase decreases with increase in abrasive flow rate. Azmir and Ahsan [10] conducted Taguchi method (TM) based experiments on AWJM of glass-epoxy composites. It was observed that hydraulic pressure and traverse speed were most significant factors affecting the surface roughness while abrasive flow rate and standoff distance were insignificant. From the literature review it can be concluded that the design of experiments (DOE) based studies on AWJC so far have considered only a single quality characteristic at a time to optimize the process performance. This paper illustrates the application of Taguchi method using the utility concept for multi-objective optimization (MOO). The utility concept [11] employs weighting factors to signal-to-noise (S/N) ratios of each response to obtain a multiple signal-to-noise ratio (MSNR) for each trial of the orthogonal array. In the present work, the above approach is employed for simultaneous minimization of surface roughness (Ra) and kerf taper (KT) in AWJC of aramid composites. 2.0 Selection of Process Parameters and Their Levels In the present work, three process parameters namely water jet pressure, abrasive flow rate and quality level each at three levels were used as shown in Table I. The dimensionless cutting quality level is defined by the mean Ra of the upper, middle and lower zones of the AWJ cut surface. The process parameters and their levels selected were primarily based on AWJ machine constraints and literature review on AWJM of aramid composites. The initial setting of parameters is: Water jet pressure–250 MPa, Abrasive flow rate–250 g/min and Quality level–3. Table I: L27 TOA process parameters and their levels used in Process Parameters Symbol Units Low Medium High Water Jet Pressure (WJP) A MPa 250 300 350 Abrasive Flow Rate (AFR) B g/min 250 325 400 Quality Level (QL) C -- 3 4 5 3.0 AWJC Process Details (Material and Measurement) The OMAX 2652 Machining Centre was used to cut 20 mm long, through cuts in a single pass on 2 mm thick specimen. The orifice diameter (0.33 mm), mixing tube diameter (0.762 mm), standoff distance (3 mm) and garnet abrasive size (80 mesh) were kept as constant fro all te experiments. In the present work, composite laminates have been prepared by the standard autoclave vacuum bagging process using bidirectional aramid- prepregs (VICOTEX 913/50%/K285, fibre volume fraction 0.50) supplied by Hexcel Composites were used. The application of this material is in impact resistant structural components of Dornier transport aircraft. Ra was measured by ‘Stylus profilometer’ (Taylor-Hobson subtronic 10) near the top and bottom surface to avoid the jet striation effect at entry and exit. Two measurements per trial were taken for Ra and their average values used. The TKW and BKW were measured using the Tool Maker’s Microscope at 20X magnification. The KWs taken are the average of two measurements of each cut. The kerf taper (in degree) is calculated as follows: 83 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S Kerf taper = 1tan− [ (Top kerf width - Bottom kerf width)/ (2 × Specimen thickness)] (1) The kerf geometry of a through cut generated by AWJ is shown in Figure 2. Fig. 2. Schematic diagram for kerf geometry 4.0 Taguchi Method TM is an important DOE tool for robust process design and provides a simple and systematic way to optimize design for performance, quality and cost. TM defines the quality of a product, in terms of the loss imparted by the product due to deviation of the product’s functional characteristic from the desired value. The uncontrollable external factors which cause the functional characteristics of a product to deviate from their target values are called noise factors such as vibration, temperature and human factors etc. The overall aim is to design a process that is robust against these noise factors. 4.1 Taguchi Design of Experiments The experimental design based on a Taguchi orthogonal array (TOA) is orthogonal. i.e. the effect of each process parameter at different levels to be separated out. TOA provides an effective way of conducting minimum number of experiments that give full information of all the parameters affecting the two responses. To select an appropriate orthogonal array for conducting the experiments, the total degrees of freedom (dof) are computed as follows [12]: dof = ((number of levels - 1) for each control para meter + (number of levels - 1) for each interaction + 1) (2) In the present case, the dof comes out to be ((3-1)× 3 + (3-1) (3-1)× 3 + (3-1) (3-1) (3-1) +1 = 27). Hence, a standard L27 (313) TOA (27 runs of a maximum of 13 control parameters at 3 levels each) is chosen for the experimental design matrix. The computed S/N ratios and the corresponding MSNR are shown in Table II. 84 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S TableII: S/N ratios of Ra and KT and the computed MSNR S No. S/N ratio (Ra) S/N ratio (KT) MSNR 1 -18.06 -9.73 -13.89 2 -16.26 -8.92 -12.59 3 -13.44 -7.55 -10.50 4 -18.17 -7.59 -12.88 5 -16.26 -6.38 -11.32 6 -13.44 -7.73 -10.59 7 -18.28 -10.24 -14.26 8 -16.65 -7.98 -12.32 9 -13.80 -7.84 -10.82 10 -17.27 -10.28 -13.77 11 -15.42 -9.89 -12.65 12 -12.67 -9.64 -11.16 13 -17.39 -10.65 -14.02 14 -15.42 -9.93 -12.67 15 -12.87 -7.57 -10.22 16 -17.50 -10.72 -14.11 17 -15.56 -10.46 -13.01 18 -12.87 -7.35 -10.11 19 -16.52 -10.43 -13.47 20 -13.44 -10.43 -11.93 21 -11.82 -7.39 -9.60 22 -16.52 -10.43 -13.47 23 -13.63 -11.45 -12.54 24 -12.26 -9.35 -10.80 25 -16.65 -10.61 -13.63 26 -14.15 -11.56 -12.86 27 -12.26 -8.32 -10.29 Mean -15.13 -9.27 -12.04 4.2 Determination of Optimal Process Parameters The optimum level for a factor is the level that results in the highest value of S/N ratio in the experimental design. Generally, there are three categories of quality characteristics for S/N ratio, i.e. smaller-is-better, larger-is-better and nominal-is-better. In the present work, the smaller-is-better quality characteristic is used for both Ra and KT as we intend to minimize them. ANOVA was then conducted to observe the effect of different process parameters on the two responses. From the S/N ratio and ANOVA analysis, the optimal combination of the process parameters can be determined. 5.0 Data Analysis and Results 85 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S 5.1 S/N Ratio The term ‘signal’ represents the desirable value (mean) of the output characteristic and the term ‘noise’ represents the undesirable value (Standard deviation) for the same output. Therefore, S/N ratio is the ratio of the mean to the S.D. Taguchi method uses the S/N ratio to measure the quality character deviation from the desired value. The S/N ratios are computed as follows: iη = −10 log10 ( 2 jy ) (3) where yj is the value of the ith experimental response for jth quality characteristics. In the utility concept [11], the MSNR (η ) is computed as follows: 2211 ww ηηη += (4) where w1 and w2 are the weights associated with S/N ratios of Ra and KT respectively. Equal weights (w1 and w2=0.5) have been assigned to Ra and KT because both are equally important for producing a better AWJ cut surface quality. 5.2 Analysis of Variance ANOVA is performed to investigate the effect of different process parameters on the quality of a process/product and their percentage contribution. The ANOVA response for MSNR is given in Table III and the contribution of different main factors and their first order interactions in decreasing order is QL (65.62%), WJP×QL (12.40%), AFR×QL (9.16%), AFR (5.82%) and WJP (3.94%) as shown in Figure 3. Table III: ANOVA response table for MSNR Factors Sum of Squares Degree of freedom Mean sum of squares Contribution (%) WJP 2.68 2 1.34 3.94 AFR 3.96 2 1.98 5.82 QL 44.63 2 22.32 65.62 WJP×QL 8.43 4 2.11 12.40 AFR×QL 6.23 4 1.56 9.16 WJP×AFR 0.00 4 0.00 0.00 Error 2.08 8 0.260 3.06 Total 68.01 26 2.62 100 86 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S Fig. 3. Percentage contribution of different factors for MSNR 5.3 Confirmation Experiments After the determination of optimal parameter settings, the final step is to predict and verify the improvement of the quality characteristic by conducting confirmation experiments at the optimal settings. The predicted S/N ratio ( )p ∧ η using the optimal levels of the control parameters are computed as follows: )( k 1i mimp ∑ = ∧ −+= ηηηη (5) where mη is the total mean S/N ratio, iη is the average S/N ratio corresponding to the ith control parameter at its optimal level, and k is the number of control parameters that affect the quality characteristic. The results of the confirmation experiments shows an increase of 4.28 dB in MSNR at the optimal settings as compared to the initial settings. A significant reduction in Ra and KT of 3.4 µm and 1.43º respectively is obtained by MOO. Computed increase in individual S/N ratios for Ra and KT is 6.44 dB and 3.11 dB respectively at the optimal settings as compared to initial settings as shown in Table IV. Figure 4 illustrates the response plot for MSNR while Figures 5 and 6 shows S/N ratio response plots for Ra and KT. Table IV: Effects of factor levels on S/N ratios MSNR w1=0.5, w2=0.5 S/N ratio (Surface roughness) S/N ratio ( Kerf taper ) Factors Level 1 Level 2 Level 3 Level 1 Level 2 Level 3 Level 1 Level 2 Level 3 WJP -11.64* -12.41 -12.07 -16.04 -15.22 -14.14* -7.25* -9.61 -10.01 AFR -12.17 -11.52* -12.43 -14.99* -15.1 -15.3 -9.36 -7.94* -9.57 QL -13.48 -12.29 -10.36* -17.37 -15.2 -12.83* -9.59 -9.38 -7.9* Optimum parameters settings A1B2C3 A3B1C3 A1B2C3 Improvement (dB) 4.28 6.44 3.11 *Optimum Level 87 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S Fig. 4. Response plot for MSNR Fig. 5. Response plot for Surface roughness Fig. 6. Response plot for Kerf taper 6.0 Conclusions The following conclusions can be drawn from the results of the present research work: 88 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 O PTIMIZATION OF SURFACE FINISH AND KERF TAPER IN DE SIGN OF EXPERIMENTS BASED ABRASIVE WATER JET CUTTING OF ARAMID COMPOSITE S 1. The optimum parameter settings for simultaneous optimization of Ra and KT are found to be water jet pressure-250 MPa, abrasive flow rate-325 g/min and quality level-5. 2. The contribution of cutting parameters by ANOVA in decreasing order is QL, WJP×QL, AFR×QL, AFR and WJP. 3. This approach will prove to be very useful for the AWJC community for simultaneous optimization of kerf quality characteristics of aramid composites. Acknowledgement The authors gratefully acknowledge the support of IIT Kanpur for extending their AWJM and measurement facilities for the experimental work. The authors also acknowledge HAL, Kanpur for extending their composite part manufacturing facilities and providing the aramid- epoxy composite samples. References [1] G.S. Choi, G.H. Choi, ‘Process analysis and monitoring in abrasive water jet machining of alumina ceramics’, International journal of machine tools & manufacture, 37, 3, 295-307, 1997. [2] M. Hashish, ‘A modeling study of metal cutting with abrasive water jets’, Journal of engineering materials and technology, 106, 88-100, 1984. [3] J.G.A. Bitter, ‘A study of erosion phenomenon’: I and II, Wear, 6, 5-21, 1963. [4] I. Finnie, ‘The mechanism of erosion of ductile metals’, in: Proc. ASME 3rd US National congress of applied mechanics, 527-532, 1958. [5] D. Arola, M. Ramulu, ‘Micro-mechanisms of material removal in abrasive water jet machining’, Processing of advanced materials, 4, 37-47, 1993. [6] A. Rahmah, A.A. Khan, M. Ramulu, ‘A study of abrasive waterjet machining of Kevlar composite’, in: Proc. 12th American water jet conference, 4-F, 1-21, 17-19 Aug., 2003. [7] W. Konig, ‘Machining of fiber reinforced plastics’, Annals of the CIRP, 34, 2, 537–548, 1985. [8] P.J. Singh, W. L. Chen, J. Munoz, ‘Comprehensive evaluation of abrasive waterjet cut surface quality’, in: Proc. 6th American water jet conference, Houston, 139-179, 24-27 Aug., 1991. [9] J. Wang, D.M. Guo, ‘A predictive depth of penetration model for abrasive water jet cutting of polymer matrix composites’, Journal of materials processing technology, 121, 390–394, 2002. [10] A. Azmir, A.K. Ahsan, ‘Investigation on glass/epoxy composite surface machined by abrasive water jet machining’, Journal of materials processing technology, 198, 3, 122-128, 2008. [11] V.N. Gaitonde, S.R. Karnik, J. Paulo Davim, ‘Taguchi multiple-performance characteristics optimization in drilling of medium density fibreboard (MDF) to minimize delamination using utility concept’, Journal of materials processing technology, 196, 73–78, 2008. [12] M.S. Phadke, Quality Engineering Using Robust Design, Prentice Hall, Eaglewood cliffs, New Jersey, 1989. 89 90 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS NEW METHODS FOR ENHANCED STUDY OF THE ELECTROCHEMICAL MACHINING PROCESS Lauren ţiu Sl ătineanu 1 , Margareta Cotea ţă 1 , Anca Dr ăghici 2 , Oana Dodun 1 , Irina Neaga 1 1. Technical University “Gh. Asachi” of Ia şi, Romania 2. “ Politehnica” University of Timi şoara, Romania Abstract Generally, the machining technique using electrochemical er osion is based on the material removal from the workpiece as a consequence of the ch emical reactions developed between the workpiece material and the electrolyt e, in the presence of the electric current. In the case of most of the electrochemi cal machining techniques, the concrete process develops in closed spaces. Therefore, the d irect observing of the material removal from the workpiece and of the gradual forming of the machined surface is not possible. In order to enhance the study of th e electrochemical erosion process, two devices are suggested and discussed in the pape r. The first device is simple yet highly suggestive in illustrating the evolution of t he electrochemical erosion process with natural depassivation. This device is based on th e use of two electrodes (test piece – anode and electrode–tool – cathode) made of thin p lates (0.1...0.3 mm). The electrodes are clamped on the internal surface of the parallelepipedic recipient made of transparent material. If the two electrodes are connec ted to the direct current source and the electrolyte (sodium chloride aqueous solution) is found in the recipient, the electrochemical process is then initiated, one having th e possibility to observe it through the recipient transparent wall. This solution does n ot permit an intense circulation of the electrolyte in the work gap. For this reason , the second device was designed and manufactured. In this case, both electrodes (mad e also of thin plates) are placed between the base piece of insulating material and th e transparent cover. The device could be placed on the guide of the universal lathe t ool slide; thus, the slow mechanical work motion of the electrode tool could be obtained. The system of existing holes in the base piece allows for the forced elec trolyte circulation, due to the presence of a hydraulic pump. The existing gap value between t he electrode-tool and the test piece could be theoretically established and prac tically verified using the two devices. Keywords: electrochemical machining, device, natura l depassivation, hydrodynamic depassivation. 91 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS 1.0 Introduction As part of some specialized directions for the educ ation of future engineers in the field of industria l engineering, at the Technical University “Gh. Asach i” of Ia şi the curriculum contains the subject named “Non- conventional Technologies”. The main objective of t his discipline is to provide a basic understanding of the aspects specific to the research and practice of ce rtain machining techniques by non-conventional meth ods and, at the same time, the development of those com petences able to allow the future engineer to handl e with problems of design and use of non-conventional mach ining methods. Generally, in universities, the non-conventional te chnologies are considered to be technologies based essentially on increasing the energy available in the work zone, through different methods, so that ei ther a traditional machining process develops under better conditions, or the machining process develops on new principles, fundamentally different in comparison w ith the basic principle of the traditional machining process ( the principle of plastic deformation; even the cutting process is based on such a princ iple, because the workpiece material is pressed by the cutting to ol until a shearing phenomenon determines the chip forming). Currently, the non-conventional technolog ies include [1]-[5] the electrical discharge machin ing, the electrochemical machining, the plasma or ion beam m achining, the laser beam machining, the electron be am machining, the water jet machining, the magnetic fi eld machining etc. The actual name of non-conventional technologies is sometimes considered to be not exactly adequate , because in its first stage, each new technology could be regarded as a non-conventional (non-traditional) technology. Therefore, in the future it is possible to consider that the name used by the pres tigious journal CIRP Annals. Manufacturing technolo gies (the name of “Electro-Physical & Chemical Processes ) or the name used within the international symposi a ISEM (name of “Electro machining”) illustrate much better the discipline content. A relatively important group of technologies includ ed within the non-conventional technologies is the group of electrochemical machining methods; of course, the group could include electrochemica l processes which imply the material removal from the workpiece (manu facturing methods by electrochemical erosion), but also some manufacturing methods which involve the additi on of material (there are electrochemical plating o r certain processes that determine a change in the ch emical composition of the surface layer). 2.0 Assessment of the Current Situation The machining methods by electrochemical erosion (u sually known as the electrochemical machining – ECM ) are based on the electrochemical (anodic) dissoluti on of the workpiece material within some characteri stic processes of electrical charges and mass changes be tween the electrolyte, the anode and the cathode. Sometimes, electrochemical machining is considered to be the reverse process of electroplating [5]. Nowadays, electrochemical machining is one of the m ostly used non-traditional machining processes by erosion [6]-[7]. The main characteristic of the pro cess is the metal removal without the use of mechan ical or thermal energy; the high material removal rate is a nother argument to explain the extent of some electrochemical machining processes. As a consequen ce of the dissolution process development, the electrolyte concentration diminishes near the two e lectrodes; the phenomenon is called the polarization of the concentration. At the same time, the products of the chemical rea ctions could gradually generate a layer with a 92 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS decreased conductivity level or even with propertie s corresponding to an insulating material; this lay er appears on the workpiece surfaces affected by electrochemic al erosion. In this way, the so-called passivating film is generated. Usually, the continuous generation and developing o f the passivating film is considered to be an inconvenience, because it determines the reduction of the material removal rate (an exception being the electrochemical surface marking, when precisely the passivating layer is the marking observed). The researchers took this into consideration and finall y found certain ways to diminish the negative effec ts of the passivating film forming or of the presence of the concentration polarization. Figure 1 presents a schematic graphical representat ion which emphasizes the process of material remova l in the case of the electrochemical machining with forc ed hydrodynamic depassivation. Usually, the monographs concerning the non-conventi onal technologies include at least a chapter intend ed to present the knowledge in the field of the electroch emical machining techniques existing at a given mom ent [2]- [5]. electrolyte - + Fig. 1. Material removal during the electrochemical erosion process. More than two hundred years ago, the English scient ists Humphry Davy and Michael Faraday laid the foundations of the science called electrochemistry; nowadays, one considers that the electrochemical machining develops essentially in accordance with F araday’s laws. The main advantages of the electrochemical machining technologies are: a) High material removal rate, especially in the case of the electrochemical machining techniques with forced hy drodynamic depassivation; b) Lack of any thermally or mechanically affected layer; c) Lack of the burrs at the ends of the machined surf aces etc. On the other hand, some disadvantages of the use of electrochemical ma chining methods are the following : 1) A more reduced machining accuracy, in comparison with some other c lassical or non-traditional manufacturing methods; 2) An increased roughness of the machined surfaces and th e possibility for the surface roughness to be influ enced by the chemical composition of the workpiece material; 3) A more difficult electrode tool design (the work g ap value is not the same along the machined surface; t ogether with many other factors, a more intense ele ctrolyte circulation can influence the material removal rate from the workpiece) etc. As previously mentioned, at the Technical Universit y “Gh. Asachi” of Ia şi, the curricula valid in the case of some so-called specializations in the field of indu strial or mechanical engineering include a compulso ry or optional subject of non-conventional technologies; to this discipline, 2 classes of course (lectures ) and 1 or 2 93 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS classes of applications are allocated on a weekly b asis, for a period of one semester (14 weeks). Und er such conditions, at least an application activity (2 cla sses) is intended to the study of the aspects conce rning the electrochemical machining methods. It is known that in the case of most of the electro chemical machining techniques, the concrete process develops in closed spaces. Therefore, the direct ob serving of the material removal from the workpiece and of the gradual forming of the machined surface is not possible, therefore this direct observing of the ma chining process can be performed in the case of some classi cal cutting machining methods (turning, milling, sl otting etc.). The didactic staff considered that it would be useful to identify practical solutions able to a llow the students the possibility to directly observe the ev olution of the electrochemical erosion process, for different theoretical considerations and relations to be thus verified. The study of the specialty literature emphasized th e existence of less information able to permit the fulfilment of the above mentioned desideratum. Usually, differ ent schematic representations are used to explain t he material removal in the case of the electrochemical machining, but without specifying a simple way to verify or study in detail the phenomena specific to this m achining method. Under such circumstances, by takin g into consideration the information available in the spec ialty literature and the authors’ own experience, t wo devices were designed and built. 3.0 Device for the Study of the Electrochemical Machining Proces s with Natural Depassivation A simple yet suggestive variant to illustrate the w ay to develop the electrochemical erosion process w ith natural depassivation is presented in Fig. 2. As an ode, one uses the test piece (the workpiece) 1 of thick steel sheet (having a thickness of 0.2 – 0.3 mm). The tes t piece 1 can also be made of other metallic materials, whose behaviour during the electrochemical erosion process must be studied. The electrode tool (the cathode) 2 is made of copper thick sheet (the thickness being also of 0.2-0.3 mm). Both electrodes (the test piece and the electrode tool) are clamped on one of the vertical transparent wall s of a parallelepipedic tank 3 . The electrodes’ clamping is ensured by means of t wo tongs 4 and 5 of insulating rigid material; the tongs ends, which are in contact with the electrodes, are made of rubber, to avoid the e lectrodes slipping or damage. The two electrodes 1 and 2 are connected to a direct current source, by means of two additional metallic tongs and connection cables. The electrodes 1 and 2 are partially immersed in an electrolyte which can be, for example, the aqueous solution of sodium chloride. F or a better and direct following of the machined su rface shape and for material removal only from a certain test piece surface, both the electrode tool 2 and the test piece 1 are covered with a transparent insulating layer on the surfaces, which in a first stage should not be affected by the electrochemical erosion process. On ly the electrodes’ narrow surfaces placed face to f ace are not covered by the insulating layer and thus they d irectly ensure the conditions necessary for the mat erial removal from the test piece during the electrochemi cal erosion process with natural depassivation. Thi s depassivation is possible as a consequence of the h ydrogen bubbles generation and their motion to the upper 94 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS free surface of the electrolyte. Nevertheless, it i s necessary to mention that during the machined sur face evolution, the maintaining of the insulating film m akes the stirring and the circulation of the electr olyte near the test piece exposed surface more difficult, as a consequence of the generation and motion of hydrog en bubbles. +- V A Fig. 2. Device for the illustration of the electroc hemical machining process with natural depassivatio n. There is no motion between the electrodes, so there is no so-called work motion. Because both electrodes are made of thick metallic sheets, the time necessary f or the experimental test is reduced to only a few m inutes and there is the direct possibility to see how the mach ined surface develops. The factors which can be cha nged in the case of an extended experimental test are the f ollowing: the type of test piece material, the test piece thickness, the shape and the dimensions of the acti ve part belonging to the electrode tool, the chemic al composition and the concentration of the electrolyt e, the voltage applied to the electrodes etc. The a ctive zone of the electrode tool can have different plane shap es (angular shape, with different values of the ang le, curvilinear shape, a concatenation of curve and str aight lines etc.), by taking into consideration the phenomena and the effects that one intends to study. An interesting aspect was emphasized by the uninten ded existence of some pores in the insulating trans parent film which covers the test piece lateral surface; d ue to the presence of the electric field, a faster dissolution of the workpiece material in the non-protected zone ge nerated a penetrated hole, having reduced enough er rors from the circular shape. 1 2 3 4 5 95 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS 4.0 Device for the Study of the Electrochemical Erosion Process with Forced Hydrodynamic Depassivation The device above presented did not permit the illus tration and the study of the electrochemical erosio n process with forced electrochemical depassivation; the only possibility for electrolyte stirring or for breaki ng and removing the passivating film was based on the gene ration and motion of the hydrogen bubbles. Fig. 4. The gap evolution during the electrochemica l erosion process with hydrodynamic forced depassiv ation. In order to illustrate the electrochemical erosion process by the use of the hydrodynamic depassivatio n, the device schematically presented in Fig. 3 was design ed and built. Essentially, there is a parallelepipe dic body 1 made of insulating material, in which a cavity perm itting the placement of the thin test piece 2 was previously designed. The electrode tool 3 , made of copper thin sheet, can have a fixed posit ion, but it may also have a Fig. 3. Device for the illustration of the electroc hemical erosion process with forced hydrodynamic de passivation. 2 1 3 9 4 6 5 10 8 7 electrolyte 96 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS rectilinear slow motion. The cavity existing in the device body 1 can be closed by means of a cover made of insulating transparent material; this transparency permits the observing of the gap size evolution and of the surface shape developing gradually as a consequence of the electrochemical erosion process (Fig. 4). T he electrode tool 3 is clamped at one of its ends by means of two scre ws 4 and 5 in a slit existing at the end of the bar 6, which can move along its axis. The bar motio n occurs into a sleeve 7 that screws in the metallic wall 8 bordering the machining space. Fig. 5 Theoretical modelling of the machined surfac e evolution during the electrochemical machining pr ocess. To ensure only a rectilinear motion to bar 6 and thus to avoid its rotation, at its end there i s a flange 9 , which can be moved along a guiding rod 1 0 . The device could be placed on the guide of the un iversal lathe tool slide; thus, the slow mechanical work motion of the electrode tool could be obtained. The system of ho les existing in the base piece permits the forced elect rolyte circulation, due to the presence of a hydrau lic pump. 5.0 Modelling of the Gap Size Evolution To theoretically model the gap size evolution, the scheme presented in Fig. 5 could be taken into consideration; the coordinate system x Oy has the origin at the intersection of the symmetry axis of the electrode tool profile (an angular shape of the el ectrode tool active zone is considered) and the lin e corresponding to the initial surface of the test pi ece. The active zone of the electrode-tool has a sy mmetric angular shape, characterized by the angle 2 α. In the case of electrochemical machining without work motion of the electrodes, after a duration t of electrochemical erosion, the machined profile c orresponds [8] to the following relation: 2 02 sCts += (1) where s 0 is the initial gap size and the constant C is given by: ( ) mF kUUAC δ ∆− = (2) A being the atomic mass of the test piece material, U – the voltage applied to the two electrodes, U pol – the voltage for the two electrodes polarization, k – the electrochemical equivalent of the test piece material, F – Faraday’s constant, δm – the test piece material density. - + electrolyte O x y 97 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS The initial gap size s0 is practically determined by the minimum distance s0 between the electrodes and the distance ∆s, generated by the inclined position of the electrode tool prof ile: sss ∆+= 0 (3) By taking into consideration the geometrical aspects, the distance ∆s can be expressed [9] as a function of the tangent to the angle α: αtg x s =∆ (4) Thus, the initial gap size s0 becomes: αtg x ss += min (5) On the other hand, the profile of the surface resulted as a conseq uence of the electrochemical machining process is: 0ssy −= (6) By considering the above written relations (1), (5), (6), t he function y can be conveyed as follows: ss tg x sCty ∆−−    ++= min 2 min2 α (7) The relation (7) provides an image concerning the t heoretical influence exerted by some factors on the profile of the machined surface (the time t, the angle α, the initial gap size smin etc.). 0 5 10 15 20 0 1 2 3 4 5 6 7 8 x, mm y, m m Theoretical values Experimental values Fig. 6 Theoretical and experimental values of y (α=90°, I=1.4 A, A=28, U -∆U =40 V, k=0.2, F=1608 A ·min, ρm=7.8 g/cm 3 , smin=0.44 mm, t=5 min). In order to verify the validity of the mathematical model represented by the relation (7), the second device (with forced hydrodynamic depassivation) was used. A photography of the obtained profile after a machi ning time t=5 min allowed one to measure the value y along the distance x =8 mm. The experimental and the theoretical values of y were graphically represented in Fig. 6. The more r educed correspondence between the theoretical and experimental values can be noticed for smaller values of x , when the real gap is smaller than the theoretical gap; this means that in reality the dissolution phenomenon is less intense than that s upposed by 98 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 N EW METHODS FOR ENHANCED STUDY OF THE ELECTROCH EMICA L MACHINING PRO CESS the relation (7). The fact could be explained, for example, by the diminution of the electrolyte conce ntration along the gap, by the more intense bubbles generati on in this zone etc. 6.0 Conclusion The electrochemical machining process develops usua lly in closed spaces; therefore, the work gap size evolution can not be directly observed. Two solutio ns to see how the machined surface develops as a consequence of the electrochemical machining proces s were proposed and applied. The first solution use s two electrodes made of thin metallic sheets, connected to a direct current source and partially immersed i n an electrolyte; the depassivation is determined throug h the electrolyte stirring as a consequence of the generation of hydrogen bubbles and their motion to the free su rface of the electrolyte. Using a hydrodynamic forc ed depassivation as a second solution, the two electro des were placed in a device having a transparent co ver. The electrolyte circulates through the gap size due to the presence of the hydraulic pump. The experiments proved the possibility to use both devices so as to direct ly observe the evolution of the machined surface du ring the electrochemical erosion process and to verify the v alidity of some theoretical considerations. A bette r understanding of the electrochemical erosion proces s by the students is possible through the use of th e two devices built. Acknowledgement The authors wish to acknowledge the assistance and support of the Romanian Managerial Agency of Scienti fic Research, Innovation and Technological Transfer and to the Romanian National University Research Counc il. References [1] Y.L., Yao, G.J. Cheng, S. Feiner, W. Zhang, K.P. Ra jurkar, R. Kovacevic. A web – based curriculum deve lopment on non-traditional manufacturing with interactive f eatures. International Journal of Engineering Education, Vol. 21, 3, pp. 546-554, 2005. [2] C. Sommer. Non-Traditional Machining Handbook. Houston: Advance Publishing, Inc., pp. 215-221, 2 000. [3] W. König. Studium und Praxis. Fertigungsverfahren. Bd. 3. Abt ragen. Düsseldorf, GW: VDI Verlag, pp. 69-118, 1990. [4] J.A. McGeough . Advanced methods of machining . London-New York: Chapman and Hall, pp. 55-88, 198 8. [5] L. Sl ătineanu, G. Nagî ţ, O. Dodun, M. Cotea ţă, F. Chinesta, A. Gonçalves-Coelho, J. Pamies Teixe ira, M. San Juan, L. Santo, F. Santos. Non-traditional Manufacturing Processes. Kishinew (Republic of Moldova): Editura Tehnica Info, pp. 61-110, 2004. [6] M. Datta. Micromachining by electrochemical dissolu tion. In McGeough, J. (ed.). Micromachining of Engineering Materials. New York – Basel: Marcel Dekker, Inc., pp. 239-27 6, 2002. [7] K.P. Rajurkar, D. Zhu, J.A. McGeough, J. Kozak, A. De Silva. New developments in Electro-Chemical Mach ining. Annals of the CIRP , Vol. 48, 2, pp. 567-579, 1999. [8] A.E. De Barr, D.A. Oliver . Electrochemical Machining (in Russian; translation in English language). Moscow: Mashinostroenie, pp. 49, 1973. [9] M. Cotea ţă, L. Sl ătineanu. Some Considerations concerning the Gap Evo lution during the Electrochemical Machining. Annals of DAAAM for 2007 & Proceedings of the 18th International DAAAM Symposium (Editor B. Katalinic), Published by DAAAM International, Wien, pp. 177-178 , 2007. 99 100 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER USE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICAL AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR- COO LED DIESEL ENGINE CYLINDER Salah H.R. Ali 1 , Hassan H.Dadoura 2 , and M.Kamal Bedewy 3 1. Engineering & Surface Metrology, National Institute of Standards, Giza (12 2 1 1 ) , Egypt SalahAli20@yahoo.com 2. Automotive Eng., Faculty of Engineering, Helwan Un iv., Mataria, Cairo (117 1 8 ) , Egypt 3. Mechanical Design & Prod. Eng., Faculty of Engineer ing, Cairo Univ., Giza (12 6 1 3 ) , Egypt KBedewy@yahoo.com Abstract Micro-scale measurements of geometrical and dimensional featur es of components require an advanced precise and accurate device such as the CMM machine. Evolved changes in the geometrical and dimensional measurements as referred to benchmark values can be employed as a reliable diagnostic tool in monitoring the functional deterioration of mechanical parts that involve working surfaces d uring their operation. It is evident that excessive wear in a cylinder bore of an internal combustion engine can dramatically affect the quality of perform ance, the sealing function, the scheme of lubrication, and eventually the servic e life span of the piston rings and in turn of the engine as a whole. In this work, precise and accurate measurements of evolved de viations in roundness, straightness, and concentricity in a cylinder bore of an air cool ed Automotive Diesel Engine using a CMM machine have been executed and analyzed. The results have been presented, discussed, and interpreted in order to d emonstrate making use of them in monitoring the status of the engine during operation. Loc ations of severe wear occurrence in the cylinder bore are then detected and in vestigated. The measurements within relevant uncertainties would reflect the quality of the engine performance, the suitability of the applied scheduled maintenance plan, and t he adverse operating conditions which may have been probably encountered during ser vice life. Thus, in the light of the findings, recommendations can be provided to the engine designer to improve his design regarding changes of material selection and/or surface treatments. 101 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER Furthermore, an innovative constructional modification may be sugges ted to homogenize the wear occurrence in the cylinder bore during oper ation. For instance, a device may be added to the construction in order to cause cont inuous slow rotation of the cylinder about its geometrical axis while the engine is running, without having to dismantle the components. This may extend the operating life span of the cylinder and in turn reduces the maintenance expenses. In addition, powe r loss due to friction and wear in the engine may be favorably affected. Keywords: Dimensional metrology, surface geometry, uncertainty, diesel engine, friction and wear. 1.0 Introduction Accurate geometrical and dimensional measurements u sing precision devices are crucial during the manufacturing processes of parts to insure their co mpliance with the design requirements. In addition, those measurements may also be employed with reference to their benchmark values to monitor the extent and severity of functional deterioration of the parts, especially those working with their surfaces during service. This helps the maintenance engineer take proper dec isions regarding his forthcoming maintenance plan a nd/or repair actions. Thus, the durability and reliabilit y of the parts and the assembly would be favorably affected. Air-cooled Diesel engine, for instance, is commonly used in heavy duty transport fleets applications d ue to their high performance, efficiency, and low fuel co nsumption. The surface contact problems between cyl inders and pistons through their rings are vital to the en gine performance within the adverse operating condi tions of high pressure, temperature rise, and high relative velocity of the contacting surfaces [1]-[2]. Fine f inish and surface treatment together with proper geometrical and dimensional tolerances standards implementation are required in order to ensure good sealing between cy linder wall and piston rings, good load-carrying ca pacity, good lubrication conditions, less friction, suitabl e wear resistance, low translated vibration levels, high engine efficiency, and longer service life span [3]-[4]. T he main function of the piston rings assembly is to provide a good dynamic sealing between combustion chamber and crankcase during compression and power strokes. Reasonable sealing minimizes power loss due to char ge escape from the combustion chamber within suitab le ring expansion gap and limited friction force. For long sealing service life, friction and wear betwee n piston rings and cylinder wall have to be properly control led [5]. They are controlled by lubrication of the interface with dry lubrication of cylinder bore material comp osition besides an oil film thick enough to separat e the asperities of piston rings and cylinder surface [2] -[5]. The friction loss varies according to piston velocity between top dead center (TDC) and bottom dead cente r (BDC), where the oil film thickness depends on th e instantaneous relative velocity of the piston ring, which varies from zero at TDC and BDC to a maximum in the midstroke section. This means that wear conditi ons will vary along the piston ring traveling dista nce, from mild to severe [6]. Normally the cylinder bore is not perfectly cylindr ical along its entire length. Practically, the bore distortion causes loss of conformity between piston rings and cylinder wall which in turn produces some troubles to oil film distribution. Variation in the oil film thickn ess exposes piston rings and cylinder to the whole spectrum of 102 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER lubrication regimes, from mixed and probably elasto hydrodynamic to full film hydrodynamic lubrication [4]- [5]. Consequently, different wear mechanisms will d evelop geometrical departures in transverse section s along the cylinder bore [6]. TDC location on the bore suf fers heavily from oil starvation more than that at the BDC and its vicinity. Although the piston at both locat ions are kinematically characterized by marginal in version velocity situations where it reaches zero before st arting to get inverted, the most severe wear is exp ected to appear at the TDC due to the oil shortage while at the BDC the oil is available either from the source or due to gravity. However, the BDC may also experience high wear rate due to the existence of hard grit and wea r debris accumulated by the gravity at this location and the neighboring area. The midstroke location an d the nearby zone, where the piston velocity reaches its maximum value, mild wear only is expected because t he oil film becomes dynamically thick enough to separate t he mating solid surfaces and prevent metal-to-metal contact [7]. Although there are many new advanced inspection equ ipment such as CMM machines of which their use is s o far only monopolized to the manufacturing fields [8 ]-[9], rare published research work yet exists in t he use of such advanced CMM metrology utilities in the field of engine health monitoring through geometrical dep arture measurements and analysis. Characterization of engine cylinder bore geometry a nd dimensions is a two manifold problem. The first is related to the applied techniques and quality stand ards adopted during manufacturing inspection proces s. This concerns the prescribed surface design parameters s uch as dimensional and geometrical tolerances, and surface roughness. The second is related to processing such data with the purpose of monitoring the changes th at happened to the surface geometry and dimensions dur ing engine service life span. This would help in tw o aspects: the first is related to maintenance decisi ons, while the second is related to design modifica tions. Research work has been done on surfaces with Gaussi an distribution roughness, but cylinder wall fine f inished surface with specified geometrical features and pro perties participate simultaneously together to cont rolling the environment that critically affects the engine functional performance and life [10]-[11]. Although the specified surface parameters represent advanced fea tures, their definition is generally unrelated to a ny physical or mathematical properties of the surface topograph y [12]. The plotted accumulation of surface asperit ies heights according to the Gaussian distribution appe ars as straight line scales. For transitional surfa ce topography, such a scale appears as two intersectin g straight lines. The slopes of the lines are propo rtional to the standard deviations of the two distributions, w hile the point of intersection represents the depth of transition from one finish to another. Difficulties encountered using this technique to apply, has rec ently solved with developing advanced calculations softwa re [8]. On the other hand, the numerical description of the changes in the operating surface geometry during s ervice life span necessitates detecting and follow up the surface geometrical deviations. However, some chang es occur in such a way that a band of surface fine wav elength may disappear. Hence, Fourier Transformatio n Analysis is needed in this case to determine the su rface power spectrum response of special software t o characterize the changes in the surface straightnes s and roundness relevant to operation environment c hanges [7]. Statistical calculation analysis of standard uncert ainty (type U A ) is also needed for CMM measurements [13]. The purpose of this work is to demonstrate employin g the accurate precise surface geometrical and 103 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER dimensional measurements to monitor and follow up t he extent of severity of wear changes in a worn out cylinder of an Automotive Diesel Engine as related to the resulted geometrical distortions in both tr ansverse directions (out-of-roundness, and derived concentri city), and longitudinal directions (out-of-straight ness). Thus, design improvements and/or correction actions to the scheduled maintenance plan could be suggest ed in the light of the analysis of the obtained measureme nts within the relevant uncertainties. Innovative d esign modification and inspired ideas may also be pointed at for the sake of extending the engine service li fe span and minimizing the running operational and maintena nce expenses. 2.0 Cylinder Forces and Surface Measurements 2.1 Dynamic Friction Forces Fig. 1. Forces acting on the cylinder bore Combustion gas pressure represents the essential ax ial force acting on the piston crown area to move i t downwards against reciprocating mass inertia. Fn is the instantaneous sum of the normal acting forces on piston pin, Fig. 1. Reciprocating piston motion on angular movable connecting rod generates a variable piston side force Fs. An axial transmitted force Fa of the crankshaft due to clutch engagement force and timi ng gear force components affect the cylinder wall. The resu ltant of piston forces Fs and Fa attacks the wall a t an angle with Fs. The angle value varies as a function of th e force amplitude to generate a resultant force cau sing rotation around the cylinder axis. Dynamic friction force Ff has been produced due to relative motion of piston rings with respect to the cylinder wall under the e ffect of the resultant force in a spiral like motio n. This causes the cylinder bore to wear at rates corresponding to the resultant force amplitude and direction to gen erate eventually a cylinder out-of-roundness (OOR) and ou t-of-straightness (OOS). 2.2 Surface Geometry Measurements Geometrical and dimensional characteristics of the cylinder bore surface have been measured using a computerized Coordinate Measuring Machine (CMM) equ ipped with a contact scanning probe and a Least Square (LSQ) computing algorithm. The CMM used thro ughout this work was Carl Ziess bridge model available at the Engineering and Surface Metrology Lab, Precision Division, Egyptian National Institut e of Standards (NIS). It is capable of producing accurate results with a reasonable repeatability and 104 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER reproducibility for the surface geometrical departu re features. The maximum permissible specific error value of the used CMM machine can be judged using the fol lowing equation: MPE E = ± {0.9 µm + (L/350)}, µm (1) Where L is the measured length in mm. The CMM measurement performance was verified accord ing to ISO-10360 [14]. An experimental investigation has been conducted on an air-cooled D iesel engine cylinder made of high quality grey cas t iron (GG 25) having initially a design diameter of 110 m m and configuration shown in Fig.2a. The chemical analysis and mechanical properties of the cylinder material are presented in Table I, where HB is the Brinell hardness and σt is the tensile strength. The piston stroke is 140 m m. Table I: Cylinder material specifications Chemical analysis, wt. % Mechanical properties C Si Mn P S Max Cr Ni HB σt, MPa 3.10 2.10 0.65 0.30 0.10 0.20 0.32 220 Min. 220 (a) Cylinder configuration (b) Locations for roundness and straightness measurements Fig. 2. Engine cylinder configuration and locations of measurements A straight Stylus tungsten carbide shaft probe with a ruby tip attached to PRISMO CMM machine was used to quantify the surface geometric and dimension dep arture characteristics of the cylinder bore. The CM M traveling speed was 40 mm/s and the probe scanning speed was 10 mm/s during measurements. The straightness measurements were carried out along fo ur longitudinal equispaced locations, 90 o apart around the circumference, at 1, 2, 3, and 4 as indicated in Fi g.2b. Cylinder bore roundness quantification was co nducted at sections I, II, III and IV nearby TDC, midway, a nd BDC planes as shown in Fig. 2b. The surface geometrical and dimensional features were represent ed by mean average values of five repeated test measurements. 3.0 Uncertainty Assessment of Measurements The mean average values and uncertainty of roundnes s and straightness measurements for the engine cyli nder bore have been presented in Table II. Where M AV is the mean value of five repeated test measuremen ts, S D is the standard deviation, and U A is the uncertainty due to measurement repeatabilit y (U A =S D /√n), where n is the 105 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER number of repeated tests for each target measuremen t [13]. It worth mentioning that type B source of uncertainty is not accounted for by U A values because of its relative insignificance to t he OOR and OOS. The accuracy and uncertainty of these measurements have been determined and found to be within the accepta ble standard limits. Table II: Measurements of both roundness and straig htness together with the uncertainty assessment Tests Points Test 1 Test 2 Test 3 Test 4 Test 5 MAV SD UA 1. Roundness, µm @ circle I 91.0 90.9 90.9 90.9 91.0 90.94 0.054 8 0.024 5 @ circle II 32.1 31.8 32.1 31.9 31.80 31.94 0.115 2 0.067 8 @ circle III 23.0 23.5 24.2 24.0 24.3 23.80 0.543 1 0.242 9 @ circle IV 18.2 18.3 18.3 18.5 18.9 18.44 0.279 3 0.124 9 2. Straightness, µm @ line 1 71.0 71.0 70.3 70.5 70.1 70.58 0.408 7 0.182 8 @ line 2 54.4 54.1 54.3 54.5 54.5 54.36 0.167 3 0.074 8 @ line 3 13.6 13.8 13.7 13.6 13.7 13.68 0.083 7 0.034 7 @ line 4 34.6 34.7 34.8 34.8 34.9 34.76 0.114 0 0.051 0 0 10 20 30 40 50 60 70 80 90 100 I II I II I V Cyl ind er Cir cle s Roundness, µm Test 1 Test 2 Test 3 Test 4 Test 5 Average Fig. 3. Bore roundness measurements at four transve rse sections along piston stroke. 0 10 20 30 40 50 60 70 80 1 2 3 4 Cylinder wall profile St ra ig ht n e ss , µm Test 1 Test 2 Test 3 Test 4 Test 5 Average Fig. 4. Bore straightness measurements at four long itudinal equispaced locations. 106 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER 0.00 0.05 0.10 0.15 0.20 0.25 Circle I Line 1 Circle II Line 2 Circle III Line 3 Circle IV Line 4 UA , µm Straightness Roundness Fig. 5. Uncertainty values of five repeatable test s of OOR and OOS The roundness and straightness results of five repe ated laboratory tests conducted on each one of the adopted four transverse sections I, II, III, and IV, and th e four longitudinal profiles 1, 2, 3, and 4, have b een processed and presented in Fig. 3 and Fig. 4, respectively. The results are also tabulated in table II and the calculated values of the relevant uncertainty are plotted in F ig. 5. 4.0 Results and Discussion Average run out measurements of worn cylinder bore; roundness, straightness, and concentricity, have b een conducted using accurate stylus surface scanning te chnique on a programmable CMM machine. The concentricity is represented by the relative roundn ess run out at the selected transverse sections I, II, III, and IV with respect to circle I taken as a datum as sho wn in Fig.2b. RMS averaged values of five similar a rrays of measurements have been considered. The results have been presented, discussed, and interpreted. 4.1 Out-Of-Roundness Measurement Results Average out-of-roundness results (R a values) have been processed for each measured circ le on the bore surface and presented in figure 6 with reference to the nom inal diameter which is numerically computed and fou nd equal 111 mm. The roundness is represented at each transverse section by the domain between the two vi rtual enveloping circles tangent to the distorted shape p rocessed using LSQ fitting technique built in the m achine as indicated in figure 6. Roundness at Circle I Ra = 9 0.84 µm D = 1 1 1.177 9 mm Roundness at Circle II Ra = 3 1.94 µm D =111.052 6 mm Roundness at Circle III Ra = 2 3.80 µm D =111.020 4 mm Roundness at Circle IV Ra = 1 8.44 µm D =111.051 6 mm Fig. 6. Roundness sample measurement records of eng ine cylinder bore (--- maximum amplitudes) 107 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER Analysis of the roundness patterns of the cylinder bore, illustrated in Fig 6, indicates the following points:  The CMM machine software establishes a reference ge ometric feature of ideal regular form, deduced numerically from one or more realistic irregular sc anned shapes. The established reference datum can b e used in the assessment process of the run out value s of the geometric features of the object under investigation.  Circle I nearby the TDC, as expected due to lubrica nt starvation, depicted the highest average distort ed dimension of D I=111.1779 mm and the largest average out-of-roundness (R a =90.84 µm).Whereas, the smallest distorted dimension was depicted at sectio n III (D III =111.0204 mm) in the vicinity of the mid-stroke point of the piston crown ring with average OOR val ue R a =23.80 µm., while the smallest out of roundness value was found nearby the BDC at circle IV (R a =18.44 µm).  Roundness of circle I has two maximum amplitudes (a rrow tips in Fig. 6) at points corresponding to loc ation of resultant surface reaction √F s 2 +F a 2 of piston side force F s and crankshaft axial force F a . The side force amplitude and direction vary according to the natur e of piston traveling displacement especially at compression and power strokes.  Amplitudes of circle III have the smallest wear var iation rate with relatively small out of roundness which may be due to good lubrication conditions and light side forces at that location. Whilst, roundness ne arby the BDC (circle IV) has different directions of pea k amplitude going with the indicated direction of c ylinder distorted shape. This may be attributed to the stud clamping force situation (magnitude and direction) when the piston passes by this location. At both BDC and TDC marginal inversion locations, the loads on the piston generates a stringent translated piston dyna mics. 4.2 Concentricity Measurements Experimentally measured values of the relative roun dness on the bore at different transverse locations (concentricity) have been found 39.10, 44.40, and 6 1.20 µm between circles I and II, circles I and III, and circles I and IV, respectively as shown in Figs 2, 6. This would reflect the distortion resulted from the extremely severe wear mechanisms to which the engin e cylinder bore was being experienced during servic e. 4.3 Out-Of-Straightness Measurement Results Figure 7 shows sample record of four averaged longitudinal profiles at equispaced locations 1, 2, 3, and 4 along the cylinder inner wall as indicated in Fig. 2 above. The maximum out of straightness value (S a ) processed from the measurements along each longitud inal profile represents the deviation domain around the relevant reference line obtained by applying LSQ fi tting technique. Straightness profile sample records shown in Fig. 7 disclose the following points:  Non uniform wear rates are exposed along all averag ed longitudinal profiles. It is clear that every po int on the cylinder bore is subjected to different concurr ent dynamic and environmental conditions of pressur e, friction, lubrication scheme, sliding velocity, con tact temperature, and contact force (orientation an d magnitude). Thus, frequent evaluation of bore surfa ce geometrical status is needed whenever possible t o help monitoring the functional degradation and diag nosing the surface failure symptoms in anticipation . So that, reasonable decisions can be taken regarding s urface treatment implementation and/or construction al design improvement inspiration. 108 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER  Maximum wear rates have been found to consistently lie within the TDC of the first pressure ring conta ct area for all averaged profile measurements of 71.58 , 54.38, 13.68 and 34.76 µm. This may be explained by the bad tribological conditions at the TDC location as aforementioned. The largest value of straightne ss departure (71.58 µm) which lies on profile 1 was formed during power strokes as a direct response to large side force reaction at high combustion temperatures . These findings are in agreement of a study carrie d out by Schneider et al [3].  Wear valleys of bore straightness has large values for profile at points 1 and 2 of power and compress ion stroke ends (nearby BDC) due to side force reaction of concentrated piston inertia, while profile of p oints 4 and 3 have shown the smallest amplitude valleys, re spectively.  An extended valley of straightness has the first pr ofile of power stroke till 70 mm long; it may be pr oduced of piston-skirt stringent side pressure and combust ion gas high temperature beside piston rotation aro und its pins under the effect of the friction force moment. Strong piston skirt dynamics accelerates the wear of the crankshaft axial movement control washers. Fig. 7. Longitudinal sample profiles of cylinder st raightness ( Sa is the averaged straightness of the profile). 5.0 Conclusions • Precision geometrical and dimensional micro scale m easurements of straightness, roundness, bore diameter, and concentricity of the internal surface of a worn out engine cylinder have been executed u sing CMM machine. Compared to its original design GT&D tolerance limits, these measurements proved to represent successfully a reliable diagnostic tool f or the wear development and aggression monitoring. Scenarios of the probable adverse operating conditi ons during service may also be drawn. The dimensional measurements of the bore diameter at di fferent transverse locations along the traveling st roke have assured previous findings using other differen t complicated measuring techniques. In turn this ma y provide feedbacks to both the engine designer for m odifications and the maintenance engineer for his forthcoming preventive and corrective maintenance p lans. Profile at line 1 S a =70.58 µm Profile at line 2 S a =54.38 µm Profile at line 3 S a =13.68 µm Profile at line 4 S a =34.76 µm 109 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER • The wear at the TDC and BDC transverse sections hav e been found much larger than the wear occurred at the middle of the stroke and in the vicinity of the BDC. This phenomenon is attributed to the continu ous existence of lubricating oil film dynamically prese rved at that location. • CMM machine precision measurements may also provide an insight in the engine dynamics that may contribute to the excessive wear occurrence in the engine cylinder. The geometrical deviation due to inhomogeneous wear has caused ovality in the bore w here (S a1 >S a2 >S a4 >S a3 ), as depicted in Fig. 8. This may inspire the engine designer to introduce an inn ovative modification to the engine by developing a controllable cylinder rotating device about its axi s probably without having to dismantle the engine p arts, so that the wear can be homogenized. Thus, the powe r loss due to friction and wear in the cylinder may be minimized and the engine operating life span may be rather prolonged. In addition, the maintenance expenses may be also reduced. 1 3 4 2 I II III IV Fig. 8. Bore geometrical deviation Fig. 8. Bore geometrical deviation References [1] Nathan W. Bolander, Brian D. Steenwyk, Ashwin K umar, and Farshid Sadeghi, "Film Thickness and Fric tion Measurement of Piston Ring Cylinder Liner Contact w ith Corresponding Modeling Including Mixed Lubrication”, Conference of the ASME Internal Combu stion Engine Division, Long Beach, California, USA, (2004). [2] Neale, M.J., "Tribology Handbook", Newnes Butterwor ths, London, (1975). [3] Schneider, E.W., Blossfeld, D. H., Lechman, D. C., Hill, R. F., Reising, R.F., and Brevick, J.E.," Eff ect of Cylinder Bore Out-Of-Roundness on Piston Ring Rotat ion and Engine Oil Consumption", SAE, 930796, (1993). [4] Ohlsson, R., "A Topographic Study of Functional Surfaces", Ph.D. Thesis, Chalmers Univ. of Tech., Sweden, (1996) [5] Andersson, P., and J. Tamminen, "Piston Ring Tribol ogy: A literature Survey", VTT Research Notes 2178, Helsinki Univ., (2002). [6] Vatavuk, J., and Demarchi, V., “Improvement of Cyli nder Liner Materials Wear Resistance”, SAE, 931671, (1993). [7] McNally, C. P., "Development of Numerical Model of Piston Secondary Motion for Internal Combustion Engines", M.Sc. Thesis, Massachusetts Institute of Tech., (2000). [8] Jiang, B.C., and S.-D. Chiu, "Form Tolerance-based Measurement points determination with CMM", Journal of Intelligent Manufacturing Vol. 13, No. 2, pp. 10 1-108, (2004). [9] Bugra Kilic, Juan A. Aguirre-Cruz, Shivakumar Raman , “Inspection of the Cylindrical Surface Feature af ter turning using Coordinate Metrology”, International Journal of Machine Tools & Manufacture, pp. 1893– 1903, 47 (2007). 110 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 U SE OF PRECISION MEASUREMENTS OF EVOLVED GEOMETRICA L AND DIMENSIONAL DEVIATIONS AS A DIAGNOSTIC TOOL FOR AIR-CO O LED DIES EL ENGINE CYLINDER [10] Rosen, B.-G, Ohlsson, R., and Thomas, T.R., "W ear of Cylinder Bore Micro-Topography: Some Methodological Considerations", Brighton, UK, (1994 ). [11] A.Ramalho, “A Geometrical Model to Predict the Wear Evolution of Coated Surfaces”, Elsevier, Wear J., (2007). [12] George A.L., and Nikolaos P.K., “Friction Model of a Marine Dies el Engine Piston Assembly”, Vol. 40, Issues 10-12, , PP. 1441-1453, Tribology I nternational, October-December 2007. [13] Keith Birch, “Measurement Good Practice No. 36: Est imating Uncertainties in Testing, An Intermediate Guide to Estimating and Reporting Uncertainty of Me asurement in Testing”, British Measurement and Testing Association, NPL, UK, March 2003. [14] International Standard: Geometrical Product Specifi cations (GPS) - Acceptance and Reverification Tests for Coordinate Measuring Machines (CMM) -- Part 2: CMMs used for Measuring Size, ISO 10360-2, 15-12- 2001. 111 112 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY AUGMENTED REALITY DESIGN METHODOLO G Y FOR COMPUTER NUMERICAL CONTROL MACHINERY Wasim Ahmed Khan, PhD, CEng, FIMechE Centre for Computer Studies, Institute of Business Administration, City Campus, Garden Road, Karachi – 7440 0 , Pakistan Abstract Virtual manufacturing is a synthetic environment exercised t o enhance all levels of decision and control in a discrete or continuous manufacturing organization. Machine tools are the key components of the discrete manufacturing sys tem. Computer numerical control machine tools provides basic human computer interface at the discrete manufacturing processes. This paper describes th e augmented reality methodology for building the virtual model of computer numerical control machine tools. Such augmented reality models are used as the micro ele ment of the global virtual manufacturing organization. The virtual manufacturing organizat ion apportions planning for a non existent manufacturing system, operat ion monitoring of the existent system, fault diagnostic and rapid maintenance, mai ntenance planning, practicing quality assurance, optimizing the local and global human computer interfaces, optimizing the flow of information, acquiring rapid response from the system, training at the manufacturing systems, distance learnin g, practicing e- commerce and possibility of implementing different production phil osophies. Keywords: Discrete Manufacturing, Continuous Manuf acturing, Computer Numerical Control, Augmented Reality, Virtual Reality 1.0 Introduction According to MSN Encarta the term ‘virtual reality’ is commonly used to express Simulated Reality, Computer Simulation, Simulation, Cyberspace, Comput er Modeling or Computer Graphics. In today’s scientific scenario, virtual reality is classified on a continuum from Real environment to its variati ons to virtual environment. These variations of virtual reality ar e from real environment to augmented reality to aug mented virtuality to the virtual environment. All the inte rmediate representations are known as mixed reality [1]. Azuma et. al. describes the Augmented Reality (AR) in their survey paper [1] as having the following properties: (i) AR combines real and virtual objects in a real environment; 113 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY (ii) AR runs interactively and in real time, and; (iii) AR registers (aligns) real and virtual object s with each other. A discrete manufacturing operation involves tangibl e activities such as machinery and its operation, u se of tools and measurement gadgets, use of pick and plac e technology and use of storage and transportation equipment etc. On the other hand, the intangible pa rt includes services such as process planning, sche duling, inventory, management information system and busine ss accounting etc [2][3]. Establishment of discrete manufacturing facility for specified range of discr ete products includes the factory and offices layou t, machinery layout, operation of design office, opera tion of new and old machinery, production planning and control, scheduling, assembly, quality assurance, i nventory, transportation, budgeting and accounting and financial activities. Monitoring of all these and o ther functions is required once the facility has be en setup and is functional. In contrast to discrete manufacturing systems, a co ntinuous manufacturing system produces liquids, gas es solids, grains or powders. These continuous systems are less flexible as compared to discrete manufact uring systems. However, the application of virtual realit y principles to continuous manufacturing systems co nsiders both tangible and intangible parameters associated with these systems. Comprehensive control of the manufacturing processe s and the manufacturing systems is of prime importa nce for the sake of materializing the goals set for man ufacturing since the industrial revolution. One of the basic control instruments that prevailed from centuries t o today for manifesting control is the man machine interface for the processes and the systems. With increased c omputer based automation it is now evolving towards human computer interaction supplemented by virtual representation of the real processes and systems. The design of virtual reality (VR) for discrete and continuous manufacturing covering manufacturing processes and systems is addressed. The discrete an d continuous manufacturing systems are considered t o be composed of physical structure for the processes an d the system, control characteristics of the proces ses and the system and dynamics involved in the operation o f the processes and the system. Virtual reality for discrete and continuous manufacturing systems involves defin ing steps required to build an Augmented Reality (A R) for a manufacturing process such as metal cutting p rocess from component to mechanism to machine level and subsequently integrating the Augmented Reality (AR) of the process into the virtual reality (VR) of th e manufacturing system such as job shop, project shop , cellular system, flow line or continuous manufact uring system [4][5][6]. This paper presents the augmented reality design methodology for Computer Numerical Control machinery. 2.0 Manufacturing System and Processes The basic configuration of discrete and continuous manufacturing systems by Chryssolouris [7] is consi dered. This comprises: 1. Job shop 2. Project shop 114 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY 3. Cellular system 4. Flow line 5. Flexible manufacturing system, and; 6. Continuous manufacturing These ideal layouts for discrete and continuous man ufacturing along with assembly methods constitute t he core of the virtual reality template required for d efining any virtual reality application in manufact uring. The processes required to build a manufacturing system are mainly adopted from Kalpakjian [8] description of current manufacturing processes comprising the foll owing general areas: 1. Metal forming processes 2. Bulk deformation processes 3. Sheet metal working processes 4. Metal cutting processes 5. Metal joining processes 6. Thermal properties modification processes 7. Surface properties modification processes 8. Fabrication of micro mechanical and micro electroni cs devices 9. Non conventional processes 10. Continuous Manufacturing Processes In a true virtual reality of discrete or continuous manufacturing system, the manufacturing processes should be defined with high level of virtuality encompassing all their functionality and should be capable of be ing configured on a chosen manufacturing system layout along with comprehensive energy, information and material flow capability [2][4][6]. The simulated processes should have the capability to mimic real manufacturing processes and are confi gured on a simulated factory layout. It is also possible to configure each of the tangible and intangible op erations in variable quantity and size depicting a real factory . The manufacturing system and the manufacturing pr ocesses chosen in the virtual domain are reconfigurable thu s allowing infinite possibilities for the developme nt of virtual manufacturing organization. The processes m ay be operated using standards such as variants of EIA 274D and JIS SLIM (Standard Language for Industrial Manipulator), techniques used in programmable logi c controllers (PLC) such as defined by IEC 61131-3, E mbedded system and Supervisory Control and Data acquisition (SCADA) and other pertinent standards [ 9]. 3.0 Automation of Production Equipment Discrete manufacturing processes are developed as m echanical artifacts to perform production. These machines produce variety of components, structures, assemblies, mechanisms and machines. Each manufacturing process implementation into mechanica l artifacts has three distinct features [6]: 1. The input to the equipment e.g. raw material type, form and feeding mechanism; final dimension of the product, energy source and other auxiliaries. 115 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY 2. The process implementation allowing transformation of raw material into required size, shape and surfa ce finish using tools (e.g. tools as used in metal cut ting, high energy beams or various types of jets); utilizing tool holding devices and work holding devices; measuring devices and manufacturing instructions. In process supplies such as lubricating oil and coolants may a lso be used. 3. The out put from the equipment comprising a compone nt (the building block of structure, mechanisms or machines); and, scrap. There may be several other features present at the production machinery to make the task of manufactur ing simpler, easy to control and resulting in high prod uction rate. Like any other discrete product, the manufacturing equipment commonly utilizes standard mechanical components, machine elements, control elements, ele ctrical and electronics components and software components. Special assemblies and other accessories utilized a t the construction of manufacturing equipment may a lso include: 1. Tool (Referring to mechanical tool, Light Beams, Wa ter Jets etc.) 2. Tool Holding Devices 3. Work holding Devices 4. Lubricating oil pump assembly 5. Coolant circulation pump assembly 6. Material Handling Equipment, and, 7. Scrap Handling Equipment Computer Numerical Control, whether implemented in open loop or closed loop configuration, is the most common human computer interface at discrete manufac turing equipment. It is widely used to control feat ures, assemblies and accessories at the discrete manufact uring machinery. Computer Numerical Control is commonly implemented at the following discrete manu facturing equipment [9][10][11][12]: 1. Mills and Machining Centers 2. Lathe and Turning Centers 3. Drilling Machines 4. Boring and Profilers 5. Electro Discharge machines 6. Punch Press and Shears 7. Flame Cutting Machines 8. Water Jet and Laser Profilers 9. Cylindrical Grinders 10. Welding Machines 11. Benders, Winding and Spinning Machines The above listed processes cover the complete horiz on of discrete manufacturing process classification as identified by Kalpakjian [8]. 116 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY 4.0 EIA 274D Standard Electronics Industries Association (EIA) RS 274 D i nterchangeable variable block data format for posit ioning, contouring and Contouring/Positioning Numerically C ontrolled Machines [1979] is the most basic and mos t common standard used in the manufacturing industry for controlling CNC Machinery. A subset of this standard, as shown in figure – 1 and figure – 2, is used to demonstrate augmented reality design metho dology for computer numerical control machinery [9]. 5.0 Axes Classification for Metal Cutting Machinery Axes classification for numerically controlled mach ines is defined by another commonly adopted EIA standard: EIA RS 267 Axis and motion nomenclature f or numerically controlled machines. In this paper t his standard is utilized for the axes classification of CNC machinery [9]. 6.0 Software Tools for Virtual Manufacturing The development of virtual model for describing the functionality of a product, to demonstrate operati on of manufacturing, pick and place or assembly equipment or to demonstrate the operation of a complete virt ual organization requires simulation of all tangible pr oduction functions while mathematical modeling of a ll related intangible functions. All these suites of p rograms should be modeled, developed and run using state of the art software modeling tools, software developme nt technique and software execution architecture [6 ]. Object-oriented software development offers a new a nd powerful model for writing computer software. Th is approach speeds the development of new programs and , if properly used, improves the maintenance, reusability, and modifiability of software. There are a large number of object-oriented program ming languages in use today. But the leading commer cial object oriented languages are far fewer in number. These are: 1. C# 2. Smalltalk 3. Java These object oriented languages can be used in virt ual reality design in conjunction with following gr aphic support tools [13][14][15][16]: 1. VRML (Virtual Reality Modeling Language) 2. OPEN GL (Low Level Graphics Library) 3. Other Graphic Design and Rendering Tools 7.0 Software Development Tools 117 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY The Augmented Reality for metal cutting machinery i s designed and developed using the following tools: 7.1 Unified Modeling Language The Unified Modeling Language (UML) is the standard language for visualizing, specifying, constructing , and documenting the artifacts of a software intensive s ystem. Complex software design that would be diffic ult to describe textually can readily be conveyed through design diagrams. Each diagram focuses on one aspect of application. One may focus on structure, another on behavior, and yet another on the physical partitio ning of the application. The model can be used to clearly c ommunicate with the members of programming team. Similarly the model can be used to automatically ge nerate source code. Modeling provides three key ben efits: visualization, complexity management, and clear com munication [17]. Fig. 1. A UML Use Case Diagram for CNC Operation 7.2 Visual Studio .Net 2003 Visual Studio .Net Version 1 is an IDE (Integrated Development Environment) developed for Windows Operating System. It is a complete suite of tools f or building both high-performing desktop applicatio ns and team-based Enterprise Web applications. It has a ri ch set of languages available like Visual C++, C#, VB .NET, J# etc. The system presented in this paper u tilizes C# .NET for developing the augmented models [18]. 118 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY Fig. 2. A UML Sequence Diagram for CNC Operatio 7.3 True Vision 3D SDK 6.2 True Vision TV3DSDK is a 3D rendering engine writte n using Direct X and provides powerful functions fo r accomplishing 3D abstraction. It helps to develop a powerful 3D application/much faster than the basic graphics libraries like Direct X. This 3D Rendering Engine is used to access its built-in functions in C# for loading and drawing complex graphics models [16]. 7.4 Direct X 9.0 Direct X is an advanced suite of multimedia applica tion programming interfaces (APIs) built into Micro soft Windows operating systems. It provides a standard d evelopment platform for Windows-based PCs to access specialized hardware features with easy programming code. The Direct X library is used through the Tru e Vision SDK to access the graphics hardware using th e .NET environment [19]. 119 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY 7.5 3D Studio Max 6.0 3D Studio Max is a 3-dimensional vector graphics an d animation program. It is a powerful 3D modeling a nd animation software used to design large scale anima tions for games, mechanical models and other 3D simulations. 3D Studio Max is used for designing th e machine models [20]. 8.0 Discussion A methodology for augmented reality representation of discrete manufacturing machinery is developed. T his methodology covers processes as identified in secti on – 3. A UML use case diagram is provided as figur e – 1 to demonstrate the working of augmented reality for Computer Numerical Control machines. Figure – 2 presents the sequence diagram for the augmented rea lity representation. Figure – 3 shows an augmented reality representation of a sawing machine. This representa tion has the capability to mimic the operation of t he machine tool and provides the shape of the final pr oduct. The augmented reality components of the machine too ls requires modeling of mechanical components (Powe r Screw), Electrical Component (Motors), Electronics Components (Limit Switches, Shaft Encoder) and Computer Science Components (Interpreter for EIA 27 4 D Standard, Graphics model of the machine tool, Graphics model of the work piece and rendering meth odology). The electronic signals generated by the interpreter are simultaneously sent to the visual d isplay unit of the machine tool controller and the controller machine tool interface is allowing higher level of human computer interaction. Such a virtual model ha s the capability to become a micro element of a virtual o rganization defining both tangible and intangible f unctions of discrete manufacturing system. Using same principles, a virtual organization for d iscrete or continuous manufacturing can be develope d to work as the proxy to the real organization. Such a scheme has the possibility of planning for a non existent manufacturing system, operation monitoring of the e xistent system, fault diagnostic and rapid maintena nce, maintenance planning, practicing quality assurance, optimizing the local and global human computer interfaces, optimizing the flow of information, acq uiring rapid response from the system, training at the manufacturing systems, distance learning, practicin g e-commerce and possibility of implementing differ ent production philosophies. The oral presentation of this paper includes AR sim ulation of several CNC machine tools such as Turnin g, Milling, Drilling and Sawing machine. The AR simula tion of Job Shop and Cellular Manufacturing amalgamating CNC machine tools with material handli ng equipment while producing specified objects shal l also be presented. 9.0 Conclusion A prototype for the dynamics of micro element of vi rtual discrete manufacturing organization in a job shop is presented. It proves that the computer and communic ation technologies have reached a level whereby the 120 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AUGMENTED REALITY DESIGN METHODO LO GY FO R COMPUTER N UMERICAL CONTROL MACHINERY realization of a virtual manufacturing organization is possible. This shall lead to the availability o f micro and macro decision variables to the mangers allowing op timum decision making for higher profitability. Fig. 3. A CNC Sawing machine augmented reality repr esentation Acknowledgements The author wish to acknowledge development of augme nted reality for computer numerical control machine s by Mr. Amir Hussain at the Institute of Business Ad ministration, Karachi, Pakistan; and, the Higher Ed ucation Commission of Pakistan for providing the travel gra nt. References [1] Azuma, R.; Bailout, Y.; Behringer, R.; Feiner, S.; Julier,S.; MacIntyre, B.; “Recent advances in Augm ented Reality”; IEEE computer Graphics and Applications; November/December 2001 [2] K. Hitomi;”Manufacturing System Engineering”; Taylo r and Francis; 1975. [3] Yeoman R. W. et.al.; “CIM in the small firm – In de sign rule for CIM systems”; North Holland; 1985. [4] Alting L.; “Manufacturing engineering Processes”; 2 nd Edition; Marcel Dekker; 1994. [5] Bedworth D.D. et.al.; “ Computer Integrated Design and Manufacturing”; McGraw Hill; 1991. [6] Khan, Wasim Ahmed; Raouf A.; “Standards for Enginee ring Design and Manufacturing”; Taylor and Francis, December 2005 [7] Chryssolouris, G., “ Manufacturing systems – Theory and Practice”; Springer Verlag; 1992. [8] Kalpakjian S., and Schmid S.R.; “Manufacturing Engi neering and Technology”; 4th ed.; Prentice Hall; 20 00. [9] Electronics Industries Association Recommended Stan dard EIA RS 274 D, 1979. [10] Leatham-Jones, B. ; "Introduction to computer numer ical control”; Pitman; 1986. [11] Kief, H.B.; “ Flexible automation – the internation al CNC reference book”; Becker Publishing; 1986 [12] Smid, P; “CNC Programming Handbook’; Second Edition , Industrial Press; 2003. [13] Hill Jr F. S.; “Computer Graphics using Open GL”; P rentice Hall; 2000. [14] Jamsa K. et.al. ;” VRML Programmers Library”; Jamsa Press; 1997. [15] Liu C.;” Smalltalk, Objects and Design”; iuniverse .com; 2000. [16] Chen, Jim; “A survey of 3D Graphics software tools” , IEEE computer society, CS ready notes; 2006. [17] Quatrni, T.; ”Visual modeling with Rational Rose 20 00 and UML”; Addison Wesley; 1999. [18] Johnson B. et. al. ; “Inside Microsoft Visual Stud io . NET 2003”; Microsoft Press; 2003. [19] Luna F.; “Introduction to 3D Game Programming wit D irect X 9.0 c: A Shader Approach”; Wordware Publish ing Inc.; 2006. [20] Murdock K. L.; “3D Studio Max: R3 Bible”; Wiley; 20 00. 121 122 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O PTIMIZATION OF RESIN FLO W IN A FLEXIBLE MOULD INFUS ION PRO CESS FO R LARGE PRESSU R E VESSELS O PTIMIZATION OF RESIN FLOW IN A FLEXIBLE MOULD INFUSION PROCESS FOR LARGE PRESSURE VESSELS Gascons, M. 1 , Blanco, N. 1 Matthys, K. 2 1. Analysis and Advanced Materials for Structural Desi gn, Universitat de Girona 2. School of Engineering and Design, Brunel University West London Abstract Permeability and fibre volume fraction of reinforcement fabri cs are two parameters that are greatly influenced by the mould compression effect during the resin infusion process for the manufacturing of textile composite parts. In c losed mould processes, such as resin transfer moulding (RTM), compression is dete rmined by mould cavity, while in open mould processes such as vacuum assisted resi n transfer moulding (VARTM) with flexible tooling on at least one side, the compr ession of the fabrics is mainly determined by the vacuum outflow pressure conditions. In our specific application of the production of a large press ure vessel, the mould set- up consists of a fixed external mould and a flexible rubber t ooling bag as the internal mould. The bag can be pressurized externally under controlled conditions in order to provide a compression force to the reinforcement fabrics bef ore infusion. The effect of the enforced pressure conditions on the flexible tooling wi ll be reflected on the fabric, as the generated compression effect will modify the permeab ility and fibre volume fraction. The object of this study is to optimize the pressure hist ory and control of the inflatable bag during an infusion process for large textile composite pres sure vessels in order to better control resin flow and minimize the effect of per meability and fibre volume fraction changes on process time and final part quality. Numeri cal simulation techniques of resin flow will assist in finding better process conditions such as flexible tooling pressure, infusion time, and final thickness of the part. Results of numerical simulations conducted show the importance of the thickness reduction phenomena in bag moulds and its influence on the real part. The inflatable bag used, considered as a rigid mould, gives improved results and manifests itself as b est choice for the 123 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 application. Simulations allowed reduced the need of expensive experimental prototype testing by means of virtual infusion process optimization. Keywords: Permeability, Resin Flow, Flexible mould 1.0 Introduction The use of composite vessels has extended widely in many industrial applications. Most of these vessel s are made of fibreglass and a polymer resin such as poly ester. Vessels have been produced since many years ago with a large variety of manufacturing techniques, t argeted to suit the focus on a specific application field. In particular, for low-tech applications, a combinatio n of spray-up and hand lay-up has been, for years, the production process that is most adapted to the nece ssities of low costs, size flexibility and final pa rt quality. This open mould process consists of a spray gun pro jecting resin over the mould, in conjunction with a roving chopping head. However, using this method, the qual ity of the part is highly influenced by thickness a nd fibre volume fraction variations that are subsequently re mediated by a manual compaction process. With the incoming of tighter cost limits, to uphold a competitive production strategy, and with ever m ore strict environmental laws, open mould processes have come strongly under pressure because of the associated h igh volatile compound (VOC) emission level. About 14.1 and 25% of styrene emissions are produced. The abov e technological and legislative challenge had formed the foundation for the creation of a new production process for large vessel manufacturing. The new process can be split in two main stages. Th e first stage includes the production of a preform from roving fibreglass, which is chopped and mixed with binder and projected over the preform mould. After a compaction process, one-layer of mat fabric with a shape and thickness close to the final part is obta ined. The second stage then introduces a moulding techniq ue in the process. Major advantages of closed mould processes are the significant reduction of volatile s and the fact that the part weight is reduced as b ecause less resin is used. Also mechanical properties are impro ved due to the reduction of voids and the better th ickness control of the part. However, disadvantages are the high tooling cost and sometimes the reduction of f inal part thickness due to the compaction effect in the close d mould The following paper uses numerical simulation to an alyze the production of a generic large vessel geom etry in a specific mould setup to investigate whether the a bove described second stage moulding process can be tter be implemented by means of a closed mould technique su ch as resin transfer moulding or via a flexible mou ld technique as is vacuum bag moulding (VBM). 2.0 Description of the Part and the Mould Setup The design of a new range of vessels, lea to the de sign of a full new production process that upholds cost limits, environmental requirements, and improves th e properties of the actual vessels suppressing the joint between the two middle-parts of the current geometr y configuration. 124 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 To achieve that, a mould setup consisting of an ext ernal two-part rigid mould and an internal bag has been developed. The production process includes the self -production of a preform made from chopped roving f ibre and with the shape of a half vessel. On the resin i nfusion stage, the bottom half-vessel preform is pl aced over the half-vessel bottom mould, and the upper perform is placed over the inflatable bag. Both are joined and the upper mould closes the system. The whole set will b e sealed to start the infusion. An illustrative ima ge of the set-up is shown in Figure 1. The untypical conception of the mould, with the int ernal mould being the flexible one, gives the oppor tunity to include pressure to this bag if necessary. Hence, d epending on the pressure of the bag, it can be cons idered as: Rigid Mould : The inside pressure of the bag is enough to consi der the system as a rigid mould during the infusion process. The compression effect is then ne gligible and the resin infusion can be described as a RTM. Flexible Mould : A silicone bag without structural resistance is u sed, and the setup becomes a typical VBM process. Compaction behaviour of the fabric must be reflected on simulations. Fig. 1 Representative configuration of the mould se tup with the two external parts of the mould, the i nflatable bag and the two preforms 3.0 Theoretical Background 3.1 Resin Transfer Moulding Modelling A typical setup for Vacuum Assisted Resin Transfer Moulding (VARTM) consists on a rigid mould in one side and a flexible mould in the other side of the cavity. Resin is conducted through the mould with t he help of a vacuum from vents. In the first setup considered, a vacuum is used with rigid moulds on each side of the mould cavity: Hence, RTM theoretical background mus t be considered. RTM simulation had been widely studied by different authors [1]-[2]-[3]. The porous media flow approac h to analyze the flow of resin is commonly used, with a combination of Darcy’s law and a continuity equatio n. The flow front advance through an incompressible mould cavity for an anisotropic, homogeneous medium can b e expressed as: 125 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008             ∂ ∂ ∂ ∂ ∂ ∂           −=           z P y P x P KKK KKK KKK w v u zzzyzx yzyyyx xzxyxx · 1 µ (1) Where u, v and w are the volume averaged flow velocity, µ is the viscosity of the fluid and the matrix K describes the permeability of the reinforcement in each direction. Pressure gradients in each directio n are represented by the last column. The fluid mass conservation equation is introduced and integrated over a control volume. Using the div ergence theorem, the control volume integral can be transfo rmed into a control surface integral, giving the fo llowing expression:   0· =           ∫∫ dS w v u nnn S zyx (2) Where n x , ny and nz are the normal components of the surface vector of the control volume. Replacing equation (1) into equation (2), the governing equation descr ibing flow through a general anisotropic media is o btained.   0··1 =             ∂ ∂ ∂ ∂ ∂ ∂           ∫∫ dS z P y P x P KKK KKK KKK nnn zzzyzx yzyyyx xzxyxx S zyxµ (3) 3.2 Vacuum Bag Moulding Modelling Vacuum Bag Moulding is understood as an infusion pr ocess with the mould cavity delimitated by, at leas t for one flexible mould. The lack of resistance of this mould implies that the dry preform placed on the in ternal cavity to a compaction pressure that modifies param eters such as thickness or fibre volume. Also, this changes influence in the infusion results, decreasing perme ability and flow front velocity. For flow through compressible media, a compacting m ould cavity must be considered, hence, following th e analysis of Gutowski et al [4]-[5]-[6]-[10], the 1D in-plane flow continuity equation that must be fol lowed is: x h t h ∂ ∂ = ∂ ∂ )·(ν (4) 126 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 Where h is the local material thickness and t is time. The combination of (1) and (4), including the fact that thickness and permeability are functions of pressur e, which is a function of the position in the mould , x (h [P(x)]) and K [P(x)], leads to the following expres sion         ∂ ∂ +   ∂ ∂   ∂ ∂ + ∂ ∂ = ∂ ∂ 2 22 ······ 1 x PKh x P P Kh P hK t h µ (5) Finally, a consideration of the flow front movement , changing pressure along the mould, must be solved . Making the domain of the equation includes a parame ter α=x/L where L is the instantaneous flow front position. This defines a domain of α being within the limits α=0 at inlet and α=1 at flow front position. Equation (5) can be written as (6) being able to be solved using an iterative finite element method to compute the pressure field, from which flow front progressi on can be determined using Darcy’s law. [ ] 2 1 2 2 ·· 1 · 1       +    −−=    = α α α α d dP dP dK KdP dh hhd Pd (6) Specific models of compaction and permeability must be included in the governing equation of flow to f ully simulate the infusion behaviour. With the mould cavity fully vacuumed, compaction pr essure over the reinforcement is defined by vacuum pressure. This pressure state changes with inlet op ening, when a wet zone is created and the compactio n behaviour can be defined as a difference between th e pressure outside of the bag (atmospheric) and the vacuum driving pressure. From previous studies [7]- [8], part thickness in flexible bag moulds is again related to the position of the flow front, decreasing from inlet to vents position. Correia et al. [9] unified most used expressions for compaction, such as the one used by Gutowski et al. [5] as, 4 0 0 11 1     −     − = a scomp VfVf Vf Vf AP (7) where As, Vf, Vf 0 and Vf a are the reinforcement spring constant, the fibre v olume fraction, the fibre volume fraction at zero compaction pressure and the theore tical maximum fibre volume fraction. Other research ers [6] used (8) as expression for compaction BPVfVf ·0= (8) Where Vf 0 is the fiber volume fraction at 1 Pa and B is the stiffening index determined by the fiber st ructure. 127 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 To describe the permeability modelling, different a uthors [8]-[9] use the Kozeny-Carman equation 2 3)1( · Vf VfkK −= (9) where k is determined experimentally for each fabric. 4.0 Simulation Conducted Simulations of the large vessels production process had been conducted using LIMS[12]-[13]-[14], a fin ite element/control volume simulation code with capacit y of analyzing 2D and 3D flows developed at the University of Delaware. LIMS also includes LBASIC, a built-in script language that enables users to mo dify different conditions during the simulation. The RTM-like process was simulated with the user in terface of LIMS, LIMS-UI, which allows direct solut ion of the case analyzed. In order to simulate the flow through a compressible media of the flexible mould consideration, it is required to update thickness, fibre volume and permeability locally as a function of compaction pressure. This can be done with LBASIC s cript, which is able to update the required changes and check iteratively for convergence by resolving the pressure field under the new conditions. The flow f ront is then advanced and the procedure is repeated. To simulate flexible mould infusions, some assumpti ons must be considered, such as a laminar flow, no pressure gradient in through-the thickness directio n and incompressible resin with constant viscosity while filling. These assumptions can be only made when, a s in our case, flow through the thickness is consid ered uniform. 4.1 Material Description The self-made preform can be defined as a fibreglas s mat. The production process of the preform includ es a compaction stage, in which the preform achieves the initial properties. The initial thickness of the p reform mat is 5 mm, with a fibre volume of 35% and a estimated permeability of 5x10 -10 m 2 . The Resin considered is a low viscosity resin, for infusion process, with a v iscosity of 0,25 Pa s. Vacuum pressure used is fixe d to 97,5 kPa. 4.2 Compaction Behaviour A specific bibliographic review had been done [12]- [13]-[14] searching for compaction behaviour of the considered materials. Special attention to the work of Kerang Han et al., [14] where bidirectional fib re mat (COFAB A118) and random fibre mat (COFAB M8610) are analyzed, equations (10) and (11) are given, being h the thickness of the fabric in inches and p the pressure acting on the fibre in psi. 128 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 )1·(00703.0 ))]log(38.01(3.2[1ker += − pang eh (10) )1·(0093.0 ))]log(191.0303.0(3.2[2ker += − pang eh (11) Hammami [15]-[16] analyzed the behaviour of differe nt configurations of dry Unifilo U850, a continuous filament mat from Vetrotex. The following equations are extracted from his work, being h the thickness in mm and P the pressure in MPa 0089.0)·ln(0005,01 +−= phHammami (12) 0068.0)·ln(0004,02 +−= phHammami (13) The variation of thickness with pressure is represe nted in Figure 2 for expressions (10) to (13). As i t can be seen in the figure, compaction behaviour can affect the final thickness of the preform notoriously and consequently mechanical properties can be affected. 0 1 0 10 20 30 40 50 60 70 80 90 100 Pressure [kPa] Co m pa cti on [ % ] Kerang 1 Kerang 2 Hammami 1 Hammami 2 Fig. 2 Compaction behavior vs. pressures in a typic al vacuum process 5.0 Simulation Results All the compaction models were analyzed and fitted to the particular vessel considered in this study. Simulations were conducted including this compactio n behaviour and compared with a typical RTM process . Figure 4 is a plot of a representative path of the vessel, where thickness reduction of 41 to 36% is o bserved, being the thicker parts near the inlets and the thi nner parts near the vents. 129 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 0 1 2 3 4 5 0 0.2 0.4 0.6 0.8 1 Distance [m] Th ic kn e ss [mm ] Hammami1 Hammami2 Kerang2 RTM Fig. 3 Thickness variation through a vessel path . 0 at inlet, 0.95 at vent The first compaction behaviour of the two described by Kerang et al. [13] at eq. (10) does not lead to convergence because the level of compaction achieve d is higher than the maximum for the composite. Consequently, this compaction behaviour had to be d iscarded from the analysis Results lead to the conclusion that in VBM processe s the compaction pressure clearly affects the final thickness of the part. Compaction behaviour results in a thickness decrease of about 36 to 41%. This i mplies a fibre volume increase of about 46 to 37%. Permeability is also affected due to compaction res ulting in a slower infusion process. Actually, flex ible mould processes are known to be typically slower than RTM ; and this is why ancillary material such as race t racking channels or high permeability layers are commonly u sed in order to reduce infusion time. In Figure 4, filling time of the vessel is plotted versus linear distanc e from inlet to vent. The compaction effect represe nts an increase in time of 12, 5 to 23, 4% 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 2 5 0 0 3 0 0 0 0 0.2 0.4 0.6 0.8 1 Distance [m] Ti m e [s] hammami1 hammami2 kerang2 RTM Fig. 4 Filling time through the geometry of the fil ter from inlet (0, top) to vent (0.95, bottom) 130 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 6.0 Conclusion The Simulations conducted lead to the conclusion th at, without auxiliary goods like flow channels or f low enhancement layers, VBM process can be considered a slow process. Despite being attractive due to a lo wer cost level in comparison to RTM, the planning of ma nufacturing vessels using VBM has been proven not suitable due to a dangerous reduction in thickness that can affect the mechanical properties of the ve ssel. Also, in terms of productivity, the VBM does not fit time requirements. Acknowledgement The authors wish to acknowledge the assistance of P oltank and the support of the Composites Centre of the University of Delaware References [1] Lin M., Thomas H., Hoon H., “A finite element simul ation of resin transfer moulding based on partial n odal saturation and implicit time integration” Composite s: Part A 29A(1998) 541-550 [2] García J.A.,Gascón, Ll.,Chinesta F., “A fixed mesh numerical method for modeling the flow in liquid co mposites moulding processes using a volume of fluid techniqu e” Comput. Methods Appl. Mech. Engrg. 192(2003) 877 -893 [3] Correia N.C., Robitaille F., Long A.C., Rudd C.D., “Use of Resin Transfer Moulding simulation to predi ct flow saturation, and compaction in the VARTM process” Tr ansactions of the ASME Vol. 126 March 2004 p210-215 [4] Gutowski,T. G.,and Dilon, G., 1997, “The elastic de formation of Fiber bundles,” Advanced Composites Manufacturing , T.G. Gutowski, ed., pp 138-139 [5] Gutowski,T. G.,Morigaki T., Cai Z., “The consolidat ion of laminate composite” Journal of Composite Materials,21,1987, p172-188 [6] Simacek P.,Advani S.G., “Modeling resin flow and fi ber tow saturation induced by distribution media co llapse in VARTM” Composites Science and Technology 67 (2007) 2757-2769 [7] Kessels J.F.A., Jonker A. S., Akkerman R., “Fully 2 ½ D flow modeling of resin infusion under flexible tooling using unstructured meshes and wet and dry compaction prop erties” Composites Part A 38 (2007) 51-60 [8] Robitaille F.,Gauvin R., “Compaction of textile rei nforcements for composites manufacturing II: Compac tion and relaxation of dry and H2O saturated woven reinforce ment” Polym Comp. 1998; 19(5) 543-57 [9] Correia N.C., Robitaille F., Long A.C., Rudd C.D.,S imacek P., Advani S.G., “Analysis of the vacuum inf usion moulding process: I. Analytical formulation” Compos ites Part A 36 (2005) 1645-1656 [10] Simacek P, Advani S.G., “A numerical model to predi ct fiber tow saturation during liquid composite mou lding” Composites Science and Technology 63 (2003) 1725-17 36 [11] Simacek P, Advani SG “Smulation three-dimensional f low in compression resin transfer moulding process. Rev Eur Elements Finis 2005; 14(6-7):777-802 [12] Simacek P, Advani SG”Desiderable features in mould filling simulations for liquid moulding processes” Polym Composites 2004;24:355-67 [13] Simacek P, Advani SG, Binetruy C “Liquid Injection Moulding Simulation (LIMS) a comprehensive tool to design, optimize and control the filling process in liquid composite moulding” JEC Composites 2004(8) 58-61 [14] Kerang H., Jiang S.,Zhang C.,Wang B., “Flow modelin g and simulation of SCRIMP for composites manufactu ring” Composites: Part A 31 (2000) 79-86 [15] Hammami A., “Effect of reinforcement structure on c ompaction behavior in the vacuum infusion process”, Polymer Composites, June 2001, Vol 22,No3 p337-348 [16] Hammami A., Gebart B.R., “Analysis of the vacuum in fusion moulding process” Polymer Composites, Feb 20 00, Vol 21 No.1 p28-40 131 132 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING EXPERIMENTAL ANALYSIS OF PROP ERTIES OF MATERIALS FOR RAPID PROTOTYPING Mladen Šerce 1 , Pero Raos 2 , Ana Pilipovi ć 1 1. Faculty of Mechanical Engineering and Naval Archit ecture, University of Zagreb, Croatia 2. Mechanical Engineering Faculty in Slavonski Brod, University of Osijek, Croatia Abstract Rapid prototyping (RP) can substantially shorten the time and reduce the c ost of developing a new product from the initial idea to production. R apid prototyping can help in recognizing the basic defects whose subsequent c orrection may prove very expensive, especially if they are made already when the product is ready for production. There are also many restrictions of RP procedures primarily i n the number of available materials and their properties, which may differ sign ificantly from the properties of end product materials. In the work, based on the stipulated standards on the machin es for 3D printing ( ZPrinter 310 Plus ) and hybrid Polyjet technique ( Objet Eden 330 ), adequate test specimens were made and with adequate equipment, the analys is of the dimensions, roughness of surfaces, and mechanical properties of prototype te st specimens was carried out. Then, based on the data obtained by testing of prope rties, a critical commentary has been provided regarding the data stipulated by their produce rs. Keywords: 3 D printing , hybrid P o lyjet technique, roughness of surface, mechanical proper ties of materials. 1.0 Introduction Rapid prototyping (RP) is a procedure of producing models. There are various methods of rapid prototyp ing. The main advantage lies in the speed of producing p hysical prototypes, and almost unlimited complexity of geometry. However, compared with CNC processing, th e main drawback of these processes is that they are 133 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING currently limited to fewer materials. The objective of the work was to find out the actual possibiliti es of RP procedures and materials for achieving maximal prec ision and accuracy of prototype dimensions. 2.0 E xperiment Scheme The test specimens were made by 3D printing procedu res on the ZPrinter 310 Plus machine and by Polyjet procedure on the Objet Eden 330 machine. The materials used for the test specimens made on the Objet Eden 330 are VeroBlack , VeroBlue and FullCure 720 , and on ZPrinter 310 Plus powder zp 102 , binding agent zb 56 and reinforcers glue Loctite 406 and Loctite Hysol 948 3 A&B . In determining the dimensions of the test specimens the digital calliper "Mitutoyo", with the measurem ent range 0 – 150/0.01mm, was used. A "Perthometer S8P" instrument was used to test the surface roughness. The surface roughness is determ ined perpendicularly to the direction of production. Mea surement is performed at three places (at the begin ning, in the middle, and at the end of the test specimen). To determine the mechanical properties the tester " Messphysik Beta 50 - 5" shall be used. Tests were performed at a temperature of 20 °C. 3.0 Shapes and Materials of Test Specimens The flexural properties of rigid and semi-rigid pol ymers in defined conditions are determined accordin g to standard ISO 178: 2001. Three-point testing is applied, i.e. the test speci men (Figure 1) has to be supported by two supports and loaded in the middle by force F, until the test specimen fractures or until the de flection reaches certain values. [1] F R 1 R 2 h 5 ° L /2 L l Fig. 1. Shape of test specimen for flexural testing [1] 134 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Dimensions of test specimens stipulated by the stan dard is length l = 80 ± 2 mm, width b = 10 ± 0.2 mm, thickness h = 4 ± 0.2 mm, loading radius R 1 = 5 ± 0.1 mm and support radii R 2 = 5 ± 0.2 mm. [1] The tensile properties of rigid and semi-rigid poly mers are determined according to standard ISO 527: 1993. Dimensions of tensile test specimens is total length l3 = ≥150 mm, length of the narrow parallel part l1 = 80 ±2 mm, radius r = 20 ÷25 mm, distance between expanded parallel part l2 = 104 ÷113 mm, width at the end b2 = 20 ±0.2 mm, width of the narrow end b1 = 10 ±0.2 mm, thickness h = 4 ±0.2 mm, measurement length L 0 = 50 ±0.5 mm and initial distance between the machine jaws L = 115 ±1 mm. [2] For measuring the elasticity modulus, the testing s peed has to be 1 mm/min. Figure 2 shows the shape o f the test specimen for tensile testing. l l l 3 r 2 1 1 h L 0 b 2 L b Fig. 2. Shape of test specimen for tensile testing [2] Test specimens made by 3D printing technique using machine ZPrinter 310 Plus, have been made of powder zp 102 , binding agent zb 56 and reinforced by adhesives Loctite 406 and Loctite Hysol 948 3 A&B . The time necessary for printing of test specimens was 31 min utes, and another 45 minutes for cleaning and reinf orcing. The white test specimens have been reinforced by cy anoacrylate Loctite 406 , whereas the yellowish ones by epoxy resin Loctite Hysol 948 3 A&B . Test specimens produced by Polyjet technique using the Objet Eden 330 , are made of the material produced by the company Objet : VeroBlack (black test specimens), VeroBlue (blue test specimens) and FullCure 720 (transparent test specimens). The time necessary fo r the production of all test specimens of a single material is about 1 hour and 10 minutes, and another 10 minutes for cleaning. 4.0 Results of Analyzing the Rapid Prototyping Materials Properties Table I presents the dimensions of test specimens f or flexural testing and for tensile testing. 135 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Table I: Dimensions for determining flexural and te nsile properties of test specimens 3 D PRINT IN G FLEXURAL PROPERTIES TENSILE PROPERTIES l[mm] b[mm] h[mm] l3 [mm] h[mm] b1 [mm] b2 [mm] 406 1 80.84 10.86 4.45 4 0 6 1 150.87 4.41 10.82 20.91 4 0 6 2 80.95 10.78 4.38 4 0 6 2 150.91 4.44 10.81 20.91 4 0 6 3 80.92 10.84 4.30 4 0 6 3 150.10 4.41 10.73 20.82 4 0 6 4 80.91 10.85 4.46 4 0 6 4 150.87 4.45 10.75 20.83 4 0 6 5 80.99 10.71 4.33 4 0 6 5 150.77 4.26 10.83 20.95 4 0 6 6 80.86 10.76 4.47 4 0 6 6 150.90 4.49 10.81 20.80 x 80.91 10.80 4.4 x 150.74 4.41 10.79 20.87 S 0.056 0.06 0.073 S 0.316 0.079 0.04 0.061 Hysol 1 80.56 10.62 4.37 Hysol 1 150.57 4.28 10.56 20.59 Hysol 2 80.66 10.66 4.25 Hysol 2 150.53 4.20 10.60 20.62 Hysol 3 80.71 10.57 4.35 Hysol 3 150.61 4.27 10.66 20.68 Hysol 4 80.69 10.70 4.30 Hysol 4 150.65 4.30 10.59 20.62 Hysol 5 80.66 10.63 4.33 Hysol 5 150.51 4.28 10.56 20.68 Hysol 6 80.62 10.61 4.28 Hysol 6 150.55 4.29 10.59 20.66 x 80.65 10.63 4.31 x 150.57 4.27 10.59 20.64 S 0.054 0.044 0.045 S 0.052 0.036 0.037 0.037 P O L Y J ET TECH N IQ U E VeroBlack 1 80.12 10.13 4.02 VeroBlack 1 150.28 4.03 10.08 20.04 VeroBlack 2 80.18 10.11 4.02 VeroBlack 2 150.28 4.01 10.04 20.04 VeroBlack 3 80.15 10.12 4.02 VeroBlack 3 150.29 4.02 9.99 20.07 VeroBlack 4 80.12 10.11 4.02 VeroBlack 4 150.27 4.00 10.06 20.01 VeroBlack 5 80.16 10.12 4.03 VeroBlack 5 150.26 4.00 10.12 20.08 x 80.15 10.12 4.02 x 150.28 4.01 10.06 20.05 S 0.026 0.009 0.005 S 0.012 0.013 0.048 0.028 FullCure 1 80.22 10.07 3.97 FullCure 1 150.18 3.99 9.99 19.97 FullCure 2 80.19 10.04 4.03 FullCure 2 150.22 4.01 10.01 20.01 FullCure 3 80.23 10.06 4.01 FullCure 3 150.19 4.02 10.00 20.02 FullCure 4 80.25 10.05 3.98 FullCure 4 150.15 4.01 9.99 19.99 FullCure 5 80.27 10.05 3.98 FullCure 5 150.22 4.00 9.98 19.99 x 80.23 10.05 3.99 x 150.19 4.01 9.99 19.99 S 0.03 0.012 0.025 S 0.03 0.012 0.012 0.021 VeroBlue 1 80.12 10.08 4.01 VeroBlue 1 150.17 4.01 10.01 20.00 VeroBlue 2 80.10 10.08 4.01 VeroBlue 2 150.19 4.00 9.98 20.03 VeroBlue 3 80.11 10.09 4.02 VeroBlue 3 150.21 4.00 10.00 19.98 VeroBlue 4 80.13 10.09 4.01 VeroBlue 4 150.21 3.99 10.00 19.97 VeroBlue 5 80.10 10.09 4.02 VeroBlue 5 150.21 4.03 9.97 20.02 x 80.11 10.09 4.01 x 150.2 4.01 9.99 20.00 S 0.013 0.007 0.007 S 0.018 0.016 0.017 0.025 136 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING The obtained results of measurement (Table I) show that the values of the dimensions obtained on ZPrinter 310 Plus ( Loctite 406 and Loctite Hysol ) are greater than on Objet Eden 330 ( VeroBlack , VeroBlue and FullCure ), because the machine Objet Eden 3 3 0 produces layers of 16 µm, and ZPrinter 310 Plus of 89 µm. For length l the deviation of 80 ± 2 mm is stipulated and the length l3 can be ≥ 150 mm , which means that all the dimensions are within tolerance limits, whereas the width b (10 ± 0.2 mm), b1 (10 ± 0.2 mm), b2 (20 ± 0.2 mm) and thickness h (4 ± 0.2 mm) on Loctite 406 and Loctite Hysol test specimens exceed the tolerance limits. Such deviations occur since the material is rougher (i.e. powder particles), but the assumption is tha t they will not affect further testing of mechanical properties . On Loctite 406 test specimens the arithmetic mean x of the mean arithmetic deviation of profile R a = 15.68 µm, and of Loctite Hysol is R a = 13.99 µm, which means that the level of roughness is N10. Substantial difference is noticed already in VeroBlack test specimens. Here is R a = 1.64 µm, which is by 88% lower than R a Loctite Hysol test specimen. The level of roughness is N7. The r oughness parameters in VeroBlue test specimens are even lower, which can be seen also in Figure 3. The level of roughness is the same as in VeroBlack test specimens N7. FullCure test specimens show the best roughness of surface. The mean arithmetic deviation of profile R a = 1.00 µm, which is in comparison with the test specimens m ade by 3D printing an as much as 94% lower value. The roughness level of N6 is determined according to R a (Figure 3). Index : red – Loctite 406 test specimen, green – Loctite Hysol test specimen, black – VeroBlack test specimen, blue – VeroBlue test specimen, pink – FullCure test specimen Fig. 3. Surface roughness of all test specimens Test specimens of powder zp 102 fractured during bending, whereas this was not the case with other materials, so that flexural stress at break σfp and flexural strain at break εfp are not calculated for them. Loctite 406 and Loctite Hysol test specimens fractured before deflection stipula ted according to standard ISO 178: 2001 of Sc = 1.5·h = 6 mm, so that the flexural stress at con ventional deflection σfC is not determined for them. The testing was done at a speed of 2 mm/min. Table II s how the flexural properties of the test specimens, whose are calculated according to the standard ISO 178: 2 001[1]. Test specimens made of powder zp 102 break at strain εfp from 0.5 to 1%. Test specimens made of materials VeroBlack , VeroBlue and FullCure did not fracture to deflection Sc = 1.5· h = 6 mm stipulated by the standard, but rather the specimen falls in between the suppor ts at: σfp = 33.7 MPa, εfp = 17.6 % and deflection S = 30 mm. 137 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Tensile properties is calculated according to the s tandard ISO 527: 1993 [2]. Table III to present the calculated values of tensile properties of test spe cimens. Table II: Flexural properties of test specimens Ef[GPa] A0 [mm 2 ] Fmax [N] σfM [MPa] εfM [%] Smax [mm] σfp [MPa] εfp [%] 4 0 6 1 2.791 48.33 32.60 14.55 0.684 1.050 14.36 0.75 4 0 6 2 2.505 47.22 30.35 14.09 0.716 1.117 13.04 0.822 4 0 6 3 2.840 46.61 36.00 17.24 0.881 1.399 16.15 0.982 4 0 6 4 2.928 48.39 33.75 15.01 0.678 1.037 14.54 0.743 4 0 6 5 2.616 46.37 29.25 13.98 0.671 1.058 13.43 0.769 4 0 6 6 2.179 48.10 40.50 18.08 0.892 1.363 17.08 0.911 x 2.643 47.50 33.74 15.49 0.754 1.171 14.77 0.83 S 0.275 0.893 4.093 1.739 0.104 0.166 1.565 0.097 Hysol 1 3.230 46.41 33.70 15.95 0.531 0.829 15.95 0.571 Hysol 2 3.168 45.30 25.85 12.89 0.404 0.648 12.34 0.436 Hysol 3 2.929 45.98 24.75 11.88 0.511 0.802 10.08 0.558 Hysol 4 2.534 46.01 28.10 13.64 0.539 0.855 12.01 0.547 Hysol 5 2.718 46.03 31.50 15.17 0.520 0.820 14.62 0.564 Hysol 6 3.337 45.41 28.10 13.88 0.524 0.835 12.77 0.538 x 2.986 45.86 28.67 13.90 0.505 0.798 13.08 0.536 S 0.315 0.421 3.385 1.482 0.050 0.076 1.876 0.05 Ef[GPa] A0 [mm 2 ] Fmax [N] σfM [MPa] εfM [%] Smax [mm] σfc [MPa] VeroBlack 1 2.430 40.72 133.8 78.46 4.856 8.246 68.58 VeroBlack 2 2.493 40.64 132.7 77.97 4.848 8.233 68.07 VeroBlack 3 2.294 40.68 134.9 79.22 5.067 8.604 68.65 VeroBlack 4 2.234 40.64 132.7 77.97 5.082 8.630 66.75 VeroBlack 5 2.326 40.30 136.1 80.45 5.161 8.742 68.48 x 2.355 40.57 134.04 78.81 5.003 8.491 68.11 S 0.105 0.171 1.469 1.048 0.142 0.235 0.791 VeroBlue 1 2.475 40.42 148.4 87.92 5.272 8.975 73.95 VeroBlue 2 2.524 40.42 145.1 85.94 4.902 8.346 73.95 VeroBlue 3 2.565 40.56 147.4 86.75 5.041 8.561 74.83 VeroBlue 4 2.527 40.46 146.2 86.50 4.834 8.229 74.52 VeroBlue 5 2.479 40.56 147.4 86.75 4.858 8.250 74.83 x 2.514 40.48 146.9 86.77 4.981 8.472 74.42 S 0.037 0.071 1.273 0.722 0.181 0.310 0.444 FullCure 1 2.628 39.98 158.6 95.93 4.908 8.439 81.63 FullCure 2 2.596 40.46 164.2 96.67 4.962 8.405 83.43 FullCure 3 2.639 40.34 155.2 92.10 4.738 8.066 80.09 FullCure 4 2.754 39.80 159.7 96.79 4.916 8.431 81.79 FullCure 5 2.798 39.80 159.7 96.79 4.962 8.511 81.79 x 2.683 40.08 159.48 95.66 4.897 8.37 81.75 S 0.088 0.308 3.22 2.020 0.092 0.175 1.183 138 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Table III: Tensile properties of test specimens E [GPa] A0 [mm 2 ] Fm [N] R M [MPa] R p [MPa] εK [%] εp [%] 4 0 6 1 3.645 47.72 321.6 6.740 6.693 0.353 0.537 4 0 6 2 2.708 48.00 304.8 6.349 6.349 0.152 0.387 4 0 6 3 3.623 47.32 275.5 5.822 5.798 0.138 0.298 4 0 6 4 4.463 47.84 343.0 7.169 7.169 0.260 0.421 4 0 6 5 3.078 46.14 262.0 5.679 5.630 0.157 0.341 4 0 6 6 2.935 48.54 307.0 6.325 6.279 0.322 0.537 x 3.409 47.59 302.3 6.347 6.32 0.230 0.420 S 0.638 0.816 29.66 0.557 0.568 0.094 0.100 Hysol 1 3.897 45.20 218.1 4.827 4.827 0.058 0.182 Hysol 2 2.902 44.52 257.5 5.784 5.784 0.095 0.294 Hysol 3 2.515 45.52 238.4 5.237 5.212 0.078 0.285 Hysol 4 2.472 45.54 253.0 5.556 5.507 0.089 0.312 Hysol 5 2.497 45.30 246.3 5.436 5.411 0.097 0.314 Hysol 6 2.819 45.43 219.3 4.827 4.827 0.103 0.274 x 2.641 45.262 242.9 5.368 5.348 0.092 0.296 S 0.203 0.425 15.03 0.362 0.357 0.009 0.017 VeroBlack 1 3.612 40.62 1868 45.98 33.00 7.047 7.960 VeroBlack 2 2.794 40.26 1826 45.36 33.24 6.090 7.280 VeroBlack 3 2.905 40.16 1859 46.29 33.82 6.466 7.630 VeroBlack 4 2.796 40.24 1842 45.77 33.25 6.524 7.714 VeroBlack 5 2.723 40.48 1875 46.31 34.42 5.610 6.875 x 2.966 40.35 1854 45.94 33.55 6.347 7.492 S 0.367 0.191 19.94 0.396 0.574 0.535 0.422 VeroBlue 1 2.7 40.14 2088 52.02 37.59 5.626 7.019 VeroBlue 2 2.726 39.92 2057 51.52 38.96 5.130 6.559 VeroBlue 3 2.505 40.0 2049 51.22 41.94 4.347 6.022 VeroBlue 4 2.459 39.9 2042 51.18 39.29 4.780 6.375 VeroBlue 5 2.699 40.18 2077 51.69 36.35 5.828 7.175 x 2.618 40.03 2063 51.53 38.83 5.142 6.630 S 0.125 0.127 19.32 0.348 2.097 0.606 0.471 FullCure 1 2.561 39.86 2196 55.10 40.0 4.820 6.382 FullCure 2 2.832 40.14 2210 55.05 36.56 5.904 7.195 FullCure 3 2.073 40.20 2224 55.33 40.20 4.527 6.465 FullCure 4 4.174 40.06 2198 54.88 36.41 6.7 7.572 FullCure 5 2.416 39.92 2200 55.10 40.25 4.717 6.384 x 2.811 40.04 2206 55.09 38.68 5.334 6.8 S 0.810 0.144 11.61 0.161 2.010 0.934 0.550 Loctite Hysol test specimen 1 features very large deviations and is therefore eliminated from further analysis. 139 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Test specimen of powder, binding and reinforced by cyanoacrylate ( Loctite 406 test specimen) shows slightly better properties than with epoxy resin ( Loctite Hysol test specimen). The analysis of flexural properties (Figure 4) and the analysis of tensile properties (Figure 5) shows that the best properties belong to the test specimens made o f FullCure materials. Compared to the properties declared by the producers [3] flexural strength σfM is somewhat higher, whereas flexural modulus Ef is higher by an average of 100 MPa. Tensile strength R M is a bit lower, whereas the modulus E is approximately the same. Tensile strain at break εp [4] is much higher than the one obtained by analys is ( εp = 6.8%). In case of Vero material the producer guarantees in VeroBlack the highest flexural modulus. The performed analysis shows that in this group of materials the flexural modulus is the lowest. In VeroBlue materials the values guaranteed by the producers almost match wit h the obtained ones, except in case of tensile stra in at break εp which is much higher ( εp = 15 – 25%) than obtained by the analysis ( εp = 6.63%). VeroBlack material has the highest tensile modulus E = 2966 MPa in the group of materials made by Polyjet technique. The lowest flexural and tensile properties are feat ured by the test specimens of powder made by 3D printing ( Loctite 406 and Loctite Hysol test specimens). In comparison with FullCure and Vero materials the flexural strength σfM is very low, as low as σfM = 15 MPa, but the flexural modulus Ef is similar. Tensile strength R M and tensile stress at break R p are substantially lower than in FullCure and Vero materials, but the elasticity modulus E in Loctite 406 test specimens is the highest of all the materials . It should be noted that their properties depend on the hardening agent which has been added to the powder and binding agent. Better properties are featured by the reinforcing agent cy anoacrylate. Fig. 4. Diagram of flexural stress - strain of all materials 140 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Fig. 5. Diagram tensile stress - strain of all materials 5 0 Conclusion Rapid prototyping allows production of parts of ver y complex shapes the production of which, until the occurrence of these techniques, had been limited. T he rapid prototyping techniques have been developin g intensively from day to day. Here, the limiting num ber is of the available materials and their propert ies, which substantially differ from the properties of materia ls of final products. Therefore, it is necessary to know the properties of the prototypes materials. The results of measuring the dimensions of test spe cimens show that the instrument Objet Eden 330 is more precise in production than ZPrinter 310 Plus . The surface of test specimens made of powder zp 102, binding zb 56 and reinforced by cyanoacrylate Loctite 406 or epoxy resin Loctite Hysol 9483 A&B is rougher than VeroBlack and VeroBlue material. FullCure material shows the lowest value of surface roughnes s. The best mechanical properties belong to test speci mens made of FullCure . For Vero material the producer stipulates that VeroBlack has the highest flexural modulus, and still the an alysis results show that it is the lowest one. In VeroBlue materials the producer’s guaranteed values almost match the obtained ones. The worst mechanical properties are featured by the test spec imens made of powder. However, their properties dep end on the reinforcing agent which is added to powder o r binding. Cyanoacrylate features better properties than epoxy resin. Manufacturer’s estimates of the properties and prop erties obtain by the experiments is approximatelly the same. It is only recommended to change tensile stra in at break to εp = 15 – 25%, as well as flexural modulus to Ef = 2400 MPa, specially for VeroBlack materials. For materials 4 0 6 and Hysol some parameters may be changed such as e.g., position of the model in the work space, layer thickness, ratio of saturation by binding etc., which also influences the properties of proto types (strength, roughness, dimensions). 141 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPERIMENTAL ANALYSIS OF PRO P ERTIES OF MATERIALS FO R RAPID PROTOTYPING Acknowledgement This work is part of the research included in the p roject Increasing Eff iciency in Po lym eric Products and Processing Develo p m e nt and in the Advanced Techno lo g ies of Direct Manufacturing of Po lym eric Products . There are included in program Rapid Production – From Vision to Reality supported by the Ministry of Science, Education and Sports of the Re public of Croatia. The authors would like to thank the Ministry for the financing of this project. References [1] ISO 178: 2001 Plastic – Determination of flexural p roperties [2] ISO 527: 1993 Plastic – Determination of tensile pr operties [3] Objet, www.2objet.com , 29.04.2006. [4] Design in Site, www.designinsite.dk , 24.04.2006. 142 Advanced Manufacturing Systems 143 144 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL ADVANCED PRODUCTIZATION REFERENCE MODEL Jussi Kantola 1 , Aulis Tuominen 2 , Hannu Vanharanta 1 1. Tampere University of Technology at Pori, Departme nt of Industrial Management, Pohjoisranta 11, PL 30 0 , 28 1 0 1 Pori, Finland 2. University of Turku, Electronics Product Developme nt Research Group, Ylhäistentie 2, 241 3 0 Salo, Finland Abstract Productization can be understood as all those actions and operati ons that ensure the process from an idea to a product that sells well. Productizat ion is a business process in companies. How well this process works basically defines the future success of the company. Products today are complex and people from several differ ent knowledge disciplines are needed to do their part and communicate wi th each other before an idea becomes business. The required steps in productizat ion are neither simple nor standard. Depending on the idea, the product in mind, the sp ecific scenario, set-up and involved parties, the process can take multitude of forms. Ever shorter time requirements on turning an idea into a ready product set addi tional challenges for all the involved parties. This paper proposes a new type of reference model for indus trial productization. Reference models and reference architectures are already u sed in many other disciplines. The proposed model targets similar goals to existi ng reference models in other disciplines. The proposed model will take an ontological format. Academic and business people who are involved in productization through a vari ety of roles can realize the benefits of developing and using such model. Keywords: Reference model, operations, management, productization, process, ontology 1.0 Introduction Every year lots of good ideas are born. Most of tho se ideas never become good business. According to a very recent study conducted by Sitra, the Finnish Innova tion Fund shows that around 80% of new ideas in Fin land never become products that sell well. At universiti es, this percentage is still much higher. According to 145 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL Tuominen productization refers to everything betwee n an idea and a successful product (a product that sells well) [1]. We can see that idea generation is not a problem in Finland but something must be wrong in how the productization is understood and utilized in practi ce. Understanding the concepts in the productizatio n domain must be the basic requirement for all stakeholders. How well productization works in practice is based on people’s perceptions on those concepts case by case , since people act and decide according to their perceptions. The approach described in this paper looks at the l ow success rate in productization in a new way. Fir stly, the idea is to clarify the real concepts in a productiz ation domain. Secondly, to make empirical research based on the “clarified” concepts in order to find out how t o improve the implementation of those concepts in r eal life. Thirdly, provide practical guidelines and decision support for managers. In this research the concepts in the productization domain are explicitly specified and organized so t hat they form a structure of classes and sub-classes. When c oncepts are organized that way, they form an ontolo gy. After the ontology is specified, it forms the conce ptual framework for the productization research and practice. Then, the empirical study towards the real cure for low success rate in productization can begin. Each class (concept) alone refers to some cure, but most impor tantly they together show a holistic way what to do . The bottom-line is that all parties involved in product ization should work, develop and manage the actual and timely concepts. This can only be achieved by utili zing productization ontology as a base for all acti vity. Such ontology naturally evolves when the world is changi ng. This research project is conducted at the Unive rsity of Turku, Product Development Research Group. The goal is to build a generic model that can be used in an y productization discipline. Similar ontology based m anagement methodology has successfully been applied to Human Resource, Investment, Business and Organizati onal process management in Finland, Poland, Spain, South Korea, UK and US [2]. The motivation for this research is to increase the success rate in produc tization. If it can be done, there will be significant benefi ts for the academia and business. 2.0 Reference Models First, we need some basic terms. A process is a sys tematic series of actions directed to some end [3]. A process in a business context is understood as a bu siness process, which is defined as a series of ste ps designed to produce a product or service [4]. According to t he Random House Webster’s Dictionary, the meaning o f a reference is: the act of referring; a source of fac ts or information [3]. Not many definitions for a r eference model can be found in the literature, but with comm on sense we can understand that a reference model i s a conceptual representation of a process or a system. Since reference models have such a fundamental rol e, they can serve as the conceptual basis to develop specif ic models and applications. These specific models a nd applications are used by practitioners and academic s. In the next paragraph some known reference model are briefly explained. • SC O R (Supply-Chain Operations Reference) model describe s the business activities associated with all phases of satisfying a customer's demand. The S COR model has three main pillars: process modeling, performance measurements, and best practi ces (plan, source and deliver) [5]. SCOR users can address, improve, and communicate supply chain management practices within and between all 146 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL the interested parties. SCOR model is developed by the Supply Chain Council members on a volunteer basis [6]. • O SI (Open Systems Interconnection) model is a basic re ference model for a communications and computer networking design. The reference model has seven layers: 1. Physical, 2. Data link, 3. Network, 4. Transport, 5. Session, 6. Presentation, and 7. Application. • F EA (Federal Enterprise Architecture) is a collection of reference models that are being developed by the US government [7]. The FEA’s foundation is t he business reference model which describes the government’s lines of business and its services and provides a common framework for improvement in key areas such as: budget allocation, informatio n sharing, performance measurement, budget/performance integration, cross-agency collab oration, e-government, and component-based architectures. The goal is to simplify processes an d unify work across the agencies. The goal is more citizen-centered, customer-focused government that maximizes technology investments to better achieve mission outcomes [7]. All the different kinds of reference models describ ed above target similar goals: improve processes, communicate and co-operate better, enable applicati on development, and serve customers better. For the productization reference model described in this pa per the goals are the same. 3.0 Ontologies Ontology is defined as the specification of the con ceptualization of a domain [8]. Conceptualization i s the idea of (part of) the world that a person or a group of people can have [9]. Ontology defines the common wo rds and concepts (meanings) used to describe and represent an area of knowledge [10]. Ontologies represent a m ethod of formally expressing a shared understanding of in formation [11]. The main parts of ontologies are cl asses (concepts), relations (associations between the con cepts in the domain) and instances (elements or ind ividuals in ontology) [9]. Using ontologies can have several benefits, such as interoperability, browsing and searching, reuse an d structuring of knowledge [12]. Ontologies also enab le the computational processing of information. Ontologies are becoming increasingly important in f ields such as knowledge management, information integration, co-operative information systems, info rmation retrieval and e-commerce [13]. When applied they serve needs such as storage, exchange of data corre sponding to an ontology, ontology-based reasoning o r ontology-based navigation [14; 15]. Ontologies prom ise a shared and common understanding of a domain t hat can be communicated between people and application systems [16]. In addition, ontologies enable comput er processing to retrieve and use information in many other ways. The two above mentioned aspects, combin ed with opportunities provided by contemporary Interne t programming technology, makes ontologies a very attractive approach to tackle productization proble ms in real life. 147 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL Ontologies have an important role in defining the c oncepts and their relations for application develop ment. The reference model of a business process works as the basis for more specific ontology development in that domain. The developed ontologies form a solid base for management, development, communication and operations in the domain - productization domain in this case. 4.0 Advanced Productization Reference Model People are “connected” to productization process th rough many different kinds of roles, such as operat ional roles, developer roles, management roles and academ ic roles. It is crucial that the people in these ro les are able to be part of the productization process to the bes t they can [17]. However, this is very difficult wi thout an independent and reliable source of the concepts in the productization process. Now, we can see that ac tually ontological approach is truly needed. Such ontology enables good participation in the process in each case from the perspective of the different roles. Typical way to use ontology is to create first a to p-level ontology that specifies the most important basic concepts and their inter-relations in the domain, F igure 1. Based on the top-level ontology, more spec ific domain ontologies can be specified. In this context , we can call these more specific domain ontologies as Productization Ontologies (POs). They define the ba sic structure for application development. The deve loped applications serve management, (process) developmen t, communications and operations [17], Figure 1. Fig. 1. The productization reference model serves a s a base for ontology development. Specific ontolog ies (POs) form the base for application development for manag ement, development, communication and operational p urposes. The approach shown on the Figure 1 also enables all the stakeholders in all the above mentioned roles to “speak the same language” when doing the productiza tion work. Management must take into account stakeholders’ per ception of productization cases during their life-c ycles. Perception plays a very important role in an organi zation’s management, since people think and act bas ed according to their perception. People also envisage the future of each productization case. This is a very important element to consider, since this tension b etween current reality and envisaged future indicat es the direction for peoples’ thoughts and actions regardi ng the case in the future [2]. An individual’s perc eption is composed of such things as accumulated knowledge an d experience, individual’s situation, values and Productization Ref. Model / Top - level o ntology PO PO PO Applicat ion Applicat ion Applicat ion Management, Development, Communications & Operations 148 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL Instanc es Explicit Action Targeted Plan Cycle (1/2/3/6 m) Instances Instanc es Focus Instanc es Instanc es attitudes regarding the specific case. Therefore, a perception of a productization case combines an in dividual’s body, mind, vision, PO, perception of the PO, conte xt and time. Modern on-line tools enable people to globally give their perception on productization ca ses for collaboration and management purposes [2]. The output of effective planning is an ever-changin g plan, one that reflects the continuous learning a nd adaptation of those who prepare it [18]. This means that also the management of productization project s must be realized in periodic cycles that are not too lon g. Continuous planning requires a continuous proces s. Such ontology based management process is shown on Figur e 2[2]. According to this process, all objects spec ified as POs can be managed in a similar way by utilizing instances through the following main steps: 1. Collecting instances within a set time window. 2. Validating (in workshops) the perceived impact o f previous development plans (if existing) and fitting them together. A. Unique instances, and B. Focus areas coming from the strategy of an organiza tion and local conditions, and C. Explicit knowledge about POs 3. Making targeted development plans for the produc tization case. 4. Taking action according to the plans. 5. Returning to Step 1 after each development cycle (1/2/3/6 months). Fig. 2. Ontology based management framework by Kant ola [2]. 149 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL These steps are generic and independent from the co ntent of PO. A PO may contain only a couple of sub- classes, tens ob sub-classes and more or less hiera rchy, but still instances are collected in the same way, which allows for the visualization of the perception of P Os. Every cycle can be seen as different, and all c ycles can, in principle, be handled in a uniform way. When we take one step upwards in decision making, w e can consider each productization case as an entit y in a certain context. The context is defined by: country , business area, company size, case type, consortiu m, etc. In order to understand these cases and find meta-knowl edge (knowledge of knowledge) for improved decision making, we need a large productization case bank wh ere the cases are stored. Then, the cases in the ca se portfolio can be examined against the case bank. Th erefore, each case can benefit from the accumulated experience of all cases, and proactive heuristic me asures can be taken in order to steer and manage ea ch case towards success. The proposed productization reference model serves many purposes: • Improve the efficiency and effectiveness of the pro ductization process. • Improve communications and co-operation between: o Academic – Academic o Academic – Business o Business – Business • Improve communications and co-operation between sta keholders in productization cases. • Serve as the best practices database. • Show the variations in productization process steps . • Provide the base for measuring of productization op erations. • Serve as a basis for specific ontology development. • Serve as a basis to develop more specific models an d applications. • Enable better development and management of the pro cess than previously. • Provide documentation. • Visualize the concepts related to the productizatio n process. 5.0 Conclusion The proposed reference model can be seen as a base for productization work in academic and business se ctors. In addition to such reference model, a management p rocess that realizes the benefits of building and developing POs is necessary. Now we can see that the success in productization c ases actually becomes a management question. By applying ontologies for “collecting” the perception s of productization cases for management and stakeh older co-operation and decision making we should be able improve the success rate. The approach described in this paper makes managers and other stakeholders more ca pable to do their work. 150 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ADVANCED PRO DU CTIZATION REFER EN C E MODEL Academics and practitioners can realize the benefit of developing and using the proposed reference mod el. Without applying such a model, the risk of failing in productization cases is significant. This articl e is a proposal and an invitation to other parties to join in this effort. Should you be interested in this r esearch, please contact us. References [1] Tuominen, A., Tuote on hyvä, kun se palvelee asiaka sta (The product is good, when it serves the custom er), Talouselämä, 52(25), 40-41, 2007. [2] Kantola, J., Ingenious Management, Doctoral thesis, Tampere University of Technology at Pori, Finland, 2005. [3] Random House Webster’s Concise Dictionary, Random H ouse, New York, 1993. [4] Rummler & Brache, Improving Performance: How to man age the white space on the organizational chart, Jo ssey- Bass, San Francisco, 1995. [5] Supply-Chain Operations reference-model, SCOR Versi on 5.6., Supply Chain Council, Inc., Pittsburg, 200 1. [6] Supply Chain Council, at http:// http://www.supply- chain.org/cs/root/home, (read March 1 st , 2008). [7] Whitehouse Internet pages, at http://www.whitehouse .gov/omb/egov/a-1-fea.html (read March 1 st , 2008). [8] Gruber, T.R., A translation approach to portable on tologies, Knowledge Acquisition, 5(2), 199-220, 199 3. [9] A. Gomez-Perez, “Ontology Evaluation”. Handbook on Ontologies. S. Staab, Studer, R. Berlin, Springer-V erlag: 251-273, 2004. [10] Obrst, L., Ontologies for Semantically Interoperabl e Systems, Proceedings of the 12th international co nference on Information and knowledge management, 2003. [11] Parry, D., A fuzzy ontology for medical document re trieval. Proceedings of the second workshop on Aust ralasian information security, Data Mining and Web Intellige nce, and Software Internationalisation, Dunedin, Ne w Zealand, Australian Computer Society, Inc, 2004. [12] Menzies, T., Cost Benefits of Ontologies, Intellige nce 10(3): 26-32, 1999. [13] Baader, F., Horrocks, I., Sattler, U., Description Logics, Handbook on Ontologies. S. Staab, Studer, R . Berlin, Springer-Verlag: 3-28, 2004. [14] Crubezy, M, Musen, M.A., Ontologies in Support of P roblem Solving, Handbook on Ontologies. S. Staab, S tuder, R. Berlin, Springer-Verlag: 322-341, 2004. [15] Oberle, D., Volz, R., Staab, S., Motik, B., An Exte nsible Ontology Software Environment, Handbook on O ntologies. Staab, B., Studer, R. Berlin, Springer-Verlag: 299- 319, 2004. [16] Davies, J., Fensel, F., Van Harmelen, F., Towards t he Semantic Web - Ontology Based Knowledge Manageme nt, Davies, J., Fensel, D., Van Harmelen, F., John Wile y & Sons, Ltd, 2003. [17] Kantola, J., Tuominen A. and Vanharanta, H., Produc tization Operations Reference Model – POR Model, Fo urth Conference On New Exploratory Technologies, NEXT, S eoul, Korea, 2007. [18] Ackoff, R.L., Management in Small Doses. New York, John Wiley & Sons, Inc., 1986. 151 152 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS P RO DUCTION SCHEDULING IN THE CORRUGATED INDUSTRY: CHALLENG ES AND SOLUTIONS Btissame Iba 1 , Essam Shehab 1 , Adrian Swindells 2 1. Decision Engineering Centre, Cranfield University, Cranfield, Bedfordshire MK43 0AL, UK 2. Abbey Corrugated, South Mills, Blunham, Bedfordshi re MK44 3PH , UK Abstract Production scheduling is a crucial and deterministic proce ss to the performance of manufacturing industries. Most companies focus on optimising this process and making it as efficient as possible. This is to lower manufac turing costs, increase production efficiency and capacity utilisation but also to impr ove customer satisfaction by delivering on time and in full. In this way, huge efforts have been focused on conducting research to improve production scheduli ng and several solutions proposed however they do not alone fully satisfy indust ry expectations. Moreover, research studies related to the corrugated industry in this are a are scarce. This paper aims to identify production scheduling issues an d practices in the corrugated industry that might be solved and improved through th e development of a decision support tool. Significant improvements and economies could be achieved through the support and assistance of human schedulers in m aking more effective scheduling decisions in order to generate better solutions in terms of cost and customer satisfaction. The investigation of a case study revealed the existence of chal lenges faced by the corrugated industry during the scheduling process. Several i ndustrial visits were conducted in the form of observation sessions, interviews and workshops with the human schedulers and the shop floor operators in order to gathe r the relevant data necessary for this research project. The analysis of these d ata showed the existence of differences in the human scheduling practices and someti mes the absence of standards related to some scheduling tasks. The case study illustrate s that several decisions involved in the search and selection of jobs to build a work m ix of customers orders for the corrugators machines such as: the evaluation of the pape r trim percentage, 153 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS deckle size, upgrading costs, target run lengths and sequenci ng of the set of orders work mix, constitute challenging tasks that require the support of a decision tool to produce better scheduling solutions. Keywords: Manufacturing; Corrugated Industry; Produ ction Scheduling 1.0 Introduction Production scheduling is all about clever and effec tive handling of information to make the right deci sions related to the efficient assignment of tasks to ava ilable resources over time [1]-[2] while satisfying production constraints and minimising costs. Interest in stud ying production scheduling initiated about half a c entury ago from a general perspective and within limited disci plines [2]-[3]-[4]. However, research in this area soon expanded to embrace more disciplines and adopt a mo re specific and detailed investigation approach. T ough extensive research studies have been done on produc tion scheduling yet studies related to the corrugat ed industry in this area are scarce. Today’s highly competitive environment made the cor rugated industry a fierce and challenging business where organisational and strategic skills became key word s of success. In addition to the highly demanding customers, there are a number of constraints faced by the industry on day-to-day basis. These constra ints are mainly related to the large variety and quantity of products ordered, the machine failures, the unpred ictable customer’s ordering behaviour, and the complex mult i-stage production process [5]. With all these cha llenges, the corrugated board industry is still required to demonstrate a high degree of flexibility and to gua rantee on time and in full (OTIF) delivery to its clients. T herefore, it is necessary to keep a certain balance between satisfying customer requirements and maintaining th e business competitive advantage. In this way, spe cial attention is to be given to the investigation of th e production scheduling process in an attempt to im prove and optimise it. This will boost the corrugated indust ry’s competitive advantage while enabling it to sat isfy its clients’ requirements. In this way, this paper wi ll investigate a case study from the corrugated ind ustry to determine the major issues related to the schedulin g process and identify solutions to cope with these . An exploratory approach based on the qualitative re search technique was adopted and the concept of deduction was relied on in the research process. I t involved the conduct of literature review, design of a questionnaire, conduct of interviews as well as obs ervation sessions for information gathering mainly from the industrial case study but also from academia. This provided enough information to understand the prod uction scheduling process at the case study company and id entify major issues faced with the production sched uling system. 2.0 Overview of the Corrugated Industry In this paper, we refer to the corrugated industry as the corrugated paperboard that has been in produ ction since the 18th century. It is a composite material that consists, in its basic form, of a three layer s structure as 154 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS illustrated in Fig.1. It is composed of an outside and inside liner holding together a corrugating me dium or flute in the middle. These three major components control the characteristics of the final corrugated board based on the type and thickness of the papers as we ll as the height and number of corrugations used. The corrugating medium is a key component in making cor rugated boards. The various flutes are combined wi th the linerboards in different ways resulting in four major forms of corrugated boards (as illustrated i n fig1): - Single face: is made by gluing one sheet of linerbo ard to another sheet of corrugating medium. - Single wall: consists of two sheets of linerboard g lued to each side of the corrugating medium. - Double wall: consists of three linerboards and two flutes glued in the following sequence: an outside liner, a flute, a center linerboard, a flute and an inside liner. - Triple wall: this form of corrugated board could be described as tree single walls glued together. It consists of four linerboards and tree corrugating m ediums. Fig. 1. Corrugated paperboard components One of the main applications of the corrugated pape rboards is packaging boxes. A multi-milliom dollar machine called corrugator is used to take wide roll s of paper, corrugate one unwinding roll and glue t he other two unwinding rolls to the fluted paper in the case of a single face board. Then takes place the slit ting and scoring of the paper lengthwise in order to cut the m to the specified measures according to clients’ requirements [6]. For instance, the case study com pany offers conventional boards that consist of 600 different combinations of paper grades and flutes b oth in single and double wall, as well as specialit y boards that are made of special materials to provide a cla y-coated surface. It differentiates itself through the provision of a wide product range on a very short lead-time t hat can reach 12hours. This is made possible throu gh the exact and efficient scheduling of its 6 world-class corrugator machines. Despite the use of an automatic scheduling system, the intervention of human planers remains essential . Hence, making manual and automatic scheduling compl ementary to each other in order to for the business to preserve its competitive advantage and satisfy its clients’ requirements. In fact, keeping a certain balance between satisfying customer requirements and mainta ining the business competitive advantage is necessa ry as 155 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS expressed by Georges Schepens, a founding partner a t OM partners: “Your customer is king, but he is n ot the emperor” [6]. 3.0 Critical Parameters in the Scheduling Process A first review of industrial documents and availabl e literature review helped draw a general high leve l view of the different parameters involved in the production scheduling process in the corrugated industry alon g with the outputs that are expected as shown in Fig2. In fact, successful automated production scheduling h as to provide an effective tool for the human schedulers to help them better manage and act upon available d ata and information in order to facilitate rapid and effici ent decision-making [7]. This is in part due to th e fact that there will always be some orders –referred to as ex ceptions in Fig2- which do not completely satisfy scheduling rules and hence cannot be processed auto matically. Fig. 3. Automated production scheduling parameters At the present day, 80% of the scheduling activity at the case study company is done by three human schedulers with the help of a system that pulls sim ilar orders together. Hence the workmix attributed to each corrugator machine is done automatically. The plan ners then manipulate the categorised orders in orde r to produce the best scheduling solutions in terms of c ost and service requirements based on their own jud gement and experience [8]. The existing off the shelf pro duction scheduling software packages are inappropri ate for the company’s business because of the use of six co rrugator machines, where boards of 1200 grades of a ny size and length with scores in any position are pro duced for next day delivery in England. The availa ble software packages do not offer the possibility of s cheduling more than one corrugator at a time. In this way, in depth study of the production sched uling system at the case-study company aims at iden tifying the key parameters used by the human planners in or der to construct scheduling solutions and determine major 156 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS issues encountered during this process. This will help in the formulation of recommendations and solu tions to improve automated production scheduling in the corr ugated industry. The investigation of the scheduli ng system involved the study of the process followed t o schedule four corrugator machines. The schedulin g process of the normal orders consists of two main s tages as illustrated in Fig.3: 1) Order allocation : is performed by the Bespoke System and consists o f assigning every order to a given corrugator mainly based on the match between orders ’ characteristics and corrugators’ characteristics in terms of flute type, width, chop length, and grade. 2) Orders sequencing and programmes formation : this stage is carried out by the human planners a nd involves the manipulation of the assigned orders in order to build the best scheduling solutions. Fig. 3. Overall view of the scheduling system phase s for the case study It is important to note that during the first stage , the filtering of orders, performed by the Bespoke System, consists of grouping similar orders together under a particular programme number. Pulling orders into the same programme number is based on a number of diffe rent criteria: - Reel size or width of paper to be used based on the order width. - Flute type. - Order due date. - Grade: each corrugator runs particular grades which can be part of the same programme. - Characteristics of the corrugators in terms of mini mum and maximum widths, maximum number out, trim, minimum lineal meters per deckle. During the second stage, the human planners need to go through the first filtering performed by the Be spoke System in order to make necessary amendments as req uired depending on the situation. For instance, th ey 157 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS may need to transfer some jobs to another corrugato r to ensure equal distribution of work between the machines. They may also need to move orders forwar d to build either the lineal meters of a specific g rade or the total lineal meters of the whole programme. Th e main objective consists of making sure the corrug ators are efficiently, cost effectively and fully utilise d by taking into consideration the following criter ia: - Minimum number of deckle changes. - Minimum percentage of trim. - Use of narrow paper widths. - Start with large widths first to avoid problems wit h line ups of papers. - Run heavy grades first. The main problem here is that some of the goals and rules for scheduling are contradictory or sometime s impossible in certain situations. It is therefore important to provide support to the human planners in delicate decision making situations to allow them to select the best and most cost effective options [9]. 4.0 Major Scheduling Issues Identified As mentioned in the previous section, human planner s face the problem of contradictory goals. For ins tance, achieving both a minimum number of paper size chang es and minimum percentage of side trim is very difficult. Practically, the only way to achieve mi nimum side trim with different combinations of orde r widths is to change roll sizes. Moreover, observation ses sions and interviews conducted with the different p lanners at the case study company enabled the researcher to: - Identify the major decision making situations encou ntered by the planners when performing the scheduling. - Determine the rules and constraints taken into cons ideration when dealing with each decision making situation. - Identify differences in decision making between the different planners. Following the analysis of the data collected from t he observation sessions and the interviews, a set o f decision making situations has been identified as illustrate d in Table4: Decision Making Situation Description Corrugator scheduling priority Corrugators are sche duled based on the types of orders available and the availability of machines Assignment of exception orders Orders which do not obey to the company standards are not automatically allocated to corrugators. They are s tored in an 158 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS to corrugators exceptions file waiting to be approv ed and assigned by schedulers. Sequencing of orders within the same programme Schedulers need to follow a number of rules such as avoiding small orders at the start or end of a programme in order to insure orders are run smoothly at the shop floor level Sequencing of programmes Schedulers need to choose the deckle sizes to run in such way that they start with a similar deckle as the last one fi nished on and they introduce deckle jumps to allow time for paper chan ge at the shop floor level Searching for jobs to build programmes The schedulers need to perform a careful and comple te search of all the orders stored to select appropriate orders and avoid upgrading orders that contribute to increasing the overall co sts Fig. 4. Corrugated paperboard components One of the main issues noticed within most of the d ecision making situations is the non existence of s tandard rules applied by all the planners which was noted a t the shop floor level when receiving different sch eduling solutions. In fact, in most cases each planner fol lows his own experience when confronted to critical choices and which does not always lead to the best solution . Another issue is related to the searching approa ch adopted by the planners to find appropriate jobs ne cessary to build programs. If the searching proced ure is not efficient it may lead to omissions of potential job s. This may induce the planner to the easy solutio n of upgrading jobs to build programs which results in a dditional costs that could otherwise be avoided. A nother issue related to the searching procedure and the se lection of inappropriate jobs is the high number of grade and flute changes that affect the quality of the schedu ling solutions. In fact, the higher the number of grade and flute changes the higher the percentage of paper br eaks and hence the amount of waste. Last but not l east, the schedulers tend to underestimate the importance of selecting the appropriate sequenced combination of jobs’ chop lengths. This is extremely important because Different chop lengths restrict the running speed o f the corrugators. 5.0 Conclusion Most of the reviewed literature related to producti on scheduling, emphasised the following issues [10] -[11]- [12]-[13]: - Computer-based solutions to manufacturing productio n scheduling are rarely successful in satisfying industry requirements. - The importance of the investigation of the domain o f integrated human and computer-based scheduling systems. 159 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P R O D U CTION SCH ED U LING IN THE CORRU GATED INDUSTRY: C HALLENGES AND SOLUTIONS In fact, more research is to be directed towards th e human factor and the development of efficient and tailored decision support tools to assist human planners in the generation of cost-effective scheduling solutio ns in the corrugated industry in particular. Indeed, Product ion scheduling in the corrugated industry has not r eceived enough attention and consideration from researchers compared to the more general domain of manufacturi ng. It is therefore critical to determine the appropria te and required computer-based support to make prod uction scheduling in the corrugated industry more efficien t and cost-productive. Acknowledgement The authors would like to express thanks to the Cra nfield University decision engineering centre and s chool of applied sciences for their support for this work un der the overseas research students awards scheme. They also wish to acknowledge the assistance, support an d vital contribution of the operations director, pl anners and shop floor crew at the case study company. References [1] Herrmann, J.W. (2004), "Information Flow and Decisi on-Making in Production Scheduling", IIE Annual Conference and Exhibition , pp. 1811-1816. [2] Rodammer, F.A.; White, K.P., Jr. (1988), "Recent su rvey of production scheduling", IEEE transactions o n systems, man, and cybernetics, vol. 18, no. 6, pg. 841 -851. [3] Crawford, S. and Wiers, V.C.S. (2002). From anecdo tes to theory: A review of existing knowledge on hu man factors of planning and scheduling. In: Human Performance in Planning and Scheduling, edited by MacCarthy, B. L., Wilson, J.R., London, pp. 15-45. [4] Melnyk, S. A., Vickery, S. K. and Carter, P. L. (19 86), "Scheduling, sequencing, and dispatching: Alte rnative perspectives", Production and Inventory Management, vol. 27, no. 2, pp. 58-68. [5] Darley, V. and Sanders, D. (2004), "An agent-based model of a corrugated box factory: The trade-off be tween finished-goods-stock and on-time-in-full delivery", Proceedings 5th workshop on agent-based simulation . Available: http://www.santafe.edu/~vince/pub/CorrugatedBoxFact ory.pdf (Accessed on 20th November 2006). [6] Kenny, J. (2006), "It takes insight", Paperboard Pa ckaging, vol. 91, no. 8, pp. 23-27. [7] Miceli, J. (2005), "Harness the power of real-time data", Paperboard Packaging, vol. 90, no. 8, pp. 19 s-21s. [8] Abbey Corrugated Limited, Welcome to the Abbey webs ite. Available: http://www.abbeycorrugated.co.uk/. [9] Cloud, F.H. (1995), The art and science of corrugat or scheduling by manual and computer methods, Jelma r Publishing Co, NY. [10] Crawford, S. and Wiers, V.C.S. (2002). From anecdo tes to theory: A review of existing knowledge on hu man factors of planning and scheduling. In: Human Performance in Planning and Scheduling, edited by MacCarthy, B. L., Wilson, J.R., London, pp. 15-45. [11] Crawford, S., MacCarthy, B.L., Wilson, J.R., and Ve rnon, C. (1999), "Investigating the work of industr ial schedulers through field study", Cognition Technology & Work, vol. 1, pp. 63-77. [12] Haessler, R. W. and Talbot, F. B. (1983), "A 0-1 mo del for solving the corrugator trim problem", Manag ement Science (pre-1986), vol. 29, no. 2, pp. 200-210. [13] Jackson, S., Wilson, J. R. and MacCarthy, B. L. (20 04), "A new model of scheduling in manufacturing: T asks, roles, and monitoring", Human factors, vol. 46, no. 3, pp. 533-551. 160 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S AN INDUSTRIAL IMPLEMENTATION OF A METHODOLOG Y FOR THE OPTIMISATION OF PRODUCT DESIGN TOLERANCES R. Eadie 1 , J.Gao 2 1. Edwards Vacuum, Dolphin Rd, Shoreham-by-Sea, UK ross.eadie@edwardsvacuum.com 2. School of Engineering, University of Greenwich, Cha tham Maritime, Kent, UK j.gao@gre.ac.uk Abstract Today, the key to commercial success in manufacturing is the t imely development of new products that are not only functionally fit for purpose b ut offer high performance and quality throughout their entire lifecycle. In princi ple, this demands the introduction of a fully developed and optimised product from the outset. To accomplish this, manufacturing companies must leverage existi ng knowledge in their current technical, manufacturing and service capabilities. T his is especially true in the field of tolerance selection and application, the subjec t area of this research. Tolerance knowledge must be readily available and deployed as an in tegral part of the product development process. This paper describes a method ology and framework, currently under development in a UK manufacturer, to achieve this ob jective. Keywords: CAD, design, knowledge, tolerance analysi s, optimisation. 1.0 Introduction An important design resource in any manufacturing o rganisation is the ability to leverage the historic al knowledge contained within the entire organisation. Figure 1 illustrates the different domains this k nowledge can originate from, and whether the sources are int ernal or external to the manufacturing organisation . Sources of information outside the company include trends of product sales and marketshare between competitors. Product analysis is important to iden tify which features are responsible for commercial success - and equally which attributes can lead to poor sales . The other socio-political drivers in Figure 1 ar e the 161 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S environmental impact of the product and its manufac turer (either actual or perceived) and whether ther e are any ethical issues associated with the product or i ts manufacture. This kind of high-level informatio n can set the key characteristics of a product such as target performance and size, and also set certain lifecyc le attributes such as service intervals and overall length of use . Figure 1 also shows the three key downstream domain s within a company, the manufacturing, inspection a nd service functions. These can yield important lifec ycle information that could be used in the front en d design process, and deployed in new product design or incr emental product redesigns. In contrast to the exte rnal sources of design information, these internal sourc es can be used to refine and optimise the design ag ainst existing manufacturing and service capability. Esse ntially, the goal should be to close the design-to- manufacture-to-inspection loop, and to return as mu ch useful experience to the design domain as possib le. The quality of ‘what’ is fed back is important. Merely provid ing raw information back to the design domain is not sufficient - the knowledge must also be filtered and matched to the context in which it is being used. Ideally, existing tools should be useable, the meth odology should be complement tolerance analyses capabilities already deployed within the design fun ction [1]. Also important to what is fed back, is the question of ‘how’, ‘when’ and ‘how often’ – in othe r words the timeliness and frequency of the informa tion is crucial to its effective utilisation. 2.0 Methodology The subject of this research programme is the creat ion and implementation of a tolerance optimisation methodology for the design environment with particu lar use of the internal sources of design informati on. A Manufacture Inspection Service Sales – Existing/competitors products Socio-political drivers Design INTERNAL EXTERNAL Fig. 1. Internal and external knowledge sources 162 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S preliminary methodology is illustrated in Figure 2. Establishing a Design Knowledgebase is central to the methodology, and comprises two key elements: design tolerance knowledge and design context knowledge. The design tolerance knowledge is represented by a set of generic features created from relevant manufacturing processes such as injection moulding, sheetmetal forming and metal machining and fabrication. The knowledgebase also holds the vari ous design contexts applicable to the definition of a product or subassembly. These design contexts may hold design representations at different levels of abstraction, from a two-dimensional basic part or p roduct layout to a three-dimensional fully defined geometric part or constrained assembly. As Figure 2 shows, the Design Knowledgebase must support the u se of actual manufacturing tolerance data obtainable f rom various manufacturing sources whether from in-h ouse production or from the supply chain. In addition, the Design Knowledgebase must receive product assem bly, build or test data (recording the performance attri butes of a product). In addition to tolerance and performance data available from the original assemb ly of a new product, the equivalent data from remanufactured and serviced products should be made available to the knowledgebase. This gives the op tion to optimise part tolerance specification from new p roduct build and test data only; from remanufacture and service data only; or from an aggregate of new prod uct and service data. The inclusion of tolerance d ata from products with a service history is important for ne w product tolerance optimisation. Wear of critical components or features may affect product performan ce over time – hence dimensional control of additio nal features to offset worn componentry maybe appropria te – or recommending shorter service intervals to replace worn parts. Downstream manufacturing process data collection Design Context (1) Design Knowledgebase Design Context (2) Design Context (3) Design Context (n) Downstream service process data collection Design Tolerance Knowledge Design Context Knowledge Product performance data Fig. 2. The Research Methodology 163 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S The developed methodology must enable the design co ntexts stored in the knowledgebase to be used transparently be adaptable to the specific product development task. Each design context should provi de the user with a representation of the features or dimen sional tolerances relevant to a particular engineer ing or product scenario. The use of a particular context from the knowledgebase should act as a high level f iltering mechanism to offer a minimal dataset of retrieved i nformation. The context therefore creates a sophis ticated set of criteria to retrieve the required informatio n from the knowledgebase. This approach is central to the widespread adoption of the methodology, and should be viewed as an integrated step within the design process. If the methodology has to be used as a se parate ‘mini project’, this could limit its utilisa tion and the overall benefit in the process. ‘Usability’ is the refore a major requirement. The architecture of the Design Knowledgebase must support the chronological additi on of both design tolerance knowledge and design contexts, as new and existing products are develope d, tested and serviced. The Design Knowledgebase w ill evolve and mature over time as both the quality and quantity of the accumulated knowledge increases. The ability to understand the maturity of each knowledg e source would be useful as a confidence level chec k, indicating how much additional data are required an d the extent to which they have been filtered into a context-useable format. 3.0 Knowledge Infrastructure Support PART INSPECTION DATABASE PRODUCT ASSEMBLY DATABASE PRODUCT TEST DATABASE MASTER MANUFACTURING DATABASE SUPPLIER PART INSPECTION DATABASE SUPPLIER 1 SUPPLIER 2 SUPPLIER 3 DESIGN KNOWLEDGEBASE NPI SEAT 1 NPI SEAT 2 NPI SEAT 3 NPI SEAT 4 DESIGN DOMAIN KNOWLEDGE CONSUMER MANUFACTURING DOMAIN KNOWLEDGE PROVIDER PART REPLACE OR REBUILD DATABASE PRODUCT TEST DATABASE PRODUCT ASSEMBLY DATABASE SERVICE DOMAIN KNOWLEDGE PROVIDER MASTER SERVICE DATABASE Fig. 3. Scope of domain data collection 164 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S An information collection and propagation strategy is required to meet the needs of the domains and th e upstream manufacturing and design functions consume rs. The framework shown in Figure 3 shows how this could be enabled, and comprises design, manufacturi ng and service domains [2]. In the manufacturing domain, there are three databases: the single part inspection database, the product assembly database and the product evaluation (or test) database. Also shown within this domain is the database storing the dime nsional data from components manufactured by suppliers (e.g ., in a multi-tiered supply chain). There are als o three databases shown in the service domain: the part rep lacement or rebuild database, the product assembly database and product test database. The data is op tionally consolidated into a master service databas e, although this would be an operational refinement. The Design Knowledgebase is the recipient of the da ta available from the manufacturing and service domain s. The next step is to describe how this informati on is deployed with the design domain. An important princ iple in the use of downstream data is whether the information is ‘pushed’ by the manufacturing domain or if it is ‘pulled’ by the CAD application where the tolerance selection process is undertaken. Another consideration is whether the use of this data is i nfluenced by the context of the design scenario in which they are used and whether the use of the information is driven interactively by the designer, or is an ‘off-line’ data retrieval process. Using the design context t o control the nature of the query to the Design Knowledgebase is a way of filtering the mass of available informatio n. CAD GEOMETRIC MODEL (PART OR ASSEMBLY) KNOWLEDGE TEMPLATE (1) DESIGN TABLE (1) TO DRIVE PARAMETERS EXCEL DOC OPTIMISATION ENGINE EXCEL DOC INTERFACE ELEMENT INTERFACE ELEMENT INTERFACE ELEMENT INTERFACE ELEMENT CAD VAULT IN DESIGN KNOWLEDGEBASE Fig. 4. Use of knowledge templates KNOWLEDGE TEMPLATE (3) KNOWLEDGE TEMPLATE (2) (GEOMETRIC ASSOCIATIVE CAD DRAWING DESIGN TABLE (2) TO DRIVE PARAMETERS EXCEL DOC DESIGN TABLE (3) TO DRIVE PARAMETERS EXCEL DOC 165 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S 4.0 CAD Knowledge Templates The retrieved dataset can then be passed directly i nto the geometric model or drawing. To achieve thi s, each generic product design configuration is captured as a knowledge template in the CAD application. Each template comprises simple geometry, such as wirefra me elements or simple volumes to represent the sema ntic layout of design. Figure 4 illustrates how a knowl edge template is used as a basis for the geometric elements of a new design. The template and the new design a re linked using specific geometric elements which a re exposed or made accessible to enable the associatio n of geometric or linear constraints. These interf ace elements are used to enable the knowledge template to drive the features in the new parts and assembli es in the same system. The designer is able to link the new design into all the features in the knowledge t emplate or may opt to only choose specific features of interes t, and may decouple some of the associations if req uired. The use of the appropriate template establishes the parameters of any query to the knowledgebase. Thi s is achieved by using the design table interface which captures the required size and tolerance parameters , which in turn is linked to a query engine. This is a spr eadsheet environment from where queries are complie d to obtain information from the knowledgebase to facili tate analysis of process capability, experimental d esign and process test data, used by the designer to opti mise the design. The returned dataset information i s used to update the design table, which updates the knowledg e template, and hence the final design. The design table can contain a flag on the parameter that has been m odified using the knowledgebase data. The new geom etric design model is driven by the knowledge template vi a the interface elements, while a design table, hel d in an Excel spreadsheet, is used to drive the tolerance p arameters. 5.0 The Optimisation Engine The optimisation engine shown in Figure 4 is a key part of the methodology, and is shown in more detai l in Figure 5. Essentially the optimisation process is conducted within a configured excel workbook the mo st appropriate software tool to accommodate the proces s. The first worksheet within the workbook stores the tolerance chain imported from the design table docu ment - these initial dimensions and tolerance value s are stored as reference values. Worksheet 1 also stores the output (from the other worksheets). The secon d worksheet is a query environment, the main purpose of which is for the user to gain an understanding o f the impact of current process capability on the critica l clearances. The tolerance chain is copied into t his worksheet from the first worksheet. For each contr ibuting tolerance in the chain, the potential (CP) and process capability (CPk) indices [3] are calculated , and each tolerance is modified to reflect this pr ocess capability. The overall clearance, now taking into account process capability, is calculated and can be compared with the initial clearance values. The do tted arrow back from worksheet 2 to worksheet one i n indicates that these revised tolerances and dimensi ons may be captured and stored as a configuration, ready for use in the knowledge template if required. 166 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S In worksheet 3 allows the use of Monte Carlo analys is of the tolerance chain to obtain an output distr ibution [4]. This is in contrast to the analysis offered o n worksheet 2 which was limited to a single clearan ce value and a tolerance. It is proposed that the analysis could be performed in two ways – using an assumed distribution, where the population was statisticall y normal and 99% of the output was within tolerance , or using the actual sample distributions available fro m the Design Knowledgebase. The output distributio n would therefore more accurately represent a typical product run because realistic statistical distribu tions were used as inputs. The user would also enjoy a contr ibution analysis, displaying the impact of each tol erance in the chain - those tolerances with minimal impact co uld be relaxed to reduce cost, while the justificat ion for imposing tight tolerances on specific features can be made. As before the dotted line depicts the cap ture and storage of this tolerance chain back to the summary sheet. Worksheet 4 is where the optimisation proce ss occurs, and comprises a combination of correlation and goal seeking of the available manufacturing and performance data accumulated in the Design Knowledg ebase. Most products are factory performance teste d prior to shipping to ensure the product will perfor m as intended for the customer, and some of these performance metrics will be affected by certain key attributes in the assembly of the product. It is proposed to use the Spearman Rank correlation coefficient as me ans of identifying those critical assembly attribut es (and their numerical values) which correspond to the bes t performing products. For each critical assembly condition contributing to ‘good’ performance, the t olerance chain can be represented with the assembly (clearance) condition being the required ‘goal’. B y imposing constraints derived from the available manufacturing capability, the goal seeking algorith m in Excel can be used to allocate tolerance values to the remaining elements in the chain. Figure 5 shows th at once the tolerance chain has been optimised in t his way, the results are passed back to the worksheet 1, off ering the user the choice of updating the design ta ble parameters with the suggested tolerance values. Th is in turn updates the dimensions and tolerances in the knowledge template which (through the interface ele ments) will update the detail design parts and driv e any required drawing updates. 6.0 Case Study Fig. 5. Structure of the o p timisation e nvironme nt DESIGN TABLE INPUT WORKSHEET 1 SUMMARY SHEET (stores tolerance chains) WORKSHEET 2 Cp/CpK QUERY ENVIRONMENT WORKSHEET 3 OUTPUT DISTRIBUTIONS WORKSHEET 4 CORRELATION /GOAL SEEKING OPTIMISATION OPTIMISATION ENGINE (Excel Workbook) 167 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S A case study is currently being conducted to valida te the proposed methodology in a manufacturing comp any. Edwards Vacuum designs and manufactures vacuum pump s and exhaust abatement systems for the semiconductor and chemical industries. The company machines core pump components in-house, but suncontracts less critical sheetmetal and plastic a ncillary items. The data collection phase is currently underway, wi th the collection and synthesis of pump build data (critical internal clearances) and performance data (ultimate vacuum, peak pumping speed and peak power consumption). The identification of the existing d ownstream databases will be identified, followed by a recommended database implementation which will be s imulated to enable the capture of manufacturing dat a from both in-house and supply chain sources. The product knowledge templates will then be modell ed (as seem in Figure 4), in this case, a represent ation of the pump assembly. These drive derivates of the ge neric design (represented as geometric models), and are linked to the optimisation engine. Fig.6. Summary worksheet The optimisation engine (Figure 5) is a key element in the case study - elements of which have been developed and tested in isolation (such as process capability query) and now require integration withi n an Excel application. Figure 6 shows the first worksh eet within the optimisation engine which stores the tolerance chain imported from the design table docu ment. The initial dimensions and tolerance values are stored as reference values (Config0), while the mod ified tolerance chain, returned by the optimisation process is stored as a specific configuration (Config1,2,3, etc). The user of the system can select which of the stored configurations are to be uploaded to the design tab le, which as previously described, is used to drive the knowledge template within the CAD system. Ultimate ly, the component drawings which are associative to the geometric model can be driven by this optimisat ion process. 7.0 Conclusion • This paper has highlighted the importance of marsha lling engineering knowledge from the different domains within a manufacturing company pr incipally from manufacturing, inspection, production testing, service and the supply chain. It has shown how this information should be centralised in a Design Knowledgebase as a means of defining knowledge templates which can be used as a means of controlling the representatio n of new or incremental designs, and hence ultimately the manufacturing drawings most systems are required to output. The proposed FEATURE DESCRIPTION DIM TOL+ TOL- DIM TOL+ TOL- DIM TOL+ TOL- DIM TOL+ TOL- DIM TOL+ TOL- 5 0.5 0.5 5 0.7 0.7 5 0.7 0.7 3 1 1 3 1.2 1.2 3 1.2 1.2 2.5 0.2 0.2 2.5 0.2 0.2 2.5 0.2 0.2 8.5 0.5 0.5 8.5 0.5 0.5 8.4 0.8 0.8 TOTAL 19 2.2 2.2 19 2.6 2.6 18.9 2.9 2.9 CONFIG4CONFIG0 CONFIG1 CONFIG2 CONFIG3 168 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INDU STRIAL IMPLEMENTATION OF A METHODO LO GY FO R T HE OPTIMISATION OF PRO DU CT DESIGN TOLE RANCE S optimisation engine provides a means of optimising the tolerance chain using existing design and manufacturing knowledge. It provides several a nalyses options for the user to enrich the design with existing pertinent tolerance knowledge. References [1] R.Eadie, J.Gao ‘A Communications Infrastructure to Facilitate Tolerance Optimisation in the Knowle dge-Based Design Environment’, Proceedings of the 6 th International Conference on Integrated Design and Manufacturing in Mechanical Engineering (IDMME 2006), 2006 [2] R.Eadie, J.Gao ‘Computer-Aided Tolerance Analys es Tool Selection Criteria’, Proceedings of the 5 th International Conference on Manufacturing Research (ICMR2007), 2007 [3] P. Drake ‘Dimensioning and Tolerancing Handbook ’, McGraw-Hill, ISBN 0070181314, 1999. [4] C. Creveling ‘Tolerance Design - A Handbook for Developing Optimal Specifications’, Addison Wesley Longman Inc, ISBN 0201634732 , 1997. 169 170 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE INDUSTRIAL ROBOTICS IN THE LEAN ENTERPRISE – A CASE STUDY Mikael Hedelind, Mats Jackson School of Innovation, Design and Engineering, Mälar dalen University Abstract The globalization and the increasing challenge from low-wage com petitors highlight the need for European industries to enhance their ability to develop and manufacture products competitively. Meeting customer demands requir es a high degree of flexibility, low-cost/low-volume manufacturing skills and an abi lity to offer short delivery times. In order to stay competitive, many manufactur ing industries are trying to implement the unique management principles and practi ces of the Toyota Motor Corporation’s with many different names as e.g. “The Toyota Product ion System” or “Lean production”. One question and debate within industry, d uring the transformation towards lean manufacturing is whether tradition al robot automation fits the principles and practices of lean? This paper pre sents a case study which has investigated if industrial robot automation has a place in a manufacturing company pursuing the lean philosophy. The case study is based on one manufacturing company in Sweden that is currently implementing a transformation t owards a lean-based production system. The case study was performed using intervi ews at the company, observation at the manufacturing plant, and workshops together w ith key-employees at the company. The results from the case study show that t here is a need to align the company’s present robotic equipment and machinery towards le an principles. The lean transformation within the company is based on increased availability, controlled buffers, a more open layout, and flow-based manufacturing with reduced batch sizes which all effect the equipment and machinery. In order for the robot automation to fit lean principles and practices there is a need for developm ent of robotized working cells with increased availability, reduced set-up times b y improving the ability for easily reconfiguration, and improved information design to clearl y present visual information and options to the operators. Keywords: Lean Production, Industrial Robotics, Cas e Study. 171 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE 1.0 Introduction An increasingly global market and a business enviro nment that is becoming harder are driving companies to become more competitive. The needs to reduce waste and create production systems that are streamlined towards producing products at lowest possible cost push companies into changing their production set u p. Many manufacturing companies are transforming their production systems towards the lean production philosophies. Case studies shows that companies tha t implement what is called lean manufacturing princ iples or just-in-time (JIT) production can reach competit ive advantage over those that does not [1]. However , these transformations are more or less successful dependi ng on how much the internal structure and culture o f the company is changed [2]. Many western companies have realized that just trying to imitate the Toyota Production System will not give them the competitive advantages they are looking for. An increasing amount of companies are realizing that they need more than just a lean transformation of their production sys tems; they need to implement lean organizations ; they need to become lean enterprises [3]. This means that the whole company will have to change, both in working methods and in a business cultural sense. The term lean means using less human effort in the factory with less manufacturing space, less investm ents in tools, less engineering hours to develop a new prod uct in shorter time, keeping less inventory, fewer defects in production, and production of a greater and ever gr owing variety of products [3]. Lean practice has pr imarily two objectives: “ eliminate waste” and “ create value for end-use customers”. Companies that adopt to lean manufacturing principles or JIT production have bee n shown to gain competitive advantage over those th at does not [1]. Studies show the more comprehensive t he adoption of JIT is at a company, the greater are the overall returns [4]. The lean production philosophies introduce extra de mands on the production systems and the workstation s that are parts of the systems. Many companies autom ated as many workstations as possible in order to r educe the manual labour cost in each product. This have h istorically, in some cases, ended up with what is c alled monumental automation; meaning that companies have automated too much and their production systems became rigid and vulnerable to changes. Today, many companies are doing the opposite and are removing the automation in order to become as flexible and robus t as possible. When conducting interviews at compan ies one can receive comments like: “ Automation and industrial robotics creates complexi ty” or “ Robotics and lean does not fit together; they rather contradict each other”. Further, many companies in Sweden regards automation, and especially industrial robotics, as unfit it lean production systems. The general belie f is that working stations including industrial robots become s too rigid and thus creates production systems tha t are inflexible and cannot adapt to changes. In many cas es the movement towards removing automation is motivated with a reference to Toyota as a company t hat does not use advanced manufacturing technologie s. This, however, is often a mistake since Toyota is a technology-based company that are among the most sophisticated users of advanced manufacturing techn ologies in the world [5]. This paper presents a case study performed at a lar ge manufacturing company in Sweden. The company is undergoing a large “lean” transformation and the qu estion whether industrial robotics is a desirable t ool in the workshop was raised. The case study aimed at mappin g the current usage of industrial robotics in the f actory, and find out why the current robot installations di d not perform as well as they should in order to fi t into the new, more productive, production system. 172 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE 2.0 Method The case study is a preferred research method when the researcher is looking for answers to “Why?” and “How?” questions [6]. The case study may also conta in a wide spectrum of rich data, in the form of documents, interview results, notes from observatio ns, etc. The case study performed in this research had two main purposes: (1) map the current state of industr ial robots at the case company, and (2) propose how the company should proceed with using industrial roboti cs in their new lean production system. These goals of the study could then be divided into a set of quest ions to answer: Q 1 – How was industrial robot automation used at the company before the lean transformation? Q 2 – Why is there a reluctance to use industrial robo tics at the company? Q 3 – How can the robotic working cells used at the co mpany be changed/improved in order to make industrial robot automation more accepted at t he company? A literature study was performed in order to set th e theoretical foundation of the research. During th is literature study online searchable databases was us ed to find relevant scientific papers, as well as b ooks. The case study contained four different phases, sho wn in Table I: (p 1 ) pilot study carried out over one month containing interviews and observations at the facto ry, (p 2 ) workshops/brainstorming, (p 3 ) intensive study at the company, and (p 4 ) workshop/brainstorming with project team and key- employees at the company. The project team from the university consisted of: (1) professor in innovative production, (2) PhD-student within the area of applied industrial robotics, (3) PhD-st udent within the area of strategic manufacturing maintenance, and (4) research assistant within the area of applied industrial robotics. Table I: Summary of the different phases in the cas e study. Phase Purpose Participants P 1 - Pilot study Initial background study, in-depth study of one cell Project team; interviewees at the company P 2 - Workshop 1 Building a vision and overall goal for the project Project team Workshop 2 Building a vision and overall goa l of the project Key-employees at the company: project manager for transformation process, purchasers/managers of production equipment, production manager; project team as moderator P 3 – In-depth study In-depth study of how industrial robotics is used at the company Project team, primarily two master thesis students; interviewees at the company P 4 – Workshop and follow-up meetings Deciding suitable concept solutions for how to better utilize industrial robotics at the company Project team; key-employees at the company: project managers for purchasing projects, employee responsible for production development. 173 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE The five phases described in Table I was performed in sequence and built on the results from previous phases. The pilot study (P 1 ) was intended to investigate the need for a resear ch project that was to focus on how to integrate industrial robotics in a lean production system. The pilot study resulted in a set of sugges ted improvements for the case company, as well as a pro posal for a research project in collaboration betwe en the case company and Mälardalen University. The objective of the first workshops (P 2 ) was to get a common understanding in the project group of what it means to have industrial robotics in a lean product ion system. A set of questions had been prepared beforehand and the group used brainstorming on a wh iteboard in order to answer the questions in sequen ce. The questions were: (1) what is the overall goal of the lean transformation at the company?, (2) What does it mean to become lean?, (3) How will the company beco me lean?, and (4) Which requirements does lean pose on the equipment used in the production system (wit h focus on industrial robots)? The results were in the form of bullet lists where some of the bullets were cons idered more important. Two workshops was carried ou t, first one with the project team from the university , and then one with key-employees from the company. The results from the two workshops were later compared to see if there were any differences in the results . The third phase (P 3 ) was performed by two master thesis students who’s main objective was to make a more in-depth study of the company’s production system, and how industrial robotics was applied in that sys tem. This phase had a time-span of almost six month, sta rting in the spring of 2007 and ending during the f all. The result of this phase was a masters’ thesis containi ng a theoretical framework, a mapping of the curren t state of production, and a proposal of checklists that can b e used by the company as guidelines for investment projects. The final phase (P 4 ) was a workshop and brainstorming session where th e project team and key-employees from the company worked together to identify which areas that was of greatest importance to the compan y. The background material used in this workshop inclu ded: the literature collected in the earlier phases , the results of earlier phases, the results of a measure ment of stops in the robotic working cells performe d by the operators, and the results that a lean consultant a gency had provided after an extensive investigation performed during 2006. The case company is a large manufacturing company s upplying drivelines to automotive industry. The company has several production plants world-wide an d this study was performed at one plant in Sweden. The case study was delimited to a certain segment of th e manufacturing system where industrial robots were used for material handling and tending CNC-machines. The product produced at this production segment was ge ars used in gearboxes. The company is currently underta king a large lean transformation where a whole production facility is being re-arranged and adapte d to lean production philosophies. The company is p art of a large group that have developed its own clone of TP S, which is near to identical to TPS. The company initiated the transformation in order to meet with a higher demand of their products, which was not su pported by its current production facilities. 3.0 Automation and the Lean Enterprise Womack et al. [3] identified in their MIT study tha t many automotive companies had a lot of automation , but the Japanese companies that had the highest product ivity also had the lowest level of automation. Howe ver, they also believed that the automotive industries w ould automate almost every operation in the factori es in the future [3, p. 102]: 174 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE “by the end of the century we expect that team-asse mbly plants will be populated almost entirely by highly skilled problem-solvers whose task will be to think continually of ways and means to make the system run more smoothly and productively.” This prediction that Womack et al. did has not real ly come true but it is still possible, and even lik ely, that it will happen in the future. Traditionally within the lean enterprise, e.g. Toyo ta, low-cost automation has been utilized [7], whil e companies that are not traditionally lean have used the more high-technology automation [8]. This type of low-cost automation is implemented to take the need s of the whole system into account, and thus become s subordinated to the rest of the system. This kind o f automation is considered to be in line with lean production philosophies [8]. However, there are several motiva tions for using more high-technology solutions such as industrial robotics, the flexibility and re-program mability of a robotic arm, for example. The industr ial robot is a very versatile machine that can be used in many d ifferent applications. One of the driving forces for using automation with in industry is reduction of cost. Within the automo tive industry there is a noticeable difference between c ompanies that reside in countries where there is a high labour cost and those in countries with a lower lab our cost. In the Daimler-Benz Mercedes A-Class seri es factory in Germany the body shop is almost 100 per cent automated, whilst Skoda’s Octavia body shop in Czech Republic, where labour costs are low, is only 15 per cent automated [9]. In the Octavia body sho p “ the strategy is to use a manual solution wherever possible, unless quality or productivity factors dictate otherwise” [9, p. 136]. The term lean automation has been defined in different situations. Some pha rmaceutical industries been looking to make their production more efficient thr ough the use of automation, and have in this contex t defined lean automation as [10, p. 26]: “Lean automation is a technique which applies the r ight amount of automation to a given task. It stresses robust, reliable components and minimises overly complicated solutions.” One of the pillars of the Toyota Production System is called Jidoka, which comprises autonomation, als o known as “automation with a human touch” [11]. Auto nomation means “[…] making equipment or processes that are ‘smart’ enough to detect an undesired, abn ormal state and stop so as not to produce a defective product […]” 1 The concept of autonomation was developed since th e Toyota Corporation saw a problem in that normal automation do not have any built in che cking for quality problems. This may lead to hundre ds of defect parts produced if automated production equip ment is producing without human supervision. “ At Toyota, a machine automated with a human touch is o ne that is attached to an automatic stopping device” [11, p. 6]. This stopping device described in [11] is not only a stopping device that is used to stop that particular cell or workstation, but it is a stoppin g device for the whole production line, i.e. a line stopping device. This means that autonomation is an importa nt part of the Visual Control system, or Management by Sight, where it is important that the current state of p roduction is always visible and any problems are br ought to attention as soon as they occur [12]. 1 Source: Babylon online dictionary, http://www.baby lon.com/definition/jidouka/Japanese, visited: 2008- 03- 31 175 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE As a result of this, the equipment may produce unat tended and without human worker presence until ther e is a defect or malfunction. The original meaning of jid ouka was “Automation” as shown in Fig 1 (a). The sentence was later changed at Toyota into the spell ing shown in Fig 1 (b), the pronunciation of jidouk a was the same but they added two extra lines, spelling h uman. This was an important statement, meaning that the automation (or autonomation) should be working the same way as a human; it should be intelligent2 . The three words in Fig 1 (a) spell out: “Self moving transfor mation”, while the extra two lines in Fig 1 (b) add s the “human touch ” . (a) (b) Fig 1. the Japanese word Jidouka; for (a) Automatio n and (b) Autonomation. 4.0 Case Study Results The case study has been carried out over 16 month i n total, with different activities as described in Section 2. The overall objective of the case study has been to investigate whether the industrial robot has a pla ce in a lean manufacturing system, and if so, how it should be used. The study was initiated as the company it self felt rather unsatisfied with the industrial robot automa tion that was used at the company at the time. The pilot study (p 1 ) aimed at identifying why the industrial robot aut omation used at the company did not work in a satisfying manner. The objective subsequent worksho ps (p 2 ) was to get a consensus within the company, and between the company and research group, of what lea n production means at the company and what demands lean philosophies puts on the production equipment used in a lean manufacturing plant. The in-depth st udy (p 3 ) aimed at doing a more thorough study of how indus trial robotics is applied at the company, and propo ses a method for how industrial automation should be in tegrated in the production system. The final phase in this study (p 4 ) aimed at identifying needs for how to make indust rial robot automation more applicable at the case company. The study was designed to answer the three question s posed in Section 2. The following part of this se ction will present the answers to the questions, based on the evidence collected at the case company. Q 1 - H ow was industrial robot automation used at the com pany before the lean transformation? The study focused on a part of the company’s produc tion system where the machining of gears and axes w as performed. In this part of the production system in dustrial robots was used to tend CNC-machines. The usual robotic working cell contained a robotic working ce ll and three of four machines. The robot picked mat erial from an EU-pallet using vision or laser guidance an d moved the material between the machines before fi nally placing it in another pallet. Some cells where conn ected using conveyors or shuttles where parts where transferred between the cells. 2 Source: Seminar with Hiroyuki Mikami, a TPS-expert originating from Toyota Corporation, 2007-04-27 176 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE One notable issue for the company is that several d ifferent external system integrators have been hire d to do different installations. The different systems inte grators have been provided with a scope of supply and a technical specification, but those documents have been rather loosely defi ned and left a lot of decisions open for the consultant. Thus, different system integrat ors have made different types of solutions, which m ean that each installation is different from the others. Thi s leads to that every installation at the company h as different solutions for programming, user interfaces, and oth er technical solutions. The company have also had a lot of consultants work ing for them when it comes to daily operation of th e robotic working cells. The competence of the operat ors when it comes to robotics and programming is qu ite low and thus a lot of external expertise is needed when introducing new products in the cells, for exa mple. Q 2 - W hy is there a reluctance to use industrial robotic s at the company? There are different issues that have been identifie d as reasons for the reluctance to use industrial r obotics. One of the major reasons is that the production managem ent argues that the robotic working cells have to l ow availability. There are several reasons for this; o ne is that the robotic working cells have a high nu mber of operations incorporated into the same cell. If ther e are 5 machines in the cell, each having an availa bility of 98%, the cells overall availability will be (0.98 ^ 4) about 92.2%. This availability will suffer even more if each stop in the cell gets longer because of low compete nce of the operators, leading to situations where e xperts (internal or external) have to be requested to fix any problems. The operators working on the factory floor (blue co llars) is reluctant to use industrial robotics base d on that they do not feel comfortable working with technolog ies they do not fully understand. The different sol utions that are implemented in each installation make the inexperienced operator unable to efficiently and ef fectively interact with each cell. Another problem was identified when a review of rep orts on production stops was conducted. The follow- up protocols used in the robotic working cells were no t providing any real information on which types of problems that were experienced. When reviewing the reports for one cell, the most commonly reason for stoppage in a machine was reported as “ No Category ” with 45.0%. In another machine in that cell the m ost commonly reported reason for stops was “ Short stop”, with 67.3%. This type of follow-up of production is of course useless since no information about the reaso n for the stop is provided. A manual measurement of the stops showed that changeover was the most common re ason for stoppage in the production process. The mo st common breakdowns reported was short stops where th e robot was unable to pick up components, or droppe d components. Even though several issues with industrial robotics were reported in the study, the overall opinion of the operators was that the roboticized workstations wor ked rather ok. Interviews with the operators reveal ed that they where satisfied with the overall performance o f the cells, even tough it was quite a lot of “quic k fixes” and “fire distinguishing” reported. The overall opi nion of the management was a bit different. They fe lt that the robotic working cells where not performing as w ell as they should, and that they brought on unnece ssary costs. Q 3 - H ow can the robotic working cells used at the com pa ny be changed /im proved in order to make industrial robot automation more accepted at the co m pany? 177 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE The objective of the last phase of the study was to identify different means for making industrial rob otics more attractive and useful for the company. The res ults of P 4 were divided into four categories, with the overall goal at making the industrial robotic worki ng cells more easy to use: (1) software used in the cells, (2) maintenance of the cells, (3) manuals for how to pu rchase, integrate, and manage the cells, and (4) visualization of the cells. The different categorie s in turn contain several issues to investigate/dev elop. Table II shows some examples of topics that fall into eac h category. Table II: Summary of the content of the different c ategories for improvement. Software support Maintenance Manuals/checklists Visualization - User interfaces - Cell programming - Interfaces between cell and surrounding IT-structure - Jidoka - Operator maintenance - Preventive maintenance - Measurements in the cell - SMED - Manual for operator - Manual for purchasers - Manual for integrators - Technical specification - Scope of supply - Andon charts - Visualization of production flow - Visualisation of current state of production 5.0 Conclusions This case study aimed at investigating whether trad itional industrial robotics is applicable in a lean production system. There is, based on the case study, no reaso n to say that industrial robotics is not applicable . However, as companies strive to become leaner and eliminate waste, complex and complicated produ ction equipment often gives disturbances due to rigid solutions. Co ntinuous flow and reduced inventory highlight ineff iciencies and poses some new demands on the equipment used in the production cells. Four main areas where the robotic working cells at the company can be improve d have been identified. As can be seen from the case study, the operators h ave little confidence in their ability to implement changes in the robotic working cells. Further, the managers feel discomfort in the fact that the company has t o rely on outside experts in order to handle day-to-day activ ities, such as introducing new products or fixing s mall problems. Those issues are all incentives for the d evelopment of robotic working cells that are more e asy to use than today’s installations. Two of the proposed categories of development and improvements are foc using on this; the software support category and the visu alization category. Those two categories are also v ery closely connected since the software support may be seen as an enabler for visualization. Traditionally, research in the area of automation h as had a long tradition on focusing on the user of automation technologies, trough concepts such as sy stem/situation awareness [13], users feeling “out-o f-the- loop” [14], and other consequences of different lev els of automation [15]. This type of research has b een focusing on large automation systems which may have fatal consequences in case of a breakdown; aviatio n, power plants, and air-traffic control for example. However, when the availability of production equipm ent becomes increasingly important, those research area s become more and more important within manufacturi ng technologies as well. The automation technologies used in world class man ufacturing plants have to comply with lean principl es such as andon, jidoka, and visual control. This pos e new demands on the production equipment, but also adds a lot of value since it brings clarity and transpar ency. If the system designers focus on ease-of-use and provide solutions for system- and situation awareness for t he operators, the usage of industrial robotics will have potential to become more of a commodity in all type s of companies and manufacturing systems. 178 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 CASE STUDY ON INDU STRIAL ROB OTICS IN THE LEAN ENTER PRISE This case study aimed at identifying the needs for further development of industrial robotic working c ells that are to be used in lean production systems. The prop osed four categories will be further investigated a nd developed in the Lean Robotics project as part of R obotdalen. One of the more hands-on developments th at are planned is a Cell-PC where the theories of Jido ka and Visual Control will be implemented. References [1] White, R., Pearson, N., Wilson, J., (1999) JIT Manufacturing: A survey of Implementations in Small and Large U.S. Manufacturers. Management Science, Vol. 45, No . 1, pp. 1-15 [2] Emiliani, M.L., Stec, D.J., (2005), Leaders Los t in Transformation, Leadership & Organization Deve lopment Journal. Vol. 26, No. 5, pp. 370-387 [3] Womack, J., Jones, D., Roos, D., (1990), The Ma chine That Changed The World. Rawsson Associates, N ew York [4] Fullerton, R. and McWatters, C., (2001) The pro duction performance benefits from JIT implementatio n. Journal of Operations Management, Vol. 19, pp. 81-96 [5] Liker, J., and Meier, D., (2006), The Toyota Wa y, Field book. The McGraw-Hill Companies, Inc., New York, USA [6] Yin, K., Robert, (1994), Case study research – Design and Methods. 2nd edition, ISBN: 0-8039-5663- 0 [7] McCarthy, D., Rich, N., (2004) Lean TPM – A Blu eprint for change. Elsevier Ltd. [8] Muffatto, M., (1999), Evolution of Production P aradigms: the Toyota and Volvo Cases. Integrated Manufacturing Systems, Vol. 10, Nr. 1, pp. 15-25 [9] Kochan, A., (1998), Automotive industry looks f or lean production. Assembly Automation, Vol. 18, N o. 2, pp. 132-137 [10] Dulchlnos, J., Massaro, P., (2005), The time i s right for labs to embrace the principles of indus trial automation. Drug World Discovery, Winter-issue, 2005-2006, pp. 25-28 [11] Ohno, T., (1988), Toyota Production System – B eyond Large-Scale Production. Productivity Press [12] Liker, J., (2004) The Toyota Way. McGraw-Hill, USA [13] Endsley, M., (2001) Designing for situation aw areness in complex systems. Proceedings of the seco nd international workshop on symbiosis of humans, arti facts and environment, Kyoto, Japan [14] Di Nocera, F., Lorenz, B., and Parasuraman, R. , (2005) Consequences of shifting from one level of automation to another: main effects and their stability. Human Factors in Design, Safety, and management, pp. 363 -376, D. De Waard, K.A. Brookhuis, R. Van Egmond, and Th. Bo ersema (Eds.) [15] Parasuraman, R. and Riley, V., (1997) Humans a nd Automation: Use, Misuse, Disuse, Abuse. Human Fa ctors, Vol. 39, No. 2, pp. 230-253 179 180 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY AN ASSEMBLY LINE INFORMATION SYSTEM STUDY Keith Case 1,2 , Gunnar Bäckstrand 1,2,3 , Dan Högberg 2 , Peter Thorvald 1,2 , Leo J De Vin 2 1. Mechanical and Manufacturing Engineering, Loughbor ough University, UK 2. The School of Technology and Society, University o f Skövde, Box 408 , SE- 5 4 1 28 , Skövde, Sweden 3. Volvo Powertrain AB, SE- 5 4 1 87 , Skövde, Sweden Abstract Assembly line information systems are designed to provide asse mbly workers with appropriate information that allows the assembly of the product in good time and good quality. In this context product quality might be defined relative to the number of internal rejects or products which need some kind of r eworking before being in a deliverable condition. This paper describes a pilot study of a heavy diesel engine as sembly line where considerable variety is presented to the assembly workers in the form of engines destined for trucks, buses, marine applications and stationary power generation each of which has to comply with a variety of national and international standards. Internal rejects might for example occur through the fittin g of sub-assemblies that are unsuited to the eventual application, and although an extensive i nformation system is currently in place the level of internal rejects is consi dered to be unsatisfactory. The objectives of the study were to understand how the assembly workers interact with information systems and the impact this has on product qualit y and productivity. A single line was studied for ten days during which 2600 engines were assembled. At four of the assembly stations the existing information system w as changed to reduce the amount of information to be assimilated by the workers, the timing of its presentation and its location. The use of simple colour-coded cards and symbols resulted in the reduction of internal rejects by 40% on tw o of the assembly stations and to zero on the other two stations. It is believed that chan ging the information system has changed the workers’ behaviour through a reduction i n cognitive stress levels. The pilot study has provided useful insights into th e basis for modifying information systems and a further study of the final assembly of heavy trucks is planned with an ultimate aim of determining a rationale for th e design of information systems for use within the assembly of customised products. 181 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY Keywords: Production, Information usability, Cognit ive ergonomics, Workplace design, Assembly quality. 1.0 Introduction Volvo Powertrain in Skövde, Sweden manufactures hea vy diesel engines destined for trucks, buses, marin e applications and stationary power generation each o f which has to comply with a variety of national an d international standards . Engines are assembled using mixed-model production where a high volume product is assembled in the main flow but low volume products are also present. To handle this situation a dynami c information system is essential so that it can (for example) refer to parts to be assembled on a speci fic engine, information regarding how to assemble a particular engine, etc. In the plant studied the information s ystem is implemented as an IT system mounted on the automati c guided vehicles (AGVs) that carry the engines dow n the assembly line. Earlier work [1] explored the re lationships between attention, interpretation, deci sion- making and acting and how this relates to the infor mation flow based on the idea [2] that the degree t o which Active Information Seeking behaviour is supported/t riggered has a large influence on the number of int ernal rejects. As an extreme example, if a trigger is not present or detected, then an active information se eking behaviour will not be present. In other words, the trigger is predecessor to active attention, and act ive attention is believed to be a state of mind that is crucial f or successful use of data and information. If it is not possible to create attention, then it is not possible to start an interpretation, decision-making and acting proce ss. Obviously, there will still be some kind of interpr etation, decision-making and acting, and the assemb ly personnel will continue to assemble engines, but th e risk of assembly errors increases if one fails to trigger the personnels’ attention [3] at the right time and to the correct data source. “Structured translation of data into action” [4] to reach a specific goal must be the ma in focus for the assembly personnel. This requires that the information is available at the right time in the r ight place and that the assembly personnel have ide ntified a need for the specific information. In the assembly environment there is evidence, internal rejects, th at the personnel do not use the information system in the most effective way. Studies made on the shop floor have concluded that the support system, from a graphical point-of-view is well, but not perfectly designed, but the users do not use them in a way that was anticipated . Input from ongoing projects indicates that one th e main reasons for this is the attention levels among the assembly personnel. 1.1 The Role of Information Triggers The purpose of any rational action should be to ach ieve a goal [5, 6]. This should normally create a d emand for information, which is triggered in some way. There are four situations regarding information need versus demand: (1) There is a need but no demand. An error will eventually occur, and a solution might be to introduce triggers to create the demand. (ii) There is a need and a demand. This situation is the pref erred one with low risk of errors due to lack of information. It is still essential that information matches the need. (iii) There is a demand but no need. This situation can b e frustrating for the personnel as they have identi fied a need and have a demand, but they are not provided w ith the information they feel they need. (iv) There is no need and no demand. 182 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY 1.2 Models Relating Action to Information The OODA loop ( Observation, Orientation, Decision, Action) [7 ] is an example of a model relating the information process to actions. However, some resea rchers [8] have suggested that “Observe, Orient, De cide and Act” should be redefined to “Observe, Interpret , Decide and Act”. This would create a more generic model more suitable for manufacturing purposes, and in this case make it possible to map “Attention, Interpreting, Decision making and Acting” to the OO DA model in a better way. 1.3 Attention Before it is possible to receive information, one m ust perceive its presence. Aspects of passive and a ctive attention are believed to be major contributors to problems in the work environment [2, 9-11]. Active attention is connected to actively processing information. e. g. during skilled machining operations one actively seeks and processes information. In passive attention, su ch as a visual inspection task we may register data/information in our surroundings without proces sing it actively until something unusual happens. T he assembly personnel studied seem to be aware of some data, and perform parts of tasks correctly, but at the same time missed other important issues connected t o the same task, although all data/information is a vailable in the context of the work station. In other words, attention plays an important role in how data is o bserved. 1.4 Interpreting In this case “interpreting” refers to the process o f transforming [12] data into information. One way to describe the difference between data and information is that data is ”a set of symbols in which the individual symbols have a potential for meaning but may not be meaning ful to a given recipient” or ”a set of symbols in w hich the individual symbols are known, but the combination i s meaningless” or “understandable symbols rejected by the recipient as being of no interest or value” [12 ]. Information however can be described as “a mess age that exists but that is not necessarily sent to, or rece ived by, a given recipient” or “a message sent to a destination or received by a destination, but not evaluated or understood” or “a message understood by the recipie nt and which changes that person's knowledge base” or “an output of the process of converting received messag es, data, signs, or signals into knowledge.” [12]. This leads to a conclusion that there is a connection b etween interpretation, data and information. The definitio n of information [12] states that it is an output o f a process. This output among other things, such as knowledge, is a base for decision-making and acting. From this it is possible to state that information in the assembly plant context could be a message that when received, read, interpreted by a recipient creates knowledge/ changes the receiver’s knowledge base so that an action can be committed by the receiver that is predicted by the sender. 1.5 Decision-making and Acting Although the subject of linking acting to decision making is an interesting topic, within the context of this paper these two topics can be discussed as one. Dec ision-making can be seen from a process perspective as a “many-to-one mapping of information to responses” [ 13] . If one considers this visualization together with the 183 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY “mapping” statement it is easy to see that there co uld be different data/information sources that prov ide assembly personnel with data/information. This of c ourse can directly affect the decision-making proce ss in a negative way. 1.6 The Importance of Triggers A trigger is the signal that creates a change of st ate regarding attention, from passive to active, an d preferably, from active to passive. The hypothesis is that on t he shop floor much of the attention is focused on t he assembly task, and not on gathering data/informatio n. This selective attention [14] is a part of the h uman nature, so an information system should provide a p ossibility to focus the assembler’s attention to th e right place (there is evidence that at the specific plant , the personnel use data/information from non-relia ble sources) and at the right time as the trigger and the data m ust be synchronised.. The information system should support a change of state between active and passive. As “i t is the nature of the attention process to generat e its own extinction” [15], a trigger must also contribute to the “resurrection” of the information demand. It i s important that a trigger is used in a way that really support s the personnel. There is evidence at plant that mi suse of quality support systems have created a feeling of i rritation leading to misuse. 2.0 Pilot Study The aim of the pilot study is to find evidence that supports the overall aim of the research which is to find or develop a prototype work method that supports the d esign of information flow based on product and proc ess demands. Two hypotheses have been formulated: Hypot hesis 1 - ‘Information Seeking Behaviour’ - The degree to which Active Information Seeking behaviour is supported/triggered has a large influence on the number of internal rejects. Hypothesis 2 – ‘use of evaluation methods’ The use of an evaluation work method in the conceptual as well as in the design phase of an information flow will affect the internal rejec ts in a positive way. The pilot study was aimed at gaining practical kno wledge regarding how triggers, active attention and passive attention affect the internal reject rate, but it also gave knowledge regarding how the assembly personnel interact with the information fl ow present in their work context. It is hoped that the knowledge gained from the pilot will be useful in t he next stage of the research which is the creation of a prototype work method. 2.1 Performance Indicator To evaluate the value of the knowledge obtained it is important to identify a performance indicator. I n this case the main performance indicator is the rate of internal rejects for which there is historic data. The historic data covers a period of about 9 months in 2006 and includes approximately 33000 records. This data inc ludes: date and time which can be used to trace the error source, engine family and variant (the data include s data connected to 6, 7, 12, 13 and 16 litre engines). Th e engine families have different variants (known as “engine types”) and approximately 500 variants exist), effe ct number (a code used for reject causes), part num ber and free text field (used to describe the cause of reje ction - unfortunately not used reliably). The rejec t historic data process starts with a discovery of a divergence fro m the order specification. This is done at the end of the line 184 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY and in the test zones. If a divergence is discovere d it is saved in a database and problem is solved i n a special reject area. The most common way of subsequently re trieving data from the database is via a Business O bjects (BO) question. This data has to be prepared for vie wing so that the reject information can be used as the basis of decision-making. In this case the data can be so rted to identify the engine family, the date and ti me (to establish a weekly reject rate) and the assembly st ation. 2.2 Experimental Assembly Environment The actual production environment under normal oper ating conditions was used in the pilot study and wa s modified by changing the trigger from the screen-ba sed alphanumeric approach. The new triggers were coloured magnetic rubber sheets attached to the Aut omatic Guided Vehicle (AGV) which is the carrier of the engine (figure 1). The sheets have five different c olours - blue, pink, orange, green and black to act as triggers for the different situations. Fig. 1. The AGV/carrier and its cargo, a diesel eng ine. 2.3 Quantitative Data The results were evaluated by comparing the perform ance indicator before and after the pilot study. As mentioned earlier, this indicator is a part of the reject handling system and had previously been meas ured for 185 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY approximately 9 months during 2006. The data gather ed during the pilot study shows that the performanc e indicator for the period is at a mean level of 1.46 % (figure 2). However, depending on the sorting of the reject data two more values with higher means of 2.77% and 3.12% must be considered. The performance indicato r with value 1.46% (H α0 ) includes only rejects with the effect numbers bei ng studied (parts missing or wrong parts assembled). The performance indicator with va lue 2.77% (H β0 ) includes H α0 and rejects that have a non- assembly related attribute. The performance indicat or with value 3.12% (H γ0 ) includes rejects from H α0 and H β0 as well as rejects connected to a series of categor ies that indicate leakage of oil and water. Parts m issing or wrong parts assembled can, if not found earlier, le ad to a leakage in the final test zone. If this hap pens the reject registration will not be in the category “pa rts missing” or “wrong parts assembled” and it will be “leakage” or “deviation from specification”. Figure 3 shows the error distribution over the stations w ithin the line during the nine month period before the pilot study. The workstations affected were stations 800 and 1100, and as there are parallel lines this makes a total of four workstations. 0,0 0 % 0,5 0 % 1,0 0 % 1,5 0 % 2,0 0 % 2,5 0 % 3,0 0 % 3,5 0 % 1 3 1 7 2 2 2 6 3 0 3 5 3 9 4 3 4 8 5 1 Reject in percent based on production volume Mean Value Mean incl. sort out rejects Incl. 30 0,4 0 0,4 2 0,4 4 0,4 5 0 Fig. 2. Performance indicator sorted per week. 0 50 100 150 200 250 300 S0100 S01 5 0 S0200 S0 2 50 S0300 S0 4 00 S05 0 0 S0 6 00 S07 0 0 S0800 S08 50 S0900 S0950 S1 0 00 S1040 S1 1 00 S12 0 0 S1 3 00 S14 0 0 S1500 S18 0 0 S1900 S2 0 00 S2100 Hα 0 Hβ0 Hγ 0 Fig. 3. Performance indicators sorted per assembly station. 186 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY 2. 4 Assembly Process AS-IS The line studied contains twenty-four assembly stat ions and is balanced so that every station has approximately eight minutes of assembly time, i.e. the assembly personnel have eight minutes to accomp lish the assembly task. The work process starts typicall y with an AGV/carrier arriving at an assembly stati on. The onboard information system would typically be displ aying up to 19 lines of text to describe the assemb ly requirements. The subsequent steps in the assembly process are (1) identification of part to assemble. (not necessary where parts are identical on all engines) , (2) retrieval of part to assemble (3) assembly of part onto engine. Steps 1 to 3 are repeated until all parts a re assembled. (4) Confirmation of task - the person nel confirm via an IT system that they have completed all assig ned tasks at the station. (5) Carrier departs from the assembly station to the next station. Step 1, “Iden tification”, involves gathering of data/information . The data/information is presented to the personnel via a computer screen and is visible when the AGV/carri er arrives at that assembly station. 2.5 Assembly Process TO-BE (During the Pilot) The TO-BE process is in a sense the same as the AS- IS process except in one important respect - that i s the trigger. The trigger is a predecessor to “1. Identi fication of parts…..”. No special arrangements influ encing the production context were made, except for the handli ng of the triggers (the magnetic sheeting). This is to satisfy the objective of testing the hypotheses in a real a ssembly environment, and not in a controlled labora tory. MINITAB 14 was used to calculate the sample size (n umber of engine assemblies to be studied). And was calculated to be 2541. This, 2541, is approximately two weeks (ten twenty-four days) production. There fore the experimental time frame within the production a rea was set to ten days. It was expected that this time frame would need updating during the experiment due for example to disturbances in production. In all experiments there is a possibility that the experim ent itself influences the results. In this pilot th ree (four) experiment groups were used with the purpose to cre ate an understanding of how this particular pilot m ight influence the personnel and thereby the results. Th e different experiment groups are divided by two di fferent variables: A: Interaction: (A1:“High” , A2:“Low”), B: Trigger: (B1:Present or B2:Absent). 3.0 Results The results can be categorized according to the sub ject groups. Control group A2B2: The results from this group are the historic data from the nine month per iod in 2006. This results are used as a reference v alue for the experiment groups. Experiment group A1B1: High interaction and trigger present. The experime nter was actively involved (asking questions, starting discu ssions about the work etc.) with the assembly perso nnel at their work stations and the new trigger was present . Experiment group A1B2: High interaction and trigger absent. As before the experimenter was actively inv olved with the assembly personnel at their work sta tions but the new trigger was not used Experiment group A2B1: Low interaction and trigger present. The experimenter was situated in the start of the assem bly line and there was no or very little interactio n between the experimenter and the assembly personal. The sin gle line was studied for a period of ten days durin g which 2600 engines were assembled. At four (out of twenty -four) of the assembly stations the existing inform ation system was changed to reduce the amount of informat ion to be assimilated by the workers, the timing of its 187 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN ASSEMBLY LINE INFORMATION SYSTEM STUDY presentation and its location. The use of simple co lour-coded cards and symbols resulted in the reduct ion of internal rejects by 40% on two of the assembly stat ions and to zero on the other two stations. On reve rsion to the original information system the rate of interna l rejects rose to equal or surpass those before the study. Full details of the results can be found in [16]. 4.0 Conclusions It is believed that changing the information system changed the workers’ behaviour through a reduction in cognitive stress levels. The pilot study has provid ed useful insights into the basis for modifying inf ormation systems and a further study of the final assembly o f heavy trucks is planned with an ultimate aim of determining a rationale for the design of informati on systems for use within the assembly of customise d products. The company expects to benefit in the lon g term through reduced rates of internal rejects br ought about by more appropriate information systems that more closely match the assemblers’ needs for inform ation. References [1] Bäckstrand, G., De Vin, L.J., Högberg, D. and Case, K. ”Attention, interpreting, decision-making and a cting in manual assembly”, in ’Innovations in Manufacturing’ , the Proceedings of the 23rd International Manufac turing Conference, University of Ulster, pp165-172, 2006. [2] Bäckstrand, G., De Vin, L., Högberg, D., Case, K. ” Parameters affecting quality in manual assembly of engines” Proceedings of the 22nd International Manufacturing Conference,’ Challenges Facing Manufacturing’, pp1 65-172, 2005. [3] Endsley R. M.“Situation awareness and human error: designing to support human performance”. Proceeding s of the High Consequence Systems Surety Conference, Albuque rque, 1999. [4] Johnstone, D., Bonner, M. and Tate, M. "Bringing h uman information behaviour into information systems research: an application of systems modelling" Information Research , 9 (4) paper 191, 2004. [5] Wharton, C. and Clayton, L. (1994) “The role of psy chological theory in usability inspection methods”. In Nielsen Jakob, Mack L. Robert (Eds.) Usability Inspection M ethods. John Wiley & Sons, Inc. New York, 1994. [6] Wilson, T.D. “Human information behaviour”, Special issue on information science research, Vol 3, No 2 , 2000. [7] Blasch, E and Plano, S.“JDL Level 5 fusion model: u ser refinement issues and applications in group tra cking,” SPIE Vol 4729, Aerosense, pp. 270 – 279, 2002. [8] De Vin, J. L, Ng, H.C. A., Andler F. S. and Oscarss on J.“Information fusion for simulation based decis ion support in manufacturing” Flexible Automation and Intelligent Manufacturing, FAIM2005, Bilbao, Spain, 2005. [9] Endsley, R. M. “Automation and situation awareness” . In Parasuraman, R. and Mouloua M. (Eds.), Automat ion and Human Performance: Theory and Applications, 1996. [10] [10] Machlup, F.“Knowledge: its creation, distribut ion and economic significance, Volume 1: Knowledge and knowledge production” Princeton University Press. P rinceton, New Jersey, 1980. [11] [Wilson, T.D. “Human information behaviour”. Specia l Issue on Information Science Research, Volume 3, No.2, 2000. [12] Meadow, T. C. and Weijing, Y.”Measuring the impact of information: defining the concepts.” Information Processing and Management, Vol. 33. No.6. pp. 697-7 14, 1997. [13] Wickens, C.D.“Engineering psychology and human perf ormance”. 3rd. Ed. Prentice-Hall Inc, 1999. [14] Donald, N.A.“The design of everyday things”. The MI T Press, London, England, 1990. [15] Reason, J.“Human error” Cambridge University Press, 1990. [16] Bäckstrand, G., Thorvald, P., de Vin, L., Högberg, D., and Case, K. ‘Information impact on work envir onment and product quality: a case study’, Nordic Ergonomics S ociety Conference (NES), Iceland, August 2008. 188 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINES FEEDING POLICIES Antonio C. Caputo, Pacifico M. Pelagagge Department of Mechanical, Energy and Management Eng ineering, University of L’Aquila, Italy Abstract Assembly line manufacturing systems require uninterrupte d availability of components and subassemblies to feed the workstations. Three policies are typically available for materials delivery to the shop floor, namely line-stoc king, kitting and kanban based (i.e. just in time supply). Each method has spec ific advantages and drawbacks also implying different performance levels in term s of work in process, material handling effort, space utilization, personnel require ments and costs. Policy selection is often a matter of qualitative judgement perhaps influenced by company- specific practices and tradition. In the paper descriptive models to design the components delivery system and compute required resources an d performance metrics are developed for the three policies. This provides production managers some quantitative decision tools to help in properly selecting the components delivery method at an early decision stage, and explore trade-offs between the three feeding policies. The possibility of adopting a mix of the delivery meth ods is also examined and a methodology to develop and evaluate hybrid line feeding policies is presen ted. Keywords: Assembly line feeding, line stocking, kit ting, kanban, model. 1.0 Introduction In assembly line manufacturing systems a decision h as to be made about the way components and subassemblies are delivered to the shop floor to fe ed the assembly stations. Three policies are typica lly available, namely line-stocking, kitting and kanban based (i.e. just in time supply). The selection am ong such policies is often a matter of qualitative judgement perhaps influenced by company-specific practices a nd tradition. However, the choice of the component del ivery policy strongly affects the performances of t he assembly system in terms of work in process, materi al handling effort, space utilization, personnel requirements and costs. In order to provide product ion managers some guidelines and quantitative decis ion tools to help in properly selecting the components delivery policies, the paper at first develops desc riptive models for the three policies. The models allow to design the components delivery system enabling the computation of size and number of containers for ea ch policy, as well as the requirements of personnel , 189 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES material handling and storage space. This allows to explore trade-offs between the three feeding polic ies for specific problem instances at an early decision sta ge in either single-product or multi-product settin gs. A comparison of cost (personnel and equipment) and pe rformances (work in process) of the three methods i s thus made possible. Subsequently, the possibility o f adopting hybrid delivery methods including a mix of the above policies is examined. A methodology to develo p and evaluate hybrid line feeding policies is then presented based on the ABC classification approach in order to properly associate the most suitable de livery policy to each component. 2.0 Overview of Line Supply Policies In a kitting policy no parts inventories are kept a t the assembly stations and the specific assortment of components required to perform the assembly operati on are grouped together and placed into a kit conta iner. Kits are prepared in a stockroom by the use of a pi ck list generated from the order’s Bill of Material s and then delivered to a nearby assembly line according to th e production schedule. In a kitting process, materi al is stored centrally which increases security in the co ntrol of physical inventory and also allows reducti on of raw materials inventory at a given service level. Kitti ng allows better control and minimization of the wo rk in process (WIP) and reduces manufacturing floor space utilization. Product changeover is simplified as o nly a change of the pick list is required and this suppor ts small batch size operations with a large product variety, while obsolete materials can be readily removed fro m the inventory. This does not happen if the invent ory is distributed on the shop floor. Material flow throug h the shop floor is simplified too as only the kits need to be moved to the assembly stations instead of individua l components containers. Kitting offers opportunity for better quality and productivity as parts are readil y available, checked and pre-positioned, and this s upports the assembler’s work. However, the kit preparation with the associated order picking is labor intensive an d non productive (with little or no value added to the pr oduct) even if automated order picking is sometimes feasible. Kit preparation increases space utilization in the stockroom. Errors in kit preparation, defective par ts included into a kit, or temporary shortage of parts may affe ct the assembly operations or the efficiency of kit ting process. Procedures for management of exceptions ar e thus required. A thorough discussion of kitting operations and related advantages or drawbacks is g iven in [1], [2], [3],[4]. In case of no kitting (c ontinuous supply), components containers may be stored along the line and resupplied in just in time fashion ado pting a kanban-based policy. Otherwise, components containe rs in bulk quantities are simply stored along the l ine and periodically replenished (line-stocking). Respect k itting these two policies save the order picking la bor but at the expense of a greater space utilization on the s hop floor, an increased containers flows and a grea ter work in process. The just in time approach allows to somewh at reduce work in process respect line stocking but requires more frequent handling of components conta iners. The main drawback of both approaches is that parts are stored at their point of use meaning that a station has to store enough quantity of all the component that are utilized at the station for every product configuration, and that if the same component is us ed in multiple stations it has to be stored separatedly a t each station. Finally, just in time and bulk line storage policies are not suited to variable product mixes. While kitting is an established practice in assembl y industries [5], [6], the choice of materials delivery will lar gely depend on the structure of the assembly proces s and from parts size and cost, kits size and containers size (some parts are simply non kittable). Johansson and Johansson [7] state that a kitting process is suitable for as sembly systems with parallelized flows, products st ructures with many part numbers, need for quality assurance and h igh value components. Kitting is considered most su itable for industries such as the electronic industry [8], [9], which deals with small parts and performs asse mbly tasks 190 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES quite often. However, it is reported that even JIT systems dealing with larger parts can benefit from kitting [8],[10],[11]. 3.0 Descriptive Models of Line Feeding Policies Among the scarce literature existing on modeling ma terial delivery to assembly lines, kitting has rece ived the greater attention. However, most available models u tilize queueing theory to analyse dynamic performan ces of kitting systems [12]-[15]. Nevertheless, these are not design methods and are scarcely useful when dif ferent component delivery methods have to be compared. Des criptive models useful for system design have been developed instead by Bozer and McGinnis [2] and Med bo [16]. Recently, Carlsson and Hendsvold [17] extended the Bozer and McGinnis model to compare ki tting with just in time kanban-based line storage a lso allowing the possibility of arbitrarily choosing th e delivery mode of each component. However, they do not take part size and weight into account. In this sec tion specific descriptive models for resource sizin g and performance estimation will be developed with refer ence to the three examined material feeding policie s. Models development is partly inspired from the earl y work of Bozer and McGinnis [2] but several distin ctive features have been included. Among the main differe nces Bozer and McGinnis do not explicitly take into account the human resources requirements and equipm ent cost or the economic value of WIP, as made in t his work, and consider only kitting vs general line sto cking policies without explicitly discriminating bu lk with just in time line supply. This work instead include s a kanban-based just in time feeding model and als o explores the possibility of mixed feeding policies. In this work all models are based under the hypoth esis of one warehouse with a single I/O point, and a single -product assembly line consisting of M workstations arranged serially. However, multi-product cases can be incorporated adopting weighted averages of requ ired data in order to reflect specific production mixes. The extension of the models to a multiple lines sy stem is straightforward. The assembly line has a constant d aily production volume and the considered time hori zon is one day. Material handling personnel is distinct fr om line staff which is in charge only of assembly o perations. Kits are prepared one at a time and delivered to th e first station of the line. Materials composing a single kit may be put into one or more separate containers acc ording to weight and space limitations. In kanban a nd line stocking systems each workstation has its own conta iners available and containers are not shared among multiple workstations. The same constant-speed vehi cle or walking operator is used to transport kit co ntainers and components containers. Empty containers are ret urned back to the central warehouse. Exceptional it ems, such as those very cumbersome and heavy requiring s pecial handling care are not included in the analys is. For meaning of symbols please refer to the nomenclature section at the end of paper. 3.1 Kitting A kit is here defined as a unit load holding all th e components required to assemble a unit of the fin ished product. However, according to weight and size of i ts components a kit may be made up of one or more distinct containers or totes. In the kitting policy it is assumed that a single kit is prepared for ea ch end product and is moved from the kit preparation zone located at the warehouse I/O station to the first station o f the line. Multiple kits can be accumulated at the start of th e line. In this work single kit preparation is cons idered (although batch kitting is allowed) and only travel ing kits (which move along the line with the assemb ly) are utilized. 191 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES The number of separate containers making up a kit i s n cont kit and it is determined on the basis of containers volume or their allowed weight. Unless the same con tainer is reused twice or more each day, the total number of containers required to manage the daily producti on D (where D is the number of kits n kit to be prepared and moved daily) is n c tot = n cont kit n kit = n cont kit D, while D is also the number of kits to be moved daily. The number of workers needed to operate the kit preparation an d transportation is N op as given in equation (2) where 2 n c tot /ω is the daily number of material handling moves fro m the warehouse to the line with full containers an d from the line to the warehouse with empty container s. Personnel cost is simply evaluated as C M = N op C op . Equipment cost C E is the sum of containers cost and transport equipm ent (i.e fork lifts, manual carts, tractor/trailers etc.) capital investment. The hold ing cost of the work in process is C WIP (equation 4) considering that ωk kits are simultaneously moved to the line and that C std i is the annual unit holding cost of i- th component. As far as space requirements on the s hop floor (S) are concerned, the space occupied by the single traveling kit can be neglected, while in cas e multiple kits are transported and accumulated at the first station S is given by equation (5) where n sl is the number of containers which can be stacked i n a column and a k and b k are the length of the containers base sides. n cont kit = max [ ∑ = N 1i k ii V nv ,∑ = N i ii p np 1 max ] (1) lop ctotmkit N 1i i psr/ op hη n] ω 2k[tn]ntN[t N ++ = ∑ = (2) lv ctot E hV 2n C L CCn Vctotc ω+= (3) 21 k N i iistdWIP nCC ω     = ∑ = (4) ( ) sln S kkitcontkk nba ω = (5) 3.2 Kanban (Just In Time) According to this policy material is resupplied at each station with a lead time LT in separate contai ners dedicated to each component type. Let define n ij as the number of items of the i-th component utili zed at the j- th station per unit finished product (n ij = 0 if component i is not utilized at station j ), then the number of containers needed at station j to hold the material needed to avoid starving during the supply lead ti me (with zero buffer stock) is n cont ij and the total number of containers utilized is n ctot . The total number of times that a 192 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES single component container is to be refilled and mo ved to the line is n m = n c tot /LT, therefore the workers requirement is N op and 2 n m /ω is the daily number of material handling moves fro m-to the warehouse. Personnel cost is computed as in the kitting case, while equipment cost is computed as shown in equati on (3) but including the storage racks cost C SR . This can be estimated as C SR = n c tot V c C SRU where C SRU is the rack cost per unit volume. The holding cost of the work in process is the holding cost of the average mater ial amount in the kanban containers along the line (equ ation 9) and the shop floor space S occupied by mat erials containers along the line is given by equation (10) .    = ii c ij ijcont p p v V nDLT n max ,min (6) ∑∑ == = N i ijcont M ij totc nn 1 (7) lop 'r/s h )2k( η ω mmfr op nttt N ++ = (8)     = ∑∑ == N i ijstd M ij WIP nDLTCC i 12 1 (9) ( ) sl1 n S ijcontcc N i M ij nba∑∑ == = (10) 3.3 Line Stocking In line stocking each station holds a different con tainer, periodically resupplied at time intervals w hich depend from the adopted containers capacity, for each sepa rate component it uses. The total number of contain ers along the line is n ctot (equation 11). The number of material handling wor kers is computed by equation (8), but since the number of replenishments is not dictated by the lead time but from the size of containers, n m is expressed as given in equation (12), and 2 n m /ω is the daily number of material handling moves fro m-to the warehouse. Personnel and equipment costs are comput ed as shown for the kanban feeding policy. The hold ing cost of the work in process is the holding cost of the average material amount in the containers along the line and is expressed as equation (13). Line storage con tainers in general will have a larger size than the corresponding containers utilized in the just in ti me policy. The shop floor space occupied by materia ls containers along the line is then computed accordin g equation (14). 193 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES ∑ = = M ij jtotc Nn (11) ∑∑ ==    = N i ii c ij M ij m p p v V nD n 1 max ,min (12)           = ∑∑ == N i ii c std M ij WIP p p v VCC i 1 max ,min 2 1 (13) ( ) sln S jcc M ij Nba∑ = = (14) 4.0 Analysis of Hybrid Feeding Policies Utilization of the above models allows to compare t he performances of the three line feeding policies in order to choose the best solution for any given scenario. However, the underlying assumption is that once th e policy selection is made all the components, apart from ex ceptions, are supplied with the same method. Nevert heless, improved results can be sometimes obtained when the feeding policy is selected at single componente le vel, meaning that hybrid feeding policies may be adopted where the entire set of component is partitioned i nto homogenous classes and a specific feeding policy is assigned to each class. While several partitioning criteria can be conceived, i.e in multi-product lines compon ents can be classified according to their commonali ty level with respect to the end products variants, in most cases a Pareto ABC classification referring to the economic value of the components can be appropriate. In this case, defining the economic value of a component a s its unit cost times the demand, components can be order ed according to diminishing values of the economic values and class A may group components having a cu mulative economic value about 80% of the total (thi s class will group about 20% of components). Class B will group remaining components having a cumulative economic value of about 15% of the total (usually a round 30% of components), while class C will compre hend the remaining 50% of components responsible for the residual 5% of overall economic value. Therefore, class A components will have a greater relevance and will be responsible for the greatest flows, WIP and hol ding costs, class C components will be scarcely relevant components with minimal flows or negligible holdin g costs, while class B components will represent the intermediate case. A hybrid delivery policy can be then identified by the string X/X/X where the sequence o f “X” indicates the policy associated to A, B, C cl asses respectively and X may assume the following values: KI = kitting, KA = Kanban-just in time, LS = Line storage. As an example the string KI/KA/LS means th at A class components are managed with a kitting po licy, class B components utilize a kanban based policy an d class C components are resupplied with a line sto rage policy. Within this framework one can choose three distinct policies for any of the three components c lasses, meaning that overall 27 options are available. Negl ecting the “pure” policies where the same method is used for all components, one has 24 theoretical hybrid p olicies. However, some are unfeasible or meaningles s in 194 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES practice. For example, the LS policy is not advisab le for class A components as it would cause excessi ve WIP and holding cost, while a kitting policy is not jus tified for class C components because their low hol ding cost allows high WIP to be tolerated and its reduction c an not offset the added cost of manual kit preparat ion. It follows that only the following options are likely to be considered for each class: Class A (KI, KA), Class B (KA, LS), Class C (KA, BL). This reduces the number of practicable hybrid policies to eight. Neverthel ess, one can consider that if KI, KA, and LS represent p olicies with a progressively reduced degree of WIP control and workforce effort, while ABC represent classes o f reducing economic relevance, it is of no use to i ncrease degree of control and effort while passing from cla ss A to C. Therefore, the choice can be further sim plified if, beyond the three pure policies, one narrows the opt ions to the following most promising hybrid policie s only, • Policy I: KI/KA/LS. This policy strictly controls WIP and holding cost by managing A class codes by kitting, while relaxes the control on Class C items which are less influencing. Class B codes are mana ged by kanban to reduce WIP but without imposing an exc essive order picking burden. • Policy II: KI/KA/KA. Most relevant codes are still managed by kitting b ut kanban control is extended to class C codes to pursue overall WIP reduction. Comm onality of Class B and C policy improves operationa l efficiency even if the handling effort is increased . • Policy III: KI/LS/LS. In this policy attention is focused on Class A cod es only and control is relaxed on Class B and C items. It is a trade off policy with lower workforce cost than the previous policies and intermediate WIP. • Policy IV: KA/LS/LS. Control is fairly relaxed except for Class A compo nents which are managed in just in time fashion. Pursues minimization of workforce cost while maintaining a degree of WIP control over most relevant components. Therefore, a strategy for selecting the optimal mat erial handling approach would include the following steps repeated for the three pure policies and the four s elected hybrid policies. A multicriteria rating of the computed performance indicators, properly weighed a ccording to the specific scenario, will then allow a comparison of the policy options and may help plant managers to make an informed decision. 1. ABC classification of materials according to econom ic value 2. Resource sizing (workforce and equipment) according to the above developed models 3. Performance estimation (WIP, holding cost, handling flows, personnel expense) 5.0 Conclusions Kitting, just in time and bulk line storage are dif ferent options to supply components and subassembli es to the workstations of assembly lines. Not a single techni que may represent the appropriate approach in any assembly situation, each having specific advantages and drawbacks,. However, to identify the cases whe re each approach is most suitable or to select the pre ferred method in a given scenario, is a challenging task especially owing to the lack of literature models a llowing comparison on a similar basis. As a results the selection of assembly lines feeding method is often a matter of qualitative judgement and common pract ice. In this paper descriptive models have been developed f or all of the three solutions enabling a plant mana ger to compare on a quantitative basis the respective perf ormances considering WIP, holding cost, equipment a nd workforce requirements and intensity of containers flows. Moreover, a procedure has been presented to enable a systematic analysis and comparison of hybrid poli cies where a components classification is carried o ut at 195 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES first and a specific material delivery method is as sociated to each class. This allows to rapidly expl ore mixed solutions which can be more suited to specific scen arios. References [1] Brynzér, H., Johansson, M.I., Design and performanc e of kitting and order picking systems, Internation al Journal of Production Economics, Vol. 41, 1995, pp. 115-125. [2] Bozer, Y.A., McGinnis, L. F., Kitting versus line s tocking: A conceptual framework and a descriptive m odel, International Journal of Production Economics, Vol. 28, 1992, pp. 1-19. [3] Medbo, L., Assembly work execution and materials ki t functionality in parallel flow assembly systems, International Journal of Industrial Ergonomics, Vol. 31, 2003, pp . 263–281. [4] Carlson, J.G.,Yao, A. C., Girouard, W.F., Role of m aster kits in assembly operations, International Jo urnal of Production Economics, Vol. 35, No. 1-3, 1994, pp. 2 53-258. [5] Butala, P., Kleine, J., Wingen, S., Gergs, H., Asse ssment of Assembly Processes in European Industry, Proc. 35th CIRP-International Seminar on Manufacturing Systems , 12-15 May 2002, Seoul, Korea. [6] Sellers, C.J., Nof, S.Y., Part kitting in robotic f acilities, Material Flow, Vol. 3, 1986, pp. 163-174 . [7] Johansson, B., Johansson, M.I., High automated kitt ing systems for small parts. A case study from the Uddevalla plant, Proc. 23rd Int. Symp. on Automotive Technolo gy and Automation, Vienna, 1990, pp. 75-82. [8] Ding, F.Y., Balakrishnan, P., Kitting in just-in-ti me production, Production and Inventory Management Journal, 1990, pp. 25-28. [9] Gunther, H.O., Gronalt, M., Piller, F., Component k itting in semi-automated printed circuit board asse mbly, International Journal of Production Economics Vol. 43, No. 2-3, June 1996, pp. 213-226. [10] Ding, F.Y., Kitting in JIT production: A kitting pr oject at a tractor plant, IE Solutions, 1992, pp. 4 2-44. [11] Das, T K., Analysis of kitting assembly systems con trolled by kanban, Proceedings of the Industrial En gineering Research Conference, 1993, pp. 133-137. [12] Ramachandran, S., Delen, D., Performance analysis o f a kitting process in stochastic assembly systems, Computers & Operations Research, Vol. 32, 2005, pp. 449–463. [13] Ramakrishnan, R., Krishnamurthy, A., Analytical app roximations for kitting systems with multiple input s, IIE Annual Conference and Exposition, 2005. [14] Som, P., Wilhelm, W.E., Disney, R.L., Kitting proce ss in a stochastic assembly system, Queueing System s, Vol. 17, 1994, pp. 471-490. [15] Takahashi, M., Osawa, H., Fujisawa, T., A stochast ic assembly system with resume levels, Asia-Pacific Journal of Operational Research, Vol. 15, No. 2, 1998, pp. 127 -146. [16] Medbo, L., Material kitting system design. Departme nt of Transportation and Logistics, Chalmers Univer sity of Technology, G.oteborg (working paper), 1999. [17] Carlsson, O., Hensvold, B., Kitting in a high varia tion assembly line. A case study at Caterpillar BCP -E, Master Thesis, Lulea University of Technology, 2008. Nomenclature C c Container unit cost n m Number of daily replenishment and transport cycles of a materials container C E Equipments capital cost n sl Number of stackable containers on the shop floor C M Personnel annual cost nc j Number of different components utilized at station j C op Annual cost of a worker N Number of different comp onents in a finished product C SR Cost of storage racks N j Number of different components assembled at j-th s tation C SRU Cost of storage rack per unit volume N op Number of operators C std i Unit holding cost of i-th component p i Weight of the i-th component C V Vehicle unit cost p max Maximum allowed weight of a container C WIP Holding cost of the work in process q i Quantity of i-th component moved replenished in ba tch D Daily demand finished products S j. Set of component codes utilized at the j-th works tation h l Daily working hours (hours/day) t fr Time to fraction bulk component cartons in warehou se i Component identification index tm Warehouse to assembly line kit containers trip tim e j Workstation index (j = 1 to M) tm’ Warehouse to assembly line containers trip time k Number of operators to move a kit t p Picking time for a single item L Length of one way transport route t r/s Time to locate and reach components in the warehou se LT Containers lead time in kanban policy v i Volume of the i-th component M Number of workstations V c Volume of a components container (a c x b c x c c , being a c , b c , c c the dimensions of the container) n cont kit Number of separate containers in a kit V k : Volume of a kit container (a k x b k x c k , being a k , b k , c k the dimensions of the container) N ctot Total number of containers V V Transportation velocity n i Mukltiplicity of i-th components in a finished product ηop Average workers efficiency 196 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ANALYSIS AND OPTIMIZATION OF ASSEMBLY LINE FEEDING POLICIES n ij Number of items i utilized at station j per each unit of finished product ω Number of containers transported simultaneously n kit Daily number of kits to be moved ωk Number of kits simultaneously moved to the line 197 198 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY DECISION SUPP O RT SYSTEM FOR PALLET UTILISATION IN THE SHEET FEEDER INDUSTRY Essam Shehab 1 , Olatz Celaya 1 , Adrian Swindells 2 1. Decision Engineering Centre, Cranfield University, Cranfield, Bedfordshire MK43 0 AL 2. Abbey Corrugated, South Mills, Blunham, Bedfordshir e MK44 3PH , UK Abstract The sheet feeder industry is a business that provides a w ide range of cardboard sheets to its customers. Pallet utilisation in such industry sect or has a direct impact on the cost and time related to the preparation and delivery of the p roduct. This paper presents a decision support system for pallet utilisation i n the sheet feeder industry. The developed system comprises a knowledge – based system, mate rial handling modules, databases and user interface. Furthermore the system encompasses pallet database, truck database and material features databases. It also tak es into account several factors that affect directly the pallet loading problem . Cardboard sheet size, type of flute, type of paper, cardboard sheet configuration, custom er restrictions and type of lorry for the product delivery are the key factors that have been considered to maximise customers’ orders onto the pallet. These data has bee n gathered by semi- structured internal interviews and questionnaires within the sheet feeder industry and visits to the shop floor. The developed system has the capability to estimate the numb er of corrugated cardboard sheets that should be loaded per pallet as well as number of stacks per pallet. Additionally the system predicts the number of she ets per stack and the type of pallet that should be used. The system provides the shop – f loor operators with an effective tool for pallet utilisation in order to reduce the ir decision time. Therefore, the developed system provides cost and time savings in terms of im provements of the pallet and vehicle volume utilisation for the preparation and d elivery of the corrugated board product. Keywords: Sheet Feeder Industry, Pallet, Cardboard Sheet, Transport, Logistics, Decision Support Systems. 199 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY 1.0 Introduction In the last few years, corrugated industry has unde rgone different mergers and acquisitions, hence the cardboard business is immersed in a very competitiv e market where each company fights for being the be st in terms of service, product quality and cost. As a re sult of this, nowadays companies are making a lot o f effort in optimising the different tasks that are involved in their business: from the management of an order to the distribution to the customer. The customers of this industry sector are very demanding because they re quire a very specific cardboard sheet sizes. Morabito et al . [1] stated that “The mission of optimisation is t o get the right goods or services to the right place, at the right time and in the desired conditions, while mak ing the greatest contribution to the company”. However, in the sheet feeder industry some improvem ents have been done in the reduction of waste mater ial. A cardboard is a lightweight and sensitive material , so many cares should be taken into account for it s transport. Sheet feeder industry is manufacturing c ardboard sheets of many sizes which require differe nt pallet sizes for their delivery. A pallet is a factor in t he material handling line and any small improvement in its use will have a big impact for the company. The less pa llets the company uses, the less time will be use t o manage and transport the products. This is traduced in tim e and cost saving. Deciding the optimal height for a pallet is not an easy task because customers sometimes have their ow n requirements. Furthermore, the pallet has its own r estrictions as well such as maximum weight and heig ht. The transport vehicle is another very important factor in this decision. The maximisation of the space ava ilable in the truck is another concern. Many pallets size is proportional to the space available in a truck. But for those pallets that have a specific size, because they are customers’ bespoke pallets, the optimal use of the space in the truck is an issue. The developed decision support system aims to provi de the shop – floor operator with the necessary information to manage customers’ orders in the most efficient way. Many variables are involved in this system such as the weight of the cardboard. The system pre sented in the paper improves the current system to distribute customers’ orders onto pallets. Less num ber of pallets will be used for the delivery of an order which means cost and lead – time reduction 2.0 Literature Review A pallet is a portable, horizontal, rigid platform used to store, stack and hand loads as a single mas s [2, 3]. Trebilcock [4] and Fornicio [5] explain that the pa llet is the most used item to transport products fr om one point to the other. The users of pallets are expect ed to buy the same number of pallets or more year b y year. The most used pallets are made from wood and plasti c. However, wooden pallets are down slightly in the last few years [6]. Different aspects of these items for the transport are looked when they are bought, lik e: durability, purchase price, availability, etc. 200 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY As well as being different materials to make a pall et, there are different pallet sizes in the market. The pallet dimensions are made in a way that they are submulti ples of the place that will be used to transport th e pallets [7]. Nowadays, there are two normalized pallet dime nsions: the Europallet (1200mm x 800mm) and the universal pallet (1200mm x 1000mm). Apart from the standard pallets that can be found, nowadays there is a tendency to make a pallet for new products. This ma kes an increase in the cost of pallet management an d this variety causes less efficiency in the distribution chain. Trade costs increase too because those produ cts that should be transported from one country to another ( countries that have different pallets standards) at the border are unloaded from the initial pallet to the second one [2]. Morabito et al. [1] argue that anot her disadvantage is the lack of well- accepted standard pallets and compatible trucks designs for these st andards that has as a consequence an undesirable empty spac e of the cargo loading. There is an option to fill these pallets when the c ustomer makes an order to fill a pallet with few ma terial, this is called mixed pallet. Instead of increasing or de creasing his amount of products of the order to fil l a pallet, a pallet with products of different customers’ produc ts is loaded. But the use of this pallet may compli cate the logistics. Firstly, the manufacturer needs to know how the products should be loaded onto that pallet; this decision is taken every time a customer order is pl aced. This pallet design should be efficient in ter ms of pallet space and it also should be stable. Because if the load is not stable enough the material can be damag e and this is the main reason [8]. Therefore, the design of th ese mixed pallets is a time consuming task. Hoffman [9] reports that if the stability is improved, the prod uctivity too, including the reduction of damage in transport. It could be found three groups of designers in the mat erial handling field: designers of handling equipme nt, designers of the pallet and the designers of the pa ckaging ([10]. White et al. [3] point out that a co st reduction in the pallet will save money in those activities a ssociated with the movement of products between the seller and the customer. For example, some designers inves t efforts in redesigning the pallets because pallet interacts with other components of the supply chain like: vib ration interactions during the loading and conveyin g, shock transferences while the forklift is handling the ma terial, etc. In spite all of this, there is a speci al concern in saving transport costs. 3.0 Methodology A combination of research methodology approaches ha s been employed in this research study. Firstly, a familiarisation stage has been conducted through co mprehensive literature review and visits to the spo nsoring company. The second phase was to conduct semi - str uctured interviews with the shop-floor operators an d operations directors to have a better understanding of the business needs. This was combined with measurements of the trucks and the pallets and inte rnal documentation from the company. In this stage, the researcher also distributed questionnaires to other sheet feeder companies to know if the system could be applied also to more companies. In the third phase, the data gathered were analysed for the subsequent development of the system. Then the system has been developed following the logic rules obtained by analysing the data. Finally, the developed system h as been validated, through real – life case studies and expert opinions. 201 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY 4.0 Overall Structure of the Developed System The aim of the Decision Support Systems is not maki ng decisions. These systems should give a serial of assessment decision tools to improve the results of the decision [11]. The developed system has the ai m of improving the management of customers’ orders onto pallets. The overall structure of the prototype sys tem is shown in Figure 1. The system consists of three databases. These datab ases have data of the different pallet and truck si zes and material features that are in the company. All thes e data interact with the user interface. It has been developed some logic rules and equation s that with the data inserted and the databases giv e a decision to the shop – floor operator. This decisio n provides information about: how many sheets per p allet should be loaded, how many full pallet are used, if there is any part pallet and which pallet size sho uld be used for the product delivery. 5.0 System Implementation and Scenario The system has been developed by means of the VBA p rogramming language. This language which embedded in Microsoft Excel which was seen as an ideal mediu m for achieving the goals of this research. Fig. 1. Overall structure of the developed system 202 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY Figure 2 illustrates the system scenario and screen shot where the user of the system should interact t o input a customer’s order. Firstly, the various data of the customer requirements are input to the system. The system prompts the user if the customer has specified the number of sheets per stack. If this the case, the u ser has to input the specified number of sheets which should b e loaded per stack. Therefore the shop-floor operat or has to follow the customer requirements. The user has to specify if the number of stacks per pallet was a customer requirement. Otherwise, the planning department decides the number of stacks per pallet. The number of stacks per pallet is detailed. More customers’ restrictions are required in the system. The overhang is a factor to consider because it af fects to the system when it has to choose the size of the pallet for the customer order. If the customer allows ove rhang the system takes a pallet some centimetres smaller than the size of the cardboard. The value of these cent imetres varies regarding the type of flute that the cardboa rd sheet is made of. Another customer restriction is the possibility of applying over sheets to the number of cardboard she ets that are loaded onto the pallet. Some customers want a s pecific amount of sheets in their orders. If the cu stomer allows over sheets a tolerance can be applied. This tolerance varies regarding the customer and the ov erall quantity of cardboard sheets per order. This tolera nce adds an extra percentage of number of cardboard sheets to the overall quantity of the order. In another fi eld of the system, this tolerance it is asked. So, that the system knows the tolerance that should be applied. The last field to fill in the customer requirements section is if the customer requires a specific pal let to use. If that is the case, it is asked which are the dimensi ons of the pallet that he wants and if it is a stan dard or bespoke pallet. In this situation, the system check s if the pallet that the customer has specified fit s properly with the order. If not, the system shows a message to the user advising him to ask the customer to req uire another type of pallet. The reason why it is import ant to ask if the pallet is standard or bespoke is because there is a difference of thickness between them. Th is difference of thickness it is traduced in the nu mber of cardboard sheets that can be loaded per pallet. The thickest the pallet is the less number of cardboar d sheets can be loaded. On the other hand, if the system has the freedom of selecting the type of pallet that c an be used only standard pallets will be taken into considerat ion. The reason for this is because only standard p allets are known beforehand. Bespoke pallets are only consider ed when the customer requires them. Bespoke pallets are purchased when the customer asks for them. The seco nd reason is that standard pallets are the ones tha t best fit into the truck because their dimensions are proport ional to the dimensions of the truck. Secondly, the features related to the order itself are required. In this section the first thing that the system requires is the overall quantity of cardboard sheet s of the order. Next, the dimensions of the cardboard sheets are in troduced: the width and the length. If the customer does not specify the type of pallet to use the system compar es the width and length of the cardboard sheet with the width and the length of the twenty four pallets ava ilable in the warehouse. And if the customer requir es a specific pallet, the system compares these dimensio ns with the dimension of the cardboard sheet. 203 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY The flute is also specified because the type of flu te affects to the thickness of the cardboard. Conse quently, to the number of sheets that can be loaded per pallet if the system has the option of making this decisio n. Fig. 2. Overall system scenario and an example of i nput window The cardboard configuration and the type of paper u sed for the manufacturing of these cardboard sheets are asked by the system. These factors affect directly to the weight and the height of the cardboard. For example, if the cardboard sheet required by the customer is single face, this kind of cardboard sheet will ligh ter and thinner than a double wall cardboard sheet. The rea son for this, it is because this kind of cardboard sheet is made of less number of papers. It influences the ty pe of paper used for the manufacturing of the cardb oard as well because not all the papers weight the same. Finally, the type of truck used for the product del ivery is another factor included in the system. Reg arding the truck, more or less space will be available for the product delivery. There four types of trucks in th e company and they have different sizes. After completing all the required fields, the syste m automatically creates the decision for pallet uti lisation. Figure 3 shows an example of the outputs generated by the developed prototype system. The system provi des the pallet width and the pallet length if the custo mer does not require a specific type of pallet. It matches the 24 standard pallets in a company to find out which pallet fit properly with the customer’s order. The reason is that these standard pallets dimensions can fit with the dimensions of a truck. 204 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY In the system is detailed as an output the number o f full pallets that are used. It is called full pal let to a pallet filled to the 100% of its maximum capacity. It also specifies if there is any part pallet. It is calle d part pallet to a pallet filled less than the 90% of its maximum ca pacity. As it has been mentioned before, if the number of c ardboard sheets in the last part pallet is less tha n the half of the maximum capacity that can be loaded, a redistri bution action is done. This means that the system redistributes these sheets onto the other full pall ets. So, this last part pallet is removed. If in th e last part pallet there are more sheets that the half of the maximum number of sheets that can be loaded per pallet, thi s last part pallet is kept. And the customer’s tolerance is app lied if it is possible. After applying the toleranc e, the system details the quantity of cardboard sheets that are i n the last part pallet. If the system does not make the redistribution ther e will be only full pallets and it could be part pa llets too. The number of cardboard sheets per stack and the nu mber of sheets per pallet in the full pallets are s pecified. Figure 3 shows the system output window. It can be appreciated that there are two outputs namely “Numb er of Sheets per Stack 1” and “Number of Sheets per Palle t 1”. They represent the number of cardboard sheets per stack and the number of sheets per pallet. The over all number of stacks of the full pallets is defined by the button “Number of Stacks Type 1”. Fig. 3. System output window If the redistribution action is not carried out the system reports the user if there is any part palle t and the number of sheets in the last part pallet. If the sy stem makes the redistribution action there will be two kinds of full pallets. The reason why there will be two kind s of full pallets is because the system proceeds to redistribute equally onto the other pallets those c ardboard sheets of the last part pallet removed. So , it will add one cardboard sheet more per stack in the full pall ets until there are no more cardboard sheets to dis tribute. So, some full pallets will have one sheet more per stac k comparing to the rest of full pallets. 205 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 D ECISION SUP P O RT SYSTEM FOR PALLET UTILISATION IN T HE SHEET FEEDER INDUSTRY 6.0 Conclusion It may be stated that the system presented enhances the production line and the planning department. Moreover, the author believes that there is not a s tandard way to choose the most suitable pallet for each order. Nowadays, sheet feeder companies have a shop – floo r that makes this decision. So, it can be stated th at this system is the first step done for the standardisati on of the pallet selection task. The weight is a new factor to consider into the bus iness of the sheet feeder industry. The conventiona l cardboard is a lightweight and strong material but with the new trends of material (microflute) appear ing in the sheet feeder industry, the weight plays an importan t role. The microflute cardboard is a very delicate material and if the pallet loads too much quantity of this k ind of material, the material can crash itself. The developed system will improve the current pract ise giving the following benefits: a maximisation o f pallet and truck utilisation, a reduction in the overall n umber of pallets used, cost and time reductions in the preparation of customers’ orders. Acknowledgement The authors would like to thank Abbey Corrugated Lt d for sponsoring this research project. Special appreciation is also extended to the operations dir ector, IT Manager and shop floor crew at the compan y. References [1] Morabito, R., Regina, S., Widmer, J. A., (2000). Lo ading Optimization of Palletized Products on Trucks . Transportation research Part E: Logistics and Trans portation Review , vol. 36, no. 4, pp. 285 – 296. [2] Raballand, G., Aldaz – Carrol, E., (2005). How do D iffering Standards Increase Trade Costs? The Case o f Pallets. World Bank Policy Research Working Paper , pp. 1 – 20. [3] Trebilcock, B., (2002). Pallets are here to Stay. Modern Materials Handling, pp. 29 – 31. [4] White, M. S., Hammer, P., (2005). Pallets Move the World. Forest Products Journal, vol. 55, no. 3, pp. 8 – 16. [5] Fornicio, H., (2000). Cutting Pallet Expenses. Candy Industry. Pp. 48 – 50. [6] Maloney, D., (2000). What Buyers Say about Pallets. Modern Materials Handling Lines , pp.66 -70. [7] Albardiaz, M. A., Castillo, J. F., (2000). Logistic s. Analysis in the Sector Pallets. Palets. Distribución y Consumo , no. 51, pp. 31 – 38. (In Spanish). [8] Fletcher, S., (1990). Technological / Physical Chan ges in the Environment. Journal of Food Research , pp. 85- 88. [9] Hoffman, W., (2006). Dow Starts Damage Study. Traffic World , pp. 1. [10] White, M. S., (1997). Optimizing - Unit Load Design – a New Systems Approach to Improving Material Handling Efficiency. Centre for Unit Load Virginia. [11] Jiancong, F., Yongquan, L., Qingtian, Z., (2006). M ethod for Designing Organization Decision Support S ystem Framework. Journal of Systems Engineering and Electronics, vol. 17, no. 4, pp. 764 – 768. 206 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION SCH EDULING OF NON-REP ETITIVE LEAN MANUFACTURING SYSTEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION Theopisti C. Papadopoulou ∗, Alireza Mousavi Brunel University, School of Engineering & Design, Uxbridge, Middlesex, UB 8 3PH , UK ∗ Corresponding Author: Tel: +44 (0 ) 18 9 5 26 5 8 8 5 , Ema il: theopisti.papadopoulou@brunel.ac.uk Abstract World-class manufacturing paradigms emerge from specific types of manufacturing systems with which they remain associated until they are obs olete. Since its introduction the lean paradigm is almost exclusively implement ed in repetitive manufacturing systems employing flow-shop layout configurations. Due to its inherent complexity and combinatorial nature, scheduling is one application domain whereby the implementation of manufacturing philosophies and best pr actices is particularly challenging. The study of the limited reported attempts to e xtend leanness into the scheduling of non-repetitive manufacturing systems with fun ctional shop-floor configurations confirms that these works have adopted a similar approach which aims to transform the system mainly through reconfiguration in order to increase the degree of manufacturing repetitiveness and thus facilitate t he adoption of leanness. This research proposes the use of leading edge intelligent agent simulation to extend the lean principles and techniques to the scheduling of non-repetitive production environments with functional layouts and no prior reconfiguration of any form. The simulated system is a dynamic job-shop with stochastic order arrivals and processing times operating under a variety of dispatching rules. The model led job-shop is subject to uncertainty expressed in the form of high priority orders unexpectedly arriving at the system, order cancellations and machine breakdowns. The ef fect of the various forms of the stochastic disruptions considered in this st udy on system performance prior and post the introduction of leanness is analysed in te rms of a number of time, due date and work-in-progress related performance metrics. Keywords: Lean Manufacturing, Just-in-Time, Schedul ing, Shop-Floor Control, Non-Repetitive Manufacturing, Job-Shops, Performance Modelling, In telligent Agent Simulation 207 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION 1.0 Introduction The lean manufacturing paradigm devised by Toyota g rew into a global phenomenon which is still attract ing the undiminishing attention of both the industry an d the academia [1]. Lean production scheduling and shop- floor control are exercised through a set of key le an concepts, techniques and tools integrated under the umbrella of Just-in-Time (JIT) pull production. Non etheless, the majority of these critical enablers w ere developed in line with the design and operational c haracteristics of flow-shop layout configurations f ound in repetitive production systems in which leanness was originally introduced. This consequently led to on ly scarce attempts to implement the lean paradigm in t he scheduling of non-repetitive manufacturing environments. Group Technology (GT) and layout reconfigurations h ave been proposed in the limited attempts reported in the literature to increase the degree of manufactur ing repetitiveness and facilitate the implementatio n of lean scheduling in complex non-repetitive production set tings. Whilst the majority of these studies report satisfactory improvement in system performance resu lting from the adoption of leanness they fall short to address the full size and complexity of real-life a pplications. More specifically they employ solution methodologies that downsize the scheduling problem considered or address a simplified version of it wh ich often ignores the openness of the system and merely deals with its deterministic version. Scheduling problems particularly those which are go od approximations of real-life systems are highly c omplex combinatorial problems the optimisation of which is classified as NP-hard. The large number of input parameters, their interdependencies as well as the stochastic nature of many of these parameters calls for modelling methodologies which offer high level repr esentation and can manage efficiently the complexit y and volume of interactions pertaining ever-evolving sch eduling systems. Constant advancements in computer technology coupled with the rapid evolution of simu lation and artificial intelligence however, call fo r the issue of the transferability of leanness into the schedul ing functions of complex non-repetitive manufacturi ng systems to be revisited. This research employs stat e-of-the-art agent-based simulation to extend lean pull production control to the scheduling of dynamic non -repetitive manufacturing job-shops which are subje ct to machine breakdowns and unexpected variations in cus tomer demand. The remainder of the paper is organised as follows: Section 2 presents a brief review of the literatur e focusing on the implementations of leanness in the schedulin g of non-repetitive production systems as well as o n applications of agent-based simulation in lean sche duling. Background information on job-shop scheduli ng and shop-floor control is presented in Section 3 al ong with a brief introduction to the push and pull production policies considered in the framework of this study. Section 4 gives an overview of the two agent-based architectures built to model the operation of the j ob-shop scheduling system under investigation and t o test its performance under push and tight pull control. The section also presents the functionalities added to the agents of both architectures to model uncertainties relate d to unexpected demand changes and machine breakdow ns. The parameters determining the experimentation sett ing in which the simulation runs were performed are analysed in Section 5. The simulation output from t he various experimentations and comparisons drawn o n the system’s performance under push and the proposed le an pull shop-floor control are summarised in Sectio n 6 which also presents brief concluding remarks on the performance of the proposed modelling methodology. 208 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION 2.0 Literature Review In spite of the general consensus in both the acade mia and industry that the lean paradigm is applicab le merely on repetitive production systems and the subsequent lack of support to its transferability, a review o f the literature reveals a small number of research works investigating the extension of leanness in the sch eduling of non-repetitive production environments. 2.1 Implementations of Lean Scheduling in Non-repetitive Manuf acturing Contexts Earlier works studying the extension of leanness in to non-repetitive production environments share a c ommon point of departure. They recognise the non-repetiti ve nature of the manufacturing operations performed in these facilities as the strongest impediment to the application of critical lean scheduling and shop-f loor control enablers. To this end, they propose functional layo ut adaptation or reconfiguration to increase the de gree of manufacturing repetitiveness in these systems and t hus facilitate the introduction of leanness into th eir scheduling functions. One of the first research wor ks highlighting the need to reform job-shops into m ore lean- friendly shop-floor configurations is presented in [2]. The author proposes a move towards the reconfi guration of functional layouts into cellular layouts, Flexib le Manufacturing Systems (FMS) or job-shop “islands ”. The utilization of MRP as a higher level planning and i nventory management system and the implementation o f JIT shop-floor control at the lower level combined with the rate per day schedules and back flushing are introduced in a reconfigured production system to s upport its lean transformation in [3]. Stockton and Lindley [4] propose process sequence cell layouts as an alt ernative to GT cells to enable the material flow to be controlled by kanbans in High Variety Low Volume (H VLV) production environments. Hybrid push/pull dual - card kanban control is implemented in different sho p-floor configurations in order to study the effect of various contextual factors e.g. batch size, materia l handling mechanisms etc and of their trade-offs o n system performance in [5]. With no prior adaptation or modification of any for m to alleviate the serious restrictions imposed by certain design and operational characteristics of non-repet itive manufacturing functional configurations, earl y studies investigating the direct introduction of leanness i nto the scheduling functions of the former focused on applications not representative of the size and com plexity of real-life problems. Despite their limita tions these studies confirm an optimised performance resulting from the adoption of leanness. One of the first comprehensive attempts to implement leanness in a n on-sequential however simplified context, is presen ted in [6]. In a similar study, Gravel and Price [7] emplo y simulation to test the performance of a job-shop under kanban control and a selection of dispatching rules developed in the framework of their work. The effe cts of pull control introduced in two alternative modes, i .e. tight pull and CONWIP on the performance of a S mall- to-Medium Enterprise (SME) job-shop operating withi n a broader Make-To-Order (MTO) supply chain are modelled and analysed in [8]. A HVLV job-shop setti ng with stochastic arrival and processing times is considered in [9] whereby the results of the agent- based simulation showed that tight pull control exe rcised by kanbans outperformed the initial push system. A bas estock pull control policy is introduced in a job-s hop setting in [10]. No machine breakdowns or unexpecte d variations in demand are considered in the agent- based modelling methodology employed to test the system p erformance after the introduction of pull control. 209 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION 2.1 Applications of Agent Based Systems in Lean Scheduling In their majority, the applications of multi-agent systems and modelling methodologies in lean schedul ing to date study the implementations of leanness in repet itive manufacturing settings utilising flow-shop la yout configurations. An agent-based approach to address the problem of minimising the JIT earliness/tardine ss weighted deviation in a parallel machine setting wi th stochastic order arrivals is proposed in [11]. A n autonomous decentralised system for minimising inte rmediate and end product storage costs, changeover costs and due date penalties for JIT scheduling is presen ted in [12]. The performance of the proposed system is tested by considering a multi-stage flow-shop and e xperimentation results confirm the effectiveness of the proposed system in meeting the aforementioned JIT s cheduling objectives while achieving considerable savings in computational time. Frey et al [13] deve lop a multi agent system for production planning an d control which they compare with other conventional centralised approaches. The benchmarking scenario adopted in their study considers the case of a mult i-level assembly where material flow is controlled by Kanbans. In a recently published study, Papadopoulo u and Mousavi [14] adopt a multi-agent modelling approach to apply lean scheduling and shop-floor in a non-repetitive functional layout with particular focus on controlling the constant work-in-progress in the sy stem. Their study considers a job-shop with dynamic order arrivals and processing times but does not account for other stochastic factors affecting the system a s it assumes negligible machine downtime, order cancella tions and rush order arrivals. 3.0 Lean Scheduling and Shop-floor Control in Job-shops Job-shops are the dominant shop-floor settings in n on-repetitive manufacturing environments. They empl oy functional layout configurations whereby equipment carrying out the same type of processing is grouped together and positioned in distinct areas of the sh op-floor. Following the introduction of lean manufa cturing, the two prevalent production control modes are push and pull with their names pointing to the way the system responds to actual customer demand. A job-shop oper ating in push mode typically comprises a number of disconnected production stages (workstations). In f ront of every workstation there is input buffer wit h theoretically infinite capacity. When actual demand information is received for a certain product type production is triggered at the first stage of its p rocess routing. If the first station in the process sequence of this job is busy the job joins the queue of waiting jobs in the input buffer in front of the workstation. J obs completing their processing at one workstation are pushed to the input buffer of the next workstation in the sequence without any consideration of its demand or workload. Each production stage in a job-shop operating in pu ll production mode can be viewed as a production- inventory station comprising an input buffer, one o r more machines and an output buffer. Apart from th e movement of parts, other types of entities that mov e within a pull production system are demand and production authorisations. A part is released from the output buffer of a preceding stage into the inp ut buffer of the subsequent stage in the sequence only if aut horisation for the release of this specific product type is available. In contrast to the physical movement of parts downstream, the movement of customer demand t akes place only logically and in the opposite direction (upstream). Production authorisations can be either physical cards (kanbans) or logical signals generated by a s oftware scheduler. 210 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION Whilst the implementation of push production contro l is quite straightforward, pull production control is more complex and can be exercised by adopting various al ternative pull production control policies [15]. Fi gure 1 below illustrates the basic principles of operation of the Kanban Control System (KCS) adopted in our study in the case of a simplified manufacturing system with two production stages in series. Fig. 1. KCS queuing network with synchronisation s tations, case of two serial stages [16] Queue PAi in the output buffer of stage i contains pairs of stage i processed parts and stage i produc tion authorisations whereas queue DAi+1 denotes pairs of demand and production authorisations for the produ ction of new stage i+1 parts. Queue Ii represents the inp ut buffer of stage i whereas the raw material buffe r and customer demand are represented as queues Po and D3 respectively. 4.0 Intelligent Agent Modelling and Simulation The agent-based simulation models developed in the framework this study, were built using JACK Intelli gent Agents™ [17]. JACK™ is a third-generation commercia l framework for building and running industrial and research multi-agent applications. The framework be nefits from the underlying JAVA infrastructure and multi- threading environment which offer high levels of pe rformance, concurrency and efficiency. A multi-age nt architecture is designed to model the operation of the scheduling system prior to the introduction of leanness and to benchmark its performance. This architecture is then modified to simulate the system’s operatio n after the introduction of lean kanban-pull production con trol, Figure 2. Both architectures incorporate unce rtainty expressed in the form of high priority orders arriv ing at the system unexpectedly, order cancellations and machine breakdowns. Fig. 2. Multi-agent architecture of the pull system Circulation of Stage 1 kanbans Circulation of Stage 2 kanbans 211 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION Fig. 3. Interface of the FMA In the architecture of the initial push model the S ystem Manager Agent (SMA) is responsible for creati ng the different job types processed in the modelled syste m. It assigns a workstation to each process step in the job’s task list and sets the associated processing time. The Job Manager Agent (JMA) carries all necessary information about the job including the data determ ined by the SMA and other time-related data collect ed during its processing. The JMA manages the job’s fl ow through the system by routing the job from one workstation (forward scheduling). When the processi ng of the entire job is completed, the JMA provides the job’s data to the Performance Monitor Agent (PMA) w hich calculates the performance metrics generated i n the simulation output. The input buffer queue in fr ont of each workstation is represented by a Worksta tion Input Buffer Agent (WIBA). Each WIBA is responsible for exchanging information with the JMA and for updating its list following the addition/removal of jobs to/from the input buffer it manages. The Work station Supervisor Agent (WSA) holds information on machine identification and status, i.e. busy/idle and is responsible for assigning jobs queuing in the works tation’s input buffer to the machines available in the workstation. The change of machine status is commun icated by the Machine Agent (MA) to the WSA whenever the machine’s status changes. The last age nt type available in this architecture is the Dispa tcher Agent (DA) which performs the selection of the next job to be processed by employing a number of dispatching rules. In order to model the proposed pull production syst em, the agent-based architecture of the initial pus h model is modified by introducing a new agent type, i.e. t he Workstation Output Buffer Agent (WOBA). These ag ents exchange information with the JMA on the availabili ty of inventory and update their databases whenever inventory is added (removed) to (from) their lists. At system initialisation their inventory lists con tain pre- determined levels of zero due date inventory for al l the different job types processed at the respecti ve workstation. Further modifications to the initial m odel concern additional functionalities performed b y the JMA. Following the arrival of a new job, its JMA re quests information on the availability of a fully p rocessed (zero due date) job from the last WOBA in the job’s task list. If confirmation is received, the JMA re places the zero due date of the already available job with the actual due date of the newly arrived job which the n removes from the WOBA’s database. After exchanging informat ion with the PMA, the job agent updates the job’s d ata by replacing its original due date with a zero due date and releases the now “zero due date” job to th e system by breaking down the job’s task list and executing it sequentially but in reverse order (backwards). H owever, if no confirmation is received, the JMA logs its re quest for fully processed job with the WOBA of the last workstation in the job’s task list and puts the rel ease of the job on hold until inventory is finally available. Modelling machine breakdowns requires the introduct ion of the Failure Manager Agent (FMA), Figure 3. I f the failure takes place whilst work is in progress without any damage caused to the part, the processi ng of the job will be resumed after the downtime period. Howe ver, in case of the work-in-progress being damaged, its JMA will remove the job from the machine and report back to the SMA and its life will be terminated. T he cancelled job will re-enter the system and start it s processing again and for that the SMA will genera te a new 212 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION job with a new due date to compensate for the time lost. Under pull production control a breakdown on a busy machine can only affect the zero due date replenish ment jobs. A machine failure resulting in the damag e of the replenishment job being processed would require the JMA to remove the affected (damaged) step of the j ob from the system and instigate the procedure for its replacement by a new part. In order to achieve its replacement, its JMA will simultaneously log a requ est for a processed part with the WOBA of all the s tages in the job’s process sequence preceding the stage wher e the breakdown occurred. Rush orders arriving at the system unexpectedly car ry a special “tag” indicating that they are high pr iority jobs. The functionality of the DA is modified sligh tly so that it initially checks whether there are a ny high priority jobs which it releases first before perfor ming its prioritising functions. Under pull product ion control, high priority orders are filled from the available inventory immediately by treating the time of their arrival as their due date and thus as the time they need to be released to the customers. If there is no sufficie nt inventory to satisfy a high priority order, a request will be logged with the last WOBA in the job’s sequence an d will be satisfied once its inventory is replenished. A canc ellation of order prior to the job’s due date will result in the removal of this job from the system and the termina tion of the life of its JMA. 5.0 E xperimentation Setting The simulated job-shop comprises 10 machines and is processing 10 jobs with diverse process routings a nd number of steps between 8 and 18. The time between the arrival of jobs follows an exponential distribu tion with µ=0.6 hours. Processing times are generated using a uniform distribution with min=3 min and max= 10 min. The due dates are calculated using the Total W ork Content Method and with the due date tightness coefficient set to 2. The dispatching rules conside red are: First Come First Served (FCFS), Shortest T otal Processing Time (STPT), Earliest Due Date (EDD) and Work Content in the Queue of the Next Operation (WINQ). In order to ensure the comparability of the output of the 8 simulation runs, three faults are introduced at time 5, 12 and 18 hours, with durations of 6, 4 and 5 minutes affecting the first, fifth and eighth workstation respectively. The probability damage is set to 100% for the case of workstations one and eight and 0% for workstation five. Two rush orders arrive at the sys tem at time 10 and 15 hours and one order is cancel led at time 22 hours. The system’s performance is evaluate d in terms of Mean Flow Time, Mean Time in Queue, Mean Absolute Deviation (MAD) of Earliness/Tardines s, Number of Tardy Jobs and WIP. 6.0 Analysis of Simulation Output and Conclusion In terms of the number of tardy jobs, the kanban-pu ll system performed better than the push system wit h the EDD dispatching rule producing the best results, Fi gure 4. However, in terms of the average number of jobs in the system at any time (WIP) the proposed kanban-pu ll system was significantly outperformed by the ini tial push system for all the dispatching rules considere d and with the best of its performance observed und er the WINQ rule, Figure 5. This is due the high levels of inventory maintained in the system to facilitate t he operation of the kanban-pull system and achieve a s atisfactory fill rate. 213 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION 0 5 10 15 20 25 30 35 No. of Tardy Jobs FCFS SIPT EDD WINQ Dispatching Rule Push Pull Fig. 4. Tardiness performance of push/pull systems 0.00 2.00 4.00 6.00 8.00 10.00 12.00 WIP Level (No. of Jobs) FCFS SIPT EDD WINQ Dispatching Rule Push Pull Fig. 5. WIP performance of push/pull systems As illustrated in Figure 6, with regards to mean ti me in queue the best pull performance was observed under the WINQ rule whereas the EDD rule produced the lea st MAD of Earliness/Tardiness. Kanban-pull produced the same output in terms of mean flow time for all the dispatching rules and was outperformed by the p ush system which produced the best output when the WINQ rule was employed. The consistent performance of t he kanban-pull system in terms of the mean flow time i s attributed to the way the pull logic is implement ed i.e. jobs in the available inventory are held until actu al demand releases them from the system at their du e date. 0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 5 [hours] Mean Time in Queue MAD Earliness/Tardiness Mean Flow Time WINQ Pull WINQ Push EDD Pull EDD Push SIPT Pull SIPT Push FCFS Pull FCFS Push Fig. 6. Time-related performance of push/pull syste ms Concluding, the employed agent-based simulation man aged the complexity and stochastic nature of the scheduling system efficiently by offering high leve l representation and performing well in terms of computational time requirements. References [1] T. C. Papadopoulou and M. Ozbayrak: “Leanness: Expe riences from the Journey to Date”, Journal of Manuf acturing Technology Management, Vol. 16, No. 7, pp. 784-807, 2005. [2] J. P. Kelleher: “JIT Application in the Job-shop En vironment”, APICS Conference Proceedings, pp. 348-3 51, 1986. [3] Sandras: “Job-shop JIT”, APICS Conference Proceedin gs, pp. 448-452, 1985. [4] D. J. Stockton and R. J. Lindley: “Implementing Kan bans within High Variety/Low Volume Manufacturing Environments”, International Journal of Operations & Production Management, Vol. 15, No.7, pp.47-59, 1 995. 214 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SC H ED U LING OF NON- R EP ETITIVE LEAN MANUFACTURING SYS TEMS UNDER UNCERTAINTY USING INTELLIGENT AGENT SIMULATION [5] J.-W. Li and D. J. Barnes: “Investigating the Facto rs Influencing the Shop Performance in a Job-shop E nvironment with Kanban-based Production Control”, Internationa l Journal of Production Research, Vol. 38, No.18, p p.4683- 4699, 2000. [6] G. A. Levasseur and R. L. Storch: “A Non-sequential Just-in-Time Simulation Model”, Computers in Indus trial Engineering, Vol. 30, No. 4, pp.741-752, 1996. [7] M. Gravel and W. L. Price: “Using the Kanban Job-sh op Environment”, International Journal of Productio n Research, Vol.26, No.6, pp.1105-1118, 1998. [8] M. Ozbayrak and T. C. Papadopoulou: “A Flexible and Adaptable Planning and Control System for an MTO S upply Chain System”, Robotics and Computer-Integrated Man ufacturing Vol. 22, pp.557-565, 2006. [9] T. C. Papadopoulou, A. Mousavi and P. Broomhead: “E xtending Lean Pull Production Control to the Dynami c Scheduling of HVLV Systems using Intelligent Agent Decision Support”, FAIM Conference Proceedings, pp. 498- 505, 2007. [10] T. C. Papadopoulou, A. Mousavi: “Performance Modell ing of Dynamic Lean Job shops with Basestock Shop-f loor Control using Intelligent Software Agents”, ICMR Co nference Proceedings, pp.185-190, 2007. [11] A. Baccalatte, A. Gozzi, M. Paolucci, V. Queirolo a nd M. Tamoglia: “A Multi-agent System for Dynamic J ust-in- Time Manufacturing Production Scheduling”, Proceedi ngs of IEEE International Conference on Systems, Ma n and Cybernetics, 2004. [12] T. Nishi, A. Sakata, S. Hasebe and I. Hashimoto: “A utonomous Decentralised Scheduling System for Just- in-Time Production”, Computer & Chemical Engineering, Vol. 24, pp.345-351, 2000. [13] D. Frey, J. Nimis, H. Worn and P. Lockemann: “Bench marking and Robust Multi-agent-based Production Pla nning and Control”, Engineering Applications of Artificia l Intelligence, Vol. 16, pp. 307-320, 2003. [14] T. C. Papadopoulou and A. Mousavi: “Control of Cons tant Work-in-Progress in Dynamic Lean Job-shops Usi ng a Multi-Agent System Approach”, International Journal of Agile Manufacturing, Vol. 10, No. 2, pp.19-28. [15] G. Liberopoulos, and S. Koukoumialos: “Tradeoffs be tween Base Stock Levels, Numbers of Kanbans and Pro duction Lead Times in Production-inventory Systems with Adv ance Demand Information”, International Journal of Production Economics, Vol. 96, No. 2 pp.213-232, 20 05. [16] G. Liberopoulos, and Y. Dallery: “A Unified Framewo rk for Pull Control Mechanisms in Multi-stage Manuf acturing Systems”, Annals of Operations Research, Vol. 9, pp .325-355, 2000. [17] http://www.agent-software.com/shared/products/index .html 215 216 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY Timo Halima Finland Tampere University of Technology, Pori Unit Abstract In a modern society, inside an organizational culture there is, or we are able to create a safety culture. However, in several cases safety culture is c onsidered too static, not being able to adjust its action to the external and internal c hanges of a predominant business environment. Therefore, we need active and dynamic m odels in order to respond to the changes of an organization, and especially to the saf ety risks seen in the future. Furthermore, the ultimate goal for an organization is to s uccessfully harness, organize and steer human beings toward the center of safety cu lture, and also toward an awareness of health, security and environment (HSE) within an organization. The current study derives from the lack of hands-on safety cultu re applications for managing this important organizational concept. The main purpose of the study is to find those safety performance drivers and characteristics, by which we are able to both understand and create a safety culture ontology in a long run . Secondly, the aim is to verify whether soft-computing is suitable technology in order to utilize constructed safety culture ontology in a dynamic business environment. Keywords: Safety, Safety Culture, Ontology, High-ri sk industry, Soft-computing. 1.0 Introduction Based on several safety reports, it is estimated th at every year organizations confront approximately 1 million accidents, which converted into euros stands for 1. 3-1.6 billion economic total costs to Finland [3]. Such economic figures illustrate that some organizations have faced severe deficiencies regarding safety pr ograms. But have some organizations achieved higher safety level than others? Furthermore, is it possible that organizations that are lacking a distinct safety cu lture program are able to learn from the pioneers? Based on the study by Juha Miettinen [7], the question is ho w to improve a safety culture within an organizatio n. According to that study, in order to reach affirmat ive safety figures, we have to enhance our safety p rograms. Already some eligible results have been gained thro ugh safety development processes, laws on industria l safety and training associated with an industrial s afety card. However, study notes that there is stil l a lot to do, mainly because a safety culture is a fuzzy concept and developing organizational structures have its o wn effect. 217 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY An organization's safety culture is seen as a way b ased on values and attitudes to carry out safety ma tters in an organization. It is noteworthy that all the related subsystems can be measured. In a safety culture su rvey an organization tries to clarify values and basic assu mptions, which have a great influence on improving safety in daily work. Furthermore, during development process es, engineers try to produce tools for management a nd personnel, by the means of which an organization ge ts its safety culture to a desired level. Since the benefits of technology in this field have not been entirely exp loited, novel kinds of theoretical frameworks, meth ods and empirical testing are required to clarify the influ ences of the development work of information techno logy, as well as the exploitable possibilities to capture ta cit knowledge (users). That is why this study thesi s aims to improve an organization's ability to create, use an d distribute new kind of knowledge in such a way th at this knowledge would be effortlessly available from a sa fety culture viewpoint. 2.0 Cultural Approach Organizational culture creates a basis for forming a safety culture. It consists of structures and con tents, which must be explored through humane, strategic and tech nical dimensions. Specifically, the technical dimen sion has become one of the important factors among organ izations operating in the nuclear, maritime and avi ation industries. Although, organizations have paid a lot of attention to technical safety, the researcher w ants to note that safety should be seen as a socio-technical mat ter, in which case also a social dimension has a si gnificant role in fostering an organization’s safety performa nce. Therefore, it is essential also to bring along side a humane approach, which attest that organizational c ulture is basically the personality of an organizat ion. By means of assumptions, values, norms and tangible si gns (artifacts), members of an organization will so on come to sense the particular culture of an organization. Furthermore, both understanding the culture, and b eing able to transform it are considered vital skills for man agement trying to achieve strategic outcomes. There fore, first an organization must create a good organizational c ulture, after which it is able to base the same rel ationship to create a sub-culture, which focuses on safety. 2.1 Safety Culture Safety culture is not merely a organization’s safet y program, policies and procedures, but the incorpo ration of safety into the informal and formal parts of the or ganization. Therefore, safety must be integrated in to every aspect of “the way of doing business”, requiring st rong commitment from its leaders, as well as showin g continuously that working in a safe manner and main taining a safe workplace are truly the core values within an organization. According to Shaw [12], especially through strong leadership an organization is able to ensure that the necessary support and training are always available, that is creating effective commun ication, providing recognition, actively both gathering inpu t and involving employees in decision-making, regul arly touring the plant, and attending safety meetings. A bove all, an organization’s safety culture must be seen as one of the boosters that gives appropriate priority to safety. Organizations have increasingly realize d that safety must in the first place be managed just like other business areas. Still, this concept confuses a lot of people. They often use this concept without underst anding the idea or the means of its possibilities. The safety culture and risks are often understood in a technic al way, in which case it mystifies people. Thereby, many people are unaware of the safety culture’s human si de. Safety culture includes all proper personnel at titudes and commitment to safety matters. This way, alongsi de with a technical approach, an organization is ab le to shape a good plan for a viable safety performance. 218 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY We are able to find from literature several differe nt kinds of safety ontologies. However, the researc her wants to stress that the safety culture concept has about as much definitional precisions as a cloud, which are at the same time both fuzzy and incongruent which each oth er. Therefore, it is common that some concepts are normally lacking something that other concepts migh t have. Although, there are lots of safety cultural structures, the core for safety consists of followi ng features [8]: • Culture dimension • Human behavior dimension • Management dimension Especially, when dissecting safety through a cultur al dimension, it can be divided into four main feat ures: organizational safety accountability, safety consci ous work environment, organizational learning and w ork planning [13]. Nonetheless, we must observe that no ne of these ontologies are exactly dynamic and vari able, but more static. Furthermore, they don’t gather inf ormation in any form, but are only considered as sp ecifiers, which are quite hard to exploit for organizations’ purposes. In order to aggregate, the researcher wan ts to point out that safety culture ontologies are hard to oper ationalize, since forming these ontologies will sta y on a conceptual level, rather than on a level based on m agnitudes or meanings. Therefore, the researcher st ates a question: “ Since just awareness of a safety culture is conside red inadequate, is it possible for an organization to operationalize and convert cultural precisions i nto dynamic intelligence, in which case readiness f or change through a continuous process of change can be executed ?” 3.0 Measuring Safety Performance through Cultural Dimension Organizations’ actions can be developed in several ways. It is essential to recognize development area s, as well to observe how they are able to succeed. One w ay for finding paths between target areas and ways of action is to create a self-evaluation tool for an o rganization’s purposes. [5] Organizational safety p erformance has been established on the foundation of “defense in depth”. This creates a basis for an organization ’s safety philosophy, through which ensuring its safety perfo rmance is guaranteed. Furthermore, it enables a pra ctical methodology for safety assurance, in which safety p rovisions are made in three completely different dimensions. The first level of safety addresses the prevention of accidents through the intrinsic desi gn features of the physical working environment. The second lev el, correspondingly, aims at providing reliable pro tection through effective devices and systems. The third le vel of safety supplement the first two through feat ures that add margins to the environment, as well as plant de sign to deal with events that are postulated to occ ur under extremely unlikely and foreseen circumstances. [11] To crystallize this notion, overall safety and its performance can be dissected through two different variables: attitudes, systems and environment. [10] Still, it is not possible to graft a safety culture onto an organization, as each organization being u nique, and the best safety systems failing without a supportive cu lture. Therefore, the researcher experiences that a n organization must create such a method or a tool, t hrough which it is able to measure attitudes, both personal and organizational, because of both crucially, affe cting the development of a safety culture in a work place. According to safety experts [10], the environment w here people work and the systems and processes in a n organization create a basis for a safety culture. E ach organization needs to consider all of these asp ects in developing and nurturing a safety culture that suit s the organization and the individuals within it. 219 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY More and more organizations are trying to measure t he intellectual capital, such as awareness of safet y, and value of human beings in the area of conscious and knowledge. Currently the success of organization de pends on how it is able to measure these two indicators i n order to effect improvements. To do this, manager s are anxious to find ways to measure the core competenci es of their employees. Furthermore, there is a coer cing need for organizations to learn to measure the infl uences of business and capital profit, as well as m anagement continuity, customer relations and the company’s in fluence on society itself. While measuring know-how, there is no single all-pu rpose indicator that is adequate to accurately port ray an organizations current and future level. The very na ture of field work and action create an impact of h ow an organization will measure and determine its know-ho w level. A few comprehensive indicators can give a quick impression, but this can hardly be considered to be waterproof. Therefore, it is appropriate that a wi de array of indicators are used simultaneously in order to get more precise and deeper knowledge giving picture of an organization’s entire business environment. Indicators based on intellectual capital measure co mpetencies of both the organization’s and its emplo yees’ capabilities. According to Rylatt [9], due to the f act that culture cannot be managed, nor owned, rese archers have engaged in extensive debate on whether culture should be experienced as capital. Nonetheless, sev eral reports claim that staff’s attitudes and presumptio ns regarding an organization and its customers dese rve particular attention. Many spokesmen feel that a so -called weak culture has a crushing impact on knowh ow and innovation in any team or business concept. Ryl att [9] shares this opinion, and remarks that in su ch situations everything gets more difficult in an org anization. It is little wonder he experiences that culture should be considered as measurable intellectual cap ital. Safety has traditionally been measured through stat istics associated with accidents, disasters and inj uries albeit subsequent to the incident. However, currently in a ddition to measuring external observable factors, organizations have created indicators, which share a purpose of measuring internal, psychological fact ors as well. In such kind of measurement organizations are first and foremost interested in reconciling emplo yees’ attitudes, skills and knowledge. Based on these res ults, the purpose is to construct methods, by which organizations pursue changes relating to attitudes and tacit expectations. The figure below illustrate s the measurable factors of a safety culture. [2] Unwanted events Working environment Organization R e a ct iv e Pro ac t iv e Safety Management System R e a ct iv e Pro ac t iv e Fig. 1. The Measurement of Safety Performance [2] 220 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY Unfortunately many indicators merely measure factor s based on external characteristics, a tendency whi ch is common with reactive organizations. Although, the r esults of quantitative indicators are easy to follo w, there is a possibility that while following statistics that represent accidents that have already occurred, an organization will never achieve a proactive stage. The common as sumption is that statistical data from every advers e incident can be used to improve safety performance, but this means that an organization does not focus on factors which relate to avoidable incidents. Henttonen [2] stresses that it is inadequate to eva luate organizational safety only by the means of ab errations, because such factors are considered flawed in chara cterizing the true stage of organizational safety. Therefore, an organization must use qualitative indicators alo ng with quantitative ones. There is indeed a clear goal to shift from reactive measuring to proactive measurin g, in which case along with measuring undesired eve nts an organization must create indicators that are capabl e of measuring both the working environment and the organization itself. [2, 8] To simplify, Henttonen [2] has outlined safety measurement into four diffe rent segments as follows: • Technical systems • Safety culture • Management systems and courses of action • Undesired events Every organization must choose applicable indicator s based on its own needs. Measuring safety related to technical systems can be carried out through risk m anagement, malfunctions and usability. Safety cultu re, on the other hand, accentuates both measuring both att itudes and an organization’s climate. By measuring the amount of training, audits and safety initiatives, an organization is able modify its management syste ms and plan courses of further action. Correspondingly, me asuring undesired events is a backward looking eval uation associated with aberrations and dangerous situation s. Due to this, Henttonen [2] indeed emphasizes tha t while fostering organizational safety, it is inappropriat e to just react to aberrations. He notes that to im prove management systems it is more rewarding to foresee aberrations and risks. 4.0 Results The purpose of the current study was as two-fold: F irstly, to accentuate the driving forces and compet ences behind a safety culture that management must observ e in order to create a strong safety culture. Secon dly, to create a novel kind of support system in order to f acilitate management’s tasks in their decision-maki ng processes. In current study the researcher has give n rise to results of creating, elaborating and pilo ting a safety- based evaluation system, the Bicorn Co-Evolute Syst em, such that an organization’s management has an effective and proactive tool in order to improve le arning and knowledge creation processes. This decis ion methodological research aims at modeling system cal led Bicorn. This application prototype focuses on creating, exploring and developing a knowledge-inte nsive safety culture for the nuclear power sector. Additionally, the existing computer applications ha ve been adapted to be applicable to safety concepts in general in industry. 221 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY Bicorn is intended to operate as an indicator, a su pportive tool to give weak signals for managers to their decision-making processes. The researcher tried to gather specific information relating to literary re views of safety cultures. This database, containing approxim ately 119 statements is used to evaluate an organiz ation’s state of the environment including both learning an d the knowledge creation of safety culture. The Bic orn- application has three different levels: a practical level, a system level and a meta-level. Features, which are presented in table 6 cover the organization’s inter nal and external targets for development and charts of improvements. According to Vanharanta [14], this ki nd of methodology allows an individual’s own evalua tion to focus on safety-related issues. The researcher n otes that by involving the whole organization in a collective effort to be aware of a comprehensive safety cultur e, it gives a throughout perception of the organiza tion’s goals. Thereby, by focusing on safety-related devel opment throughout the organization, it can enhance its core competences and turn safety activities into a compe titive advantage, i.e. added value process. Following the evaluation of 119 statements, the Bic orn-application’s goal is to convert a practical bo ttom-up view of different classification to the meta-level. The following table (Table 1) illustrates the cont ent of a viable safety culture, which enable the responsive environment for both learning and knowledge creatio n processes of safety culture. Furthermore, Table 2 i llustrates the system level ontology, which embodie s those systems, which are considered to be vital in the pu rsuit of developing an organization’s environment t oward a more responsive safety culture. Table 1: The Created Safety Culture Ontology Table 2: System Level Ontology 222 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY This kind of analysis has been made through the use of fuzzy sets. Fuzzy logic is a subset of conventi onal (Boolean) logic that has been extended to handle th e concept of partial truth - truth values between completely true” and “completely false” [6]. Generally speakin g, the principles guiding of fuzzy logic as follows : a set of input data from an array of sensors are fed into th e control system, after which the values undergo a process termed as “fuzzification”, which converts the discr ete values into a range of more specific values. Fu zzified inputs are evaluated against a set of production ru les. Whichever production rules are selected will g enerate a set of outputs. Output data is “defuzzified” as dis tinctive control commands. [1] With the help of Figure 2, we will show how the ind ividuals saw the system’s features from system repo rt. Fig. 2. Competences Indicated by Bicorn-Application . As Figure 2 illustrates, the system exhibits two di fferent results. The first one, the darker element, represents current state, and the second, the lighter one, rep resents the desired level. The difference between t hese two is considered to represent the development potential ( cf. creative tension). By means of these 119 statem ents and four maintaining systems we have been able to descr ibe an organization’s state as a responsive environ ment for learning and knowledge creation. Also this meta -classification facilitates the decision-making pro cesses of organization’s management to cope with a turbulent environment, while at the same time getting specifi c information from a systematic and value-added viewp oint. There is evidence that suggest that by systematically and continuously testing individuals within an organization, it enables a process of ch ange and an awareness of possible problem areas. Bicorn has so far only been tested in one organizat ion. The first test was conducted by students as a laboratory case at Tampere University of Technology’s Pori Uni t in Finland. This laboratory test was organized wi th 223 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY students analyzing an organization’s safety culture . The second test, which will hopefully give the tr ue indicators, will be carried out in near future. The researcher’s purpose is to gather some real-world cases. The first test results have been positive, in that they have shown that Bicorn has the potential of giving organizations several indicated competences concern ing the development of a responsive environment for learning and knowledge-creating safety culture. This application was first tested in cooperation wi th our data base administrator concerned with verif ying the technical functionality and validity of Bicorn. The results were effective and showed themselves to be reliable in accordance with Cronbach’s alfa value. According to Jussi Kantola [4] alpha’s coefficient values va ry from zero to one. He remarks that a value closer to zero indicates a greater diversity of values. The diver sity indicates a more random range of statements, wherea s the closer the value is to one the more consisten t are the statements. According to recent studies, values bet ween 0.6 and 0.80 represent values, which indicate a high degree of convergence toward statements. The alphas have been calculated within every application of t he Evolute-platform, and they all fall between the rec ommended guidelines. However, the researcher questions whether an alpha value is a sufficient measure in validating the sta tements of a larger perspective. Furthermore, is it enough to assess an organization’s safety culture only thr ough the alpha value, or do we need a more meaningful tools to analyze the validity of statements? However, in the absence of other tools for validation and for the p urposes of this thesis, we have to rely on the alph a value theory. After verifying functionality and validity, we formed a test group of 8 individuals. The resul ts were extremely clear and showed no need to remold the Bi corn-application. However, with the test group’s si ze being so small, the results can only give a directi onal sense and therefore cannot be interpreted too literally . Based on the first test results, we are extremely p leased to discover that this application can be use d as a common tool for management to evaluate an organizat ion’s capability to learn and create safety knowled ge. These results give us the possibility to identify t he important characteristics, properties and creati ve tensions, which can lead to the successful planning and impro vement of safety culture in high-risk industries. Furthermore, this evaluation system is considered t o be a consultative application, by which an organi zation is able to develop a more responsive environment for l earning and creating information related to a viabl e safety culture. However, this kind of experiment or resear ch requires a larger scale, scope and diversity of organizations to be examined. Therefore, our inclin ation is toward developing this application in co-o rdination with case-companies. We also hope that someday we w ill have the possibility to incorporate neural nets (Self- Organizing Maps/SOM) into analysis in order to faci litate management’s resources for allocating target s for development. The early results based on SOM have in dicated that by using these methods, we are able to structure, analyze and visualize large amounts of m ultidimensional data, regarding the responsive envi ronment from a safety culture viewpoint. 5.0 Discussion and Conclusion In recent years organizations have witnessed a grow ing concern over the issue of a safety culture with in the nuclear power industry and other complexes, and hig h-risk industries. Many reports claim that a safety culture must be static, pervasive, understood and practiced throughout an organization by its members. Further more, 224 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 SAFETY CULTURE CHANGE IN A HIGH-RISK INDUSTRY we are able to confirm that it should be considered to be a continuous process. Both organizational le arning and systems thinking have recently been linked with an ordered safety culture to enable to model, anal yze and engineer just like physically complex systems. Thro ugh these models, organizations are able to utilize improvements in both risk management and evaluation processes, which can lead to important competitive advantages. Additionally, there has also been in re cent years a tendency towards an effective manageme nt process in order to control and prevent risks assoc iated with health and safety in an organization’s environment. However, many organizations have also noticed that there are some limits to what can be achieved simply through hardware and technology. Th erefore, from an organization’s point of view, ther e is a proactive vision to create a supportive and respons ive organizational culture through information tech nology for people working for the benefit of organizationa l success. According to this discipline, the advantages brough t by the technology have not been implemented prope rly. Therefore, novel kinds of theoretical frameworks, m ethods and empirical testing are required to clarif y the influences of the development of information techno logy and the exploitable possibilities to capture t acit knowledge (users). The researcher has highlighted a new methodology to utilize existing theories in or der to better understand the interaction between humans, t echnology and culture. The ultimate goal is to reco ncile that everything must both be in harmony and in acti ve development such that security is optimized and security-related concerns are minimized. Thereby, w e can create a safer future. The study is in most part considered to be theoreti cal and therefore a new safety culture framework ha s been created. The reason for opting for a literature bas ed approach, was the researcher’s ambition to get a clear picture regarding dimensions of a safety culture, a s well as to see from the literature the new possib ilities to create a framework where the new technology can pla y an important role. References [1] California Agricultural Technology Institute, 2000. http://cati.csufresno.edu [2] Henttonen, T. 2000. Turvallisuuden mittaaminen. TUKES-julkaisu 7. Turvatekniikan keskus. Helsinki . [3] http://tyosuojelupiiri.net/tilastot (read 1.3.2008) [4] Kantola, J., Paajanen, P., Piirto, A. & Vanharanta, H. 2004 . Responsive organizations with genius management applications. Conference paper. EURAM 2004. [5] Koistinen, P. 2005. Itsearviointi turvallisuukulttuurin kehittämisen vä lineenä . Verkkolehti. eYtimekäs 2/2005. [6] Lotfi, Z. 1965. Fuzzy Sets . Information and Control, vol.8., pp. 8: 338-353. [7] Miettinen, J. E. 2002. Yritysturvallisuuden käsikir ja, Kauppakaari Oyj. [8] Ruuhilehto, K. & Vilppola, K. 2000. Turvallisuuskulttuuri ja turvallisuuden edistäminen yrityksessä . TUKES – julkaisu 1/2000, Turvatekniikan keskus, Helsinki. [9] Rylatt, A. 2003. Aineettoman pääoman mittaaminen . Artikkeli. Yritystalous 5/2003. [10] SafetyLine 2007. What is a safety culture? Work Safe Newsletter, September 2007. [11] Saji, G. 1991. Total safety: a new safety culture to integrate nuc lear safety and operational safety. Nucl Safety 32/3 (1991), pp. 416–423 [12] Shaw, M.C. 2005. New Approaches to Establishing a Safety Culture Or ientation in the Workplace. AWCBC, Public Forum, Vancouver 2005. [13] USNRC 2005. NRC Safety Culture Initiatives . Public Stakeholder Meeting 2005. [14] Vanharanta, H. & Pihlanto, P. 2001. A New Theoretical Framework to Support Project Mana gers’ Personal Mastery. First draft report. 225 226 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACTURING INDUSTRY B. Jones, I. Jenkinson, J. Wang Liverpool John Moores University, Liverpool, UK. Abstract This paper has been written to give a methodology of applying delay-time analysis (DTA) to a maintenance and inspection department. The aim of this paper is to give a brief overview of DTA together with its uses as a tool to achieve cost effective inspection maintenance. This is achieved by reducing downtime of plant i tems or reducing maintenance and inspection costs by removing unnecessary inspe ctions. A case study of a company producing carbon black has been included to demonstrate the proposed methodology. Keywords: Maintenance, Inspection maintenance, Dela y-time analysis. 1.0 Introduction to Delay-time Analysis Concept Delay-Time Analysis (DTA) is a concept whereby the time h between an initial telltale sign of failure u and the time to actual failure can be modelled in order to establish a maintenance strategy. Delay-time is the period of time when inspection or maintenance could be car ried out in order to avoid total failure. Figure 1 illustrates the delay-time concept (Christer and Waller (1984)) . Fig. 1. The delay-time concept 2.0 Methodology In order to develop a maintenance model using delay -time analysis a methodology needs to be developed in order to give the process a framework. Delay-time a nalysis can be used as a tool for reducing the down time, D(T) (Christer et al. (1995)) of a machine or a piece o f equipment based on an inspection period T, given the probability of a defect arising within this time fr ame b(T) . For a particular plant item, component or series of 227 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY machines, delay-time analysis is useful because the equipment in question is generally high volume and high capital expense, therefore any reduction in downtim e due to breakdown or over inspection can be benefi cial. As with the modelling of downtime per unit time, it is also possible to establish a cost model, C(T) (Leung and Kit-leung (1996)), again based on an inspection per iod T and probability b(T) , this model estimates the expected cost per unit time of maintenance. This mo delling has also been used for safety criticality ( Pillay and Wang (2003)) on a fishing vessel giving safety crit icality of a failure and operational safety critica lity. A methodology for applying delay-time analysis is p roposed as follows: • Understand the process. • Identify the problems. • Establish data required. • Gather data. • Establish parameters. • Validation of the delay-times and the distribution. • Establish assumptions. • Establish a downtime model D(T) and cost model C(T) . When the probability distribution function of a del ay-time f(h) follows an exponential distribution, i.e. when the failure rate λ or 1/MTBF is constant over a specified time period , the distribution function, as shown in equation (1), is used to calculate the probability of a defect arising b(T) : hehf λλ −=)( (1) The probability of a defect leading to a breakdown failure b(T) can be expressed as follows in equation (2). dhhf T hTTb T )()( 0 ∫    − = (2) Combining the distribution function f(h) into the breakdown failure probability b(T) this gives: dhe T hTTb h T λλ −∫    − = 0 )( (3) This term can be further simplified as: ( ) dhehT T Tb h T λλ −∫ −= 0 1)( (4) It is important to note that b(T) is independent of the arrival rate of a defect per unit time (k f) but it is dependant on the delay-time h. 228 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY 2.1 Downtime Model D(T) It has been demonstrated (Leung and Kit-leung (1996 )), (Pillay et al. (2001)) that with establishing a probability for breakdown failure b(T) it is also possible to establish an expected downt ime per unit time function D(T) as shown in equation (5).     + + = dT dTTbkd TD bf )()( (5) where; d = Downtime due to inspection. kf = Arrival rate of defects per unit time. b(T) = Probability of a defect arising. db = Average downtime for a breakdown repair. T = Inspection period. Substituting b(T) from equation (4) into equation (5) gives:       +     −+ = ∫ − dT ddhehT T Tkd TD b T h f 0 )(1 )( λλ (6) 2.2 Cost Model C(T) Similarly, given the cost of inspection Cost i, the cost of a breakdown C B and the cost of inspection repair C IR , the expected cost per unit time of maintenance of t he equipment with an inspection of period T is C(T) , giving the equation: [ ]{ }[ ] )( )(1)()( dT CostTbCostTbCostTk TC iIRBf + +−+ = (7) where; C(T) = The expected cost per unit time of maintaining the equipment on an inspection schedule of period of time T. Cost B = Breakdown repair cost. Cost IR = Inspection repair cost. Cost i = Inspection cost. The cost of an inspection is shown in equation (8). 229 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY Cost i = (Cost ip + Cost d) Tinsp (8) where; Cost ip = Cost of inspection personnel per hour. Cost d = Cost of downtime per hour. Tinsp = Time taken to inspect. The cost of a breakdown is calculated as the cost o f the failure plus the costs of corrective action t o bring the equipment back to a working condition. The details of a breakdown repair are shown in equation (9). Cost B = (M staff + Cost d) (T insp + T repair) + S p + S e (9) where; Mstaff = Maintenance staff cost per hour. Trepair = Time taken to repair. Sp = Spares and replacement parts cost. Se = Special equipment / personnel / hire costs. The cost of an inspection repair is somewhat identi cal to the breakdown repair cost apart from the fol lowing: • Inspection repair will not generally have equipment hire costs ( Se). • The time to repair will be of shorter duration for inspection repair. The time for an inspection repair having a shorter duration is mainly due to a breakdown having a grea ter knock-on effect. The equation for inspection repair is shown in equation (10). Cost IR = (M staff + Cost d) (T insp + T repair) + S p (10) A point to note regarding the cost model C(T) (equa tion 7) is that it describes a worst case scenario. This worst case scenario is a fault leading to failure b efore an inspection takes place or a fault being de tected at inspection. Conversely, a best case scenario would be no failure taking place before inspection and no fault being present at inspection. 3.0 Case Study In order to demonstrate the above models for downti me D(T) and cost C(T) a case study of a factory producing carbon black in the UK is given. This particular process of creating carbon black is made up of three units A, C & D. The three units c over the whole process stream from the reactor, MUF (Main Un it Filter) which collects & separates the product f rom the gasses produced and conveying of the carbon bla ck into storage containers. A low pressure air and natural gas produce a flame of high temperature (1500 degre es centigrade) in the combustion zone of the reacto r. 230 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY Heavy oil, which is known as feedstock, is sprayed into the flame and the carbon black reaction occurs . After the feedstock is exposed to the high temperature it is quenched with water in order to stop the carbon black formation reaction. At this point the basic form of carbon black is formed, carbon black powder. The f ilter is a bag type filter measuring approximately 10cm in dia meter and 2.5m in length. The cost of a filter is a round £28 each with a life expectancy of three to four ye ars. There is a second manufacturer of the filter t hat has a cost of around £7.50 but it has a life expectancy o f between 12 to 14 months with a lower tolerance to acid than the more expensive filter. 3.1 Costs of a Failure When a filter bag is to be changed the compartment has to be closed down. This requires 8 hours of coo l down followed by a period of 6 and 24 hours downtime for repair and replacement then a further 2 hours to w arm the unit back up, if a total re-bag is required dow ntime is generally around 7 days. When a unit is br ought off- line it continues to burn gasses in order to keep t he temperature in the reactor constant thus wasting energy. Also the system allows any energy created can be us ed by the facility and any surplus energy is sold b ack to the national grid, therefore any downtime can be co stly in respect of not just wasting energy but also potential income from surplus energy. Sometimes specialist ma intenance crews need to be brought in to deal with the problem. A typical example of a breakdown which too k 7 days to repair and replace all bags is demonstr ated below. • Loss of production per hour: £1,500 • Burn of gasses per hour: £238 • Loss of export of energy per hour: £26 • Cost of maintenance personnel per hour: £28 • Cost of supervisor per hour: £36 • Cost of replacement filters (205): £40,180 • Jetting crew: £710 • Jetter hire: £300 • Cherry picker hire: £2,500 This gives a total cost for a breakdown resulting i n 1,435 filters being replaced effecting 1 MUF for a period of 7 days to be £350,794. 3.2 Establishing a Delay-time Analysis In order to establish a delay-time analysis for thi s example several parameters need to be known. The parameters used in this example are as follows: • The arrival rate of a defect, kf - 0.28 per day. • Mean time between failure (MTBF) - 3 years. • Downtime for an inspection, d - 0.1 days. • Downtime for breakdown repair, db - 7 days. 231 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY • Breakdown repair cost, Cost B - £350,974. • Inspection repair cost, Cost IR - £5,000. • Inspection cost, Cost i - £67. Applying the parameters to equation (6) it is possi ble to establish an inspection interval where a min imum downtime is of primary concern as illustrated in fi gure 2. D(T) Expected downtime per unit time 0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 1 3 5 7 9 11 13 1 5 1 7 1 9 21 2 3 2 5 2 7 29 31 33 3 5 3 7 39 41 Days D(T) Fig. 2. Optimal inspection period based on minimum downtime D(T). As illustrated in figure 2 the minimum inspection i nterval based on minimum downtime D(T) is 14 days. When the cost C(T) is of primary concern the optimu m inspection interval is 11 days with a cost of £94 0 as shown in figure 3. If the inspection interval was m oved to 14 days in line with minimum downtime the c ost would rise to £977 which is a nominal increase of £ 37. C(T) Expected cost per unit time £0 £500 £1,000 £1,500 £2,000 £2,500 £3,000 £3,500 £4,000 £4,500 £5,000 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 Days C(T) Fig. 3. Optimal inspection period based on minimum cost C(T). 4.0 Validation In order to analyse the effect of change to the res ults of D(T) and C(T) a sensitivity analysis was ca rried out on each model. The analysis varied certain input data by 5% and 10% resulting in the following. 232 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY 4.1 Validation of D(T) The optimal inspection interval remains very close to the original interval given an increase and decr ease of 5% and 10%. The sensitivity analysis for D(T) is sh own graphically in figure 4. Sensitivity analysis based on D(T) 0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 Days D(T) -10% - 5% 0% +5% +10% Fig. 4. A graphical representation of the sensitivi ty analysis. 4.2 Validation of C(T) A sensitivity analysis was carried out on the cost of an inspection repair and the cost of an inspecti on in order to analyse the effect of a change in the costs. The cost of an inspection repair and an inspection has been increased and decreased by 5% and 10%. The sensitiv ity analysis is shown graphically in figure 5. The optimal inspection interval remains very close to the origi nal interval given an increase and decrease of 5% a nd 10%. Sensitivity analysis based on C(T) £0 £ 1,000 £ 2,000 £ 3,000 £ 4,000 £ 5,000 £ 6,000 1 3 5 7 9 1 1 13 1 5 17 1 9 21 2 3 2 5 27 2 9 31 3 3 35 3 7 39 41 Days C(T) -10% -5% 0% +5% +10% Fig. 5. A graphical representation of the sensitivi ty analysis. 5.0 Discussion 233 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLICATION OF THE DELAY-TIME CONCEPT IN A MANUFACT URING INDUSTRY It has been demonstrated in this case study that an optimal inspection interval taking into account a minimum downtime D(T) of 14 days has been established using the delay-time analysis technique. Using minimum c ost C(T) as the criteria an inspection interval of 11 d ays with a cost of £940 was calculated. Current practice at the company is that of a weekly inspection interval involving a flame check and a cloth check. It can be argued that this inspection interv al could move to an interval of weeks but given the nature of the two inspection checks and the fact that it does not stop production, a weekly inspection interval appears reasonable. 6.0 Conclusion This case study looked at a company in the UK produ cing carbon black. This paper demonstrates the dela y- time concept for the use of minimising downtime and costs, setting inspection intervals to achieve thi s. Information was gathered from historical data as we ll as expert judgement, with parameters established from this information in order to develop the delay-time models. Acknowledgements The authors wish to thank Mr G.Wright and Mr A.Whit ehead for their kind help for providing data and ot her necessary information. References [1] Christer A.H. and Waller W.M. (1984). Delay-time mo dels of industrial inspection maintenance problems. Journal of the Operational Research Society. Vol.33. pp 401-40 6. [2] Arthur N. (2005). Optimization of vibration analysi s inspection intervals for an offshore oil and gas water injection pumping system. Journal of Process Mechanical Engin eering. Vol.219. Part E. pp 251-259. [3] Christer A.H., Wang W. and Baker R.D. (1995). Model ling maintenance practice of production plant using the delay time concept. IMA Journal of Mathematics. Applied i n Business and Industry. Vol.6. pp 67-83. [4] Leung F. and Kit-leung M. (1996). Using delay-time analysis to study the maintenance problem of gearbo xes. International Journal of Operation and Production M anagement. Vol.6. No.12. pp 98-105. [5] Christer A.H., Wang W., Choi K. and Sharp J. (1998a ). The delay-time modelling of preventive maintenan ce of plant given limited PM data and selective repair at PM. I MA Journal of Mathematics. Applied in Business and Industry. Vol.15. pp 355-379. [6] Christer A.H., Wang W., Sharp J. and Baker R.D. (19 98b). A case study of modelling preventative mainte nance of a production plant using subjective data. Journal of the Operational Research Society. Vol.49. pp 210-19 . [7] Christer A.H., Lee C. and Wang W. (2000). A data de ficiency based parameter estimating problem and cas e study in delay-time PM modelling. International Journal of P roduction Economics. Vol.67. pp 63-76. [8] Pillay A. and Wang J. (2003). Technology and safety of marine systems. Elsevier Science Publishers Ltd . Essex, UK. ISBN:0 08 044148 3. pp 149-164, pp 179-199. [9] Pillay A., Wang J., Wall A.D. and Ruxton T. (2001). A maintenance study of fishing vessel equipment us ing delay- time analysis. Journal of Quality in Maintenance En gineering. Vol.7. No.2. pp 118-127. [10] International Carbon Black Association (ICBA). (200 4). Carbon Black users guide. Safety, health and en vironmental information. [11] The environmental protection act 1990. Chapter 43. [12] Integrated pollution prevention control (IPPC) Euro pean commission. December 2006. http://ec.europa.eu/environment/ippc. [13] OREDA (2002). In offshore reliability data. 4th edi tion. SINTEF Industrial Management. 234 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS MODEL BASED DESIGN OF ECONOMY OF SCOP E AND SCALE PRODUCTION SYSTEMS Zihua Cui, Richard Weston MSI Research Institute, Loughborough University, Le icestershire, UK Abstract Increasingly often enterprises are required to realise econ omies of scope production, i.e. be capable of efficient re-programming & reconfiguration to realise multiple value streams with a common set of resources. Key business models used in the furniture industry are explicitly documented using an ISO Enterprise Modelling (EM) technique. This has enabled an analysis of impacts of product dynamics to be observed in the furniture industry and has informed the development of a specific case EM which documents the business processes of a particular company making ‘fixed furniture’. Also explained is how the specific case EM was used to helped stucture the design of number of ‘context dependent Simulation Models (S Ms)’. These SMs can be reused on an ongoing basis to inform ‘process’, ‘resource’ and ‘ product’ aspects of production systems design. In this way business benefits ar ising from alternative configurations of production systems can be predicted. This d eveloped modelling approach has enabled the relative performance (in terms of valu e generation, process costs & lead times) of economy of scope and scale production sys tems to be compared with counterpart systems capable only of economy of scale product ion. The modelling methods and concepts described promise model driven means of coping with possible future impacts of product variance and product volume variation, l eading to production system designs capable of efficient mass customisation of prod ucts. Keywords: Product dynamics, Economy of scope, Econo my of scale, Enterprise modeling, Simulation modeling and Production systems design 1.0 Introduction and Product Dynamic Manufacturing Enterprises (MEs) are complex and cha nge continuously [1, 2]. Therefore responding rapid ly and cost effectively are key features to survival a nd remaining successful in any business environment . Four types of Product Dynamics have been classified by C ui & Weston [3], namely: product (class, type and feature) variances ; production volum e variations ; product mix variances and new product introduction . This paper describes a new approach to using integr ated modelling techniques to create re-usable model s of production systems and to subject these systems in a virtual environment to impacts arising from produ ct 235 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS dynamics. Thereby candidate production system desig ns can be computer executed and design decisions ma de rapidly and effective to reduce lead-time and minim ize investment risks. 2.0 State-of-the-Art manufacturing Philosophy and Research Assumptions M ade According to Ben Wang, there are two aspects of res ponsiveness: 1) reducing the manufacturing cycle to meet the market demands; 2) shortening the product devel opment cycle to meet the market opportunities [1]. GT is a manufacturing philosophy which advocates simplifi cation and standardization of similar entities whic h include (parts, assemblies, process plans, tools an d instructions). The underlying idea of GT is to ta ke advantage of similarities that exist among items to increase effectiveness during production [5]. Grouping allows similar, recurring activities to be conducte d more efficiently, like part family scheduling. He re a part family is all parts in a family that require similar trea tment and handling methods. Efficiencies are achiev ed by processing the parts together, so that productivity can be increased by reducing set-up times, realisi ng benefits of part family scheduling, achieving improved proce ss control and by using standardized process plans [6, 7]. RMS is a manufacturing philosophy centred on being cost-effective and responsive [2]. This is because the machines in RMS are designed with an adjustable str ucture to enable the system to be scalable in respo nse to market demands. RMS principles seek to combine the high throughput of Dedicate Manufacturing Lines wit h the flexibility of Flexible Manufacturing Systems. There are some outstanding advantages which can ben efit industry: 1) the adjustability of RMS machines allo ws flexibility not only in producing the variance o f parts, but also the reconfiguration of the system itself. 2) The customized flexibility of RMS can lead to fa ster throughput and higher production rate [2, 8]. Figur e 1 was constructed by the authors to characterize econo my of scope production systems . This shows three cases of production system confi guration. Case 1 is not considered to realise economies of scope, because i t can only deal with one product ‘type’ within a ‘c lass’. The terms ‘type and ‘class’ are defined by Figure 2 . It follows that if significant production volume variation occurs in a single product Case 1 system then the e conomy of that system may come into question. Howev er if the volumes are sufficiently high then economies of scale can be realised by Case 1 systems. also assumed is that all three cases require design & build time recomposition B 1 multi-product realising systemp roduct f amily B A1 A2 A3 B2 C1 C2 value st ream B A1 A 2 A3 B1 B2 C1 C2 it is presumed that multi-product value stream gene ration can be realise in the following three ways process A resource system A process B resource system B process C resource system C A1 A2 A3 A1 A2 A3 B1 B2 C1 C2 B1 B2 C1 C2 A1 A2 A3 B1 process D resource system D B2 C2 C 2 A1 A2 A3 B1 B2 C1 C 2 process A1 resource system A1 A1 A1 process A2 resource system A2 A 2 A2 process C1 resource system B1 B 1 B1 process C2 resource system C2 C 2 C2 it is assumed that these two cases will require run-time re-configuration & re-progr amming Case 1 Case 2 Case 3 Fig. 1. Possible alternative configurations of mult i-product realization systems and related ‘flexibil ity’ assumptions to be tested 236 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS Case 2 systems can realise both economies of scope and scale because they can realise more than one va lue stream associated with a number of product types wi thin a class (or product family). The notion of eco nomy of scope comes from the fact that if production volume s fall for one product type within its scope it may be possible to compensate by an increased volume requi rement for another within the systems’ scope . Natu rally for this reason Case 3 production systems have the greatest scope and therefore have the greatest pote ntial (in theory) to realise economies of scope. Figure 2 ill ustrated possible impacts on production systems des ign arising from product dynamics. Furn iture Category - a class of furniture types such as cabinet, table or chair Furniture Type -a common type of furniture variant within a class such as a kitchen cabinet, a bedroom wardrobe or a dinning room cabinet Furniture Feature -which can differentiate a product within a type and category norm such a key constructional, dimensional or finish feature difference significant process differences are needed to realise different product categories and that such differences will likely require both re- configuration & re-programming of resource systems less significant process differences are needed to realise different product types and that such differences may require resource system re-configuration but often will only require re-programming relatively minimal process differences are needed to cater for different product features and it is likely that such differences can be catered for by resource system re-programming alone It is assumed that process differences will likely take forms of * changes in elemental activities comprising the pr ocess * changes in processing routes comprising the proce ss * changes in required resource competencies & behav iours * changes in rules & information needs governing ac tivity execution * changes in execution times (& therefore consumpti on) of resources It was also presumed th at multi-s k illed human resou rces can possess ‘funct ional’ & ‘cha nge’ competencies needed to realise multi- product flows ( swi tc hin g between furni ture varia nt s) a s required, but also in general they will operate more slowly an d less ac curately th at special purpose machi nes Fig. 2. Assumptions about causal links between prod uct and process change to be tested 3.0 Choice of Case Studies Company and Enterprise Modeling The furniture industry was chosen as a subject of s tudy, for which three prime business models are in use namely ‘Fixed furniture’ production; ‘Flat Pack fur niture’ assembly and ‘Custom furniture’ fitting. Fi gure 3 was constructed to illustrate some of the purposes, common features and a common products produced by this industry and indicates the predominant business mod els use to realise them. Figure 4 shows examples of two of a total of seven graphical modelling templates that were populated w ith generic furniture industry information so as to cre ate a part of a ‘Generic EM’ covering all three fur niture industry BMs. Subsequently this model informed aspe cts of ‘product variance’ in that industry and esta blished an explicit link to ‘production systems design’. Th e EM was modeled using standard CIMOSA modelling constructs [9].The generic ‘context diagram’ shown in Figure 4 illustrates how six generic actors work collaboratively to realize furniture; while the ‘in teraction diagram’ in Figure 4 shows how material, products and information and money are transferred between d omain processes (DPs) and domains (DMs When modelling it was observed that distinctive ‘structu re diagrams’ needed to be constructed for each BM t ype. The project report details how for each BM the vari ous DPs were decomposed into CIMOSA conformant Business Processes (BPs) and Enterprise Activities (EAs). By studying these structure diagrams it was 237 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS possible to gain explicit understandings of the bus iness context of working and to derive context depe nded information about the general sequential ordering o f BPs and EAs belonging to each DP. * standard ‘spaces’ (can standardise most furniture dimensions) * standard ‘people ergonomics’ (can standardise many furniture features) * functionality then aesthetics (in most kitchen furniture) * reliability & safety key k it chen * non-standard ‘spaces’ (exceptionally can standardise dimensions) * standard ‘people ergonomics’ * BOTH aesthetics & functionality bedroom dini ng room lounge office (a t home) * non-standard ‘spaces’ (exceptionally can standardise dimensions) * standard ‘people ergonomics’ * mainly aesthetics then functionality * non-standard ‘spaces’ (exceptionally can standardise dimensions) * standard ‘people ergonomics’ * mainly aesthetics then functionality * non-standard ‘spaces’ (exceptionally can standardise dimensions) * standard ‘people ergonomics’ * functionality & ergonomics important Area of Home Common Features Common Products Applicable Business Models * cabinets * tables * chairs * work tops * wardrobes * dressing tables * beds * chairs * cabinets * tables * chairs * sofas * occasional tables * chairs * book cases * computer desks * tables * chairs * shelving * flat pack (main) * fixed (main) * custom (minor) * flat pack (minor) * fixed (main) * custom (main) * flat pack (minor) * fixed (main) * custom (minor) * flat pack minor) * fixed (main) * custom (minor) * flat pack (fairly minor) * fixed (main) * custom (minor) value aesthetics & customisation Fig. 3. ‘Drivers’ for product dynamics realise furniture for the home materials suppliers manufacturers retailers customer builders & fitters users disposers DM1 DM2 DM3 DM4 DM5 DM6 materials supply process DP5 manufacturing process DP4 retail process DP3 moneyraw material paper work use process DP1 disposal process DP56 moneyunwantedproducts paper work money fixed or flat pack products paper work money fixed or flat pack products paper work assembly/fitting process DP2 money semi-raw material hand crafted products paper work paper work money - th is domain process ( DM2) or ‘Role’ is only required for BM2 4.0 Integrated Enterprise and Simulation Modelling in Fix Furn iture Assembly Section A detailed study was made of the sequence of operat ions used to assemble Farm House and Drop Leaf tabl e products. Those sequence part of BP71-2 Assemble Ca rcasses & Fit Component’, which is realized by the assembly section of a ‘fixed furniture’ Case Study Company. These types of table product are two of a total of 18 product types produced by the Case Company. BP71 -2 was decomposed into 4 EAs, which are illustrated in Figure 5. Each of these EAs was studied in deta il to develop groundwork knowledge needed to create SM1 which related to Farm House table assembly. Here de tailed studied centred on processing activities and routes, and on needed human and other resources to realise each EA, and on possible activity grouping into ‘ro les’ that can be assigned to Work Centers (WCs) incorpor ated into SM1. In the case of SM1, enterprise activ ity EA 7.1.2.1 is executed by WC1 Collect component; EA 7.1.2.2 is executed by WC2; EA 7.1.2.3 is executed by WC3; and finally EA7.1.2.4 is executed by WC4. F ollowing the same method, the assembly process of Drop Leaf tables was modeled based on the capture o f specific company information; the aim here was to use Fig. 4. Generic Enterprise Models (CIMOSA-context a nd interaction diagram) for the Furniture Industry 238 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS the same modeling approach to create SM2, which is illustrated by figure 6. Both SM1 and SM2 are econo my of scale systems. The next production system studied was a possible f uture configuration able to realize dual economies of scope and scale. Farm House and Drop Leaf tables ar e fed into the same simulated production system, modeled by SM3; such that they share one assembly p rocess with the same set of work centers that are resourced by common underlying human and technical systems. The screen shot of SM3 is shown in Figure 7. When comparing the simulation results of the single and multi-product assembly systems, it was decided that there are three main performance measures that shou ld be taken into account. Choice here was made bear ing in SM1 SM2 SM3 Fig. 6. Drop Leaf table assembly processes simulati on model Fig. 7. Economy of Scope configuration of multi-pro duct model Fig. 5. Farm House table assembly processes simulat ion model 239 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS mind the general aims and objectives of the authors , namely to assess: 1) Lead time spending in each m odel; 2) Utilization of machines and human resources; 3) Rev enue and Cost comparisons. These factors were calculated for different simulation runs and exampl e results are compared graphically in Figures 8, 9 and 10. 58.6 135.4 357.2 404.8 0 200 400 600 FH model DL model Multi-FHMulti-DL time in system In SM3, FH and DL table work items have to share th e available time of machines and human resources us ed to realise processing operations. For example in WC 5 Bench1, they share the vertical sander to get bot h types of leg sanded. Because resources are shared during FH and DL table production the utilization of share d resources is increased. The operation times for FH and DL are distinct hence in SM3 it was necessary t o distinguish the processes of the different product flows by using numeric labels. Figure 9 compares th e utilization of Bench1 (one of the WCs) which has co mpetencies assigned to carry out the operations nee ded to assemble under frames and fit under frame to tops. This illustration shows that in the multi-product m odel SM3, the utilization of both bench and human resour ces (associated to WC Bench1) will be higher than t hat in either SM1 or SM2. Revenue generation and Cost cons umption comparison between multi and single product models is illustrated in figure 10. These two figur es show that economy of scope systems can perform m ore cost effectively. With more product types the effec tiveness will increase. 0% 50% 100% 150% FH model DL model Multi model Bench1 Human resource 0 10000 20000 30000 40000 FH model DL model multi model Revenue Cost 5.0 RMS Embed in Economy of Scope-Component Based Model SM3 was therefore designed so that it demonstrated economy of scope phenomenon, i.e. by generating mul ti- products values with a common set of resources. SM 3 model experimentation led to the design of furthe r SMs Fig. 8. Comparison of Lead-time between economy of scale an d scope models Fig. 9. Comparison of unitization between economy o f scale and scope models Fig. 10. Comparison of profit between economy of sc ale and scope models 240 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS with an inherent capability to realize increased ec onomies of scope. This includes a study of Case 3 production system. Literature review states that th e manufacturing approach of RMS can in principle ad dress the industrial need for rapid & effective change. H ence when designing and building further SMs, RMS principles were combined with those of GT principle s to inform the design of a ‘component based’ simul ation model SM4. SM4 is illustrated by Figure 11. Experim ents performed with SM4 showed significant increase d in economy of scope production. In SM4, furniture ( table) products are grouped into parts, with respec t to similarities of their attributes and operations tim es. SM4 design was centred on the use of 3 differen t processing groups (or work centers) to which altern ative resource configurations can be assigned. Also the routes taken between processing groups by different products was made flexible by the use of label and visual logic programming facilities. In this way, SM4 desi gn was structured to enable computer execution and quantitative illustration of benefits and dis-benef its of alternative RMS and GT principles. Therefore it proved possible for different versions of SM4 to be subjec ted to similar product dynamic patterns of change, so that multi product systems can be visualized and optimal ly designed over a specific period of operation 6.0 Reflections and Conclusions process A resource system A process B resource system B process C resource system C A1 A2 A3 A1 A2 A3 B1 B2 C1 C2 B1 B2 C1 C2 A1 A2 A3 B1 process D resource system D B2 C2 C2 A1 A2 A3 B1 B2 C1 C2 process A1 resource system A1 A1 A1 process A2 resource system A2 A2 A2 process C1 resource system B1 B1 B1 process C2 resource system C2 C2 C2 Ca se 1 Case 2 Case 3 Fig 12 Testing approaches of assumptions. SM4 SM3 SM2 SM4’ Fig. 11. component based simulation model 241 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL BASED DESIGN OF ECO NOMY OF SCO P E AND SCALE PR O DU CTION SYSTEMS Figure 12 summarizes how the integrated EM and SM a pproach was used to systematically and quantitative ly model single and multi-product systems. SM2 experim ents were conducted to provide a benchmark against which a multi-product SM experiments could be compa red. This is because SM2 modeled a single value flo w (economy of scale Case 1) production system configu ration. Then SM3 was built to model a production system that can assemble both Farm House and Drop L eaf tables. Hence SM3 model is a (Case 2) dual economy of scope and scale production system. SM4’ is a development of the SM4 component based model and is currently being developed, so that a full Ca se 3 production system can be modeled. In this Cas e3 production system, product components that are comm on to both ‘table’ and ‘cabinet’ types are processe d with common resources to realise additional economies of scope and scale relative to SM4. Therefore the pro ject has case study tested a novel way of systematically designing and quantitatively predicting performanc es of economy of scope systems. When so doing it had cont ributed new understandings about: (A) How different types of ‘model’ can usefully charact erize the ‘business context’ of economy of scope systems. Characterization is made in respect to: ‘p roduct variance’ in that industry; creating text descriptions of different business models used by t hat industry; and by using an ISO EM technique to explicitly model the network of business processes used by the industry concerned. (B) How a specific case of EM can be used to structure and inform the creation of ‘context dependent simulation models’. The subsequent computer executi on of these SMs has generated alternative behaviors of both single product and multiple produ ct realizing assembly systems. This has allowed performance criterion of alternatively configured e conomy of scope and scale production systems to be compared (C) How generic reference models of economy of scope an d scale production systems, can help guide the design and testing of ‘economy of scope product ion systems’ and can inform and quantify benefits gained from applying GT and RMS principles . The use of new understandings generated by this pro ject might ultimately guide industry in regard to: investment planning in existing and new production systems; planning of new product introductions, e.g . into existing or new production systems; and planning an d scheduling of existing production systems subject ed to on-going product dynamics. Reference [1] Ben wang, Kerang Han, Julie Spoerre and Chun Zhang 1997, Integrated Product, Process and Enterprise De sign., Department of Technology, Southern Illinois, Univer sity USA [2] Y.Koren, et al., 1999, Reconfigurable Manufacturing Systems, Annal of the CIRP Vol. 48/2 19 9 9 [3] Cui. Z and Weston R.H.(2008),' Model Based Design of Economy of Scope Systems ', MSc Report, Loughborough University, UK. [4] Mason-Jones R, Naylor J B, Towill D R. (2000), Engineering the Leagile Supply chain- International Journal of Agile Manufacturing System. Vol 2, No.1, pp54-61. [5] S.Angra et al., 2008, Cellular Manufacturing-A time based analysis to the layout problem. Int. J. Production Economies 112 (2 0 0 8) 427-43 8 . [6] Apple,J.M., 1977, Plant Layout and Material Handling , John Wiley and Sons, New York. [7] Colin.M, Reha.U, Y.H.Y. (1995), Manufacturing Cells: A Systems Engineering View , Taylor & Fransic, London [8] R.G.Landers, B.-K.Min, Y.Koren, 2001, CIRP Annals - Manufacturing Technology , Volume 50, Issue 1 , 2 0 0 1 , Pages 269-274 [9] AMICE (1993). CIMOSA: Open System Architecture for CIM , 2 nd extended and revised version, Springer-Verlag, Berlin. 242 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAI R TRADE MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE Lena Dzifa Mensah 1 , DR Denyse Julien 2 , DR Richard Greenough 3 1. Department of manufacturing, Cranfield University, Bedford, MK43 0AL. 2. Department of manufacturing, Cranfield University, Bedford, MK43 0AL. 3. Department of manufacturing, Cranfield University, Bedford, MK43 0AL. Abstract Fair Trade has emerged as an alternative form of international tr ade supporting equitable and sustainable models of trade that benefits prod ucers, consumers, industry and the environment. Smallholder producers of cocoa in West Africa combine to give the region a 69 % share in the world cocoa product ion and have began taking advantage of the benefits associated with forming cooperatives and b eing Fairtrade certified. Ensuring fair prices for Fairtrade certified cocoa producers is m orally just but is it the best that can be obtained for the producers? Would moving up th e value chain of cocoa in developing countries yield better proceeds? What is th e role of Fairtrade in such an initiative? This paper reviews the different facet s of Fair Trade and attempts to answer the questions raised above. Keywords: Value chain, Fairtrade, sustainable, ethi cal consumption 1.0 Background The continent of Africa sits on a trove of natural resources. Economic commodities like Cocoa, Cotton, Coffee and Gold abound in Countries like Ghana, Cote d’Ivo ire, Nigeria, and Cameroon. While most of these resources are grown and extracted in Africa, it is predominantly processed and consumed in the Western world [1]. From fig. 1 , it can be deduced that 69% of the world productio n of cocoa comes from West Africa. Fig. 2 reveals a meagre percent share of West African coun tries in the world cocoa consumption. 243 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAI R TRADE The economies of most producer countries of cocoa s uch as Cote d’Ivoire and Ghana are heavily dependen t on this commodity for sustenance [2]. The Government o f Ghana for instance, has set up a development plan known as ‘Vision 2015’ towards achieving middle-inc ome status by the year 2015. The Success of this economic development plan depends on the expected e arnings from exports [3] like cocoa, Gold and fresh fruits. However, Ghana Export Promotion Council (GE PC) suggests that targets set in this vision are unachievable as long as Ghana continues to depend o n traditional (unprocessed commodities) exports. West Africa has experienced significant improvement over the years in the cocoa bean value chain howev er, the quantities of intermediaries and final products from cocoa exported from the region is still low. Cote d'Ivoire 38% Cameroon 5% Ecuador 3% Others 10% Nigeria 5% Indonesia 13% Brazil 4% Ghana 21% Malaysia 1% Fig. 1. The share of countries in total cocoa beans production (2005/2006) [4] The Food and Agricultural Organization (FAO) has fo recasted that the benefits of cocoa processing in a dding value will continue to be enjoyed mainly by the imp orting countries that are also consumers if produce r countries do not take up the challenge of value add ition to the commodity. Benefits gained moving up t he value chain are economic growth through higher retu rns on the commodities, reduction in unemployment, as well as future export earnings. North , Central & South America 35% Europe 50% Africa 3% Asia and Oceania 12% Fig. 2. The share of regions in total cocoa consump tion in 2004/2005 [5] Reviewing years from 2001 to 2005 shows a consisten t trend of cocoa processing in West Africa. Analysi ng table I reveals the quantities of intermediaries [l iquor, cocoa butter, cake, cocoa powder] exported b y West African countries whiles table II reveals the quant ities of raw cocoa exported by Ghana. 244 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE Table 1: Intermediaries of cocoa exported from Wes t Africa Table 2: Raw cocoa exported from Gh ana 2001-2005[5] 2001- 2005 [6] Year 2001 2002 2003 2004 2005 Crop Year Metric Tonnes Country Qty in ton. Qty in ton. Qty in ton. Qty in ton. Qty in ton. 2000/2001 389,772 Cote d'Ivoire 176172 188685 201264 227219 247989 2001/2002 340,562 Ghana 51778 45807 30361 43443 46312 2002/2003 496,846 Nigeria 8690 9494 10653 10718 12362 2003/2004 736,911 Cameroon 25514 22098 17874 16426 16978 2004/2005 600,000 Total 262154 266084 260152 297806 323641 Table 3: Percentages of processed cocoa exported from Ghana Crop Year % of Processed Cocoa Exported 2000/2001 13.3 2001/2002 13.5 2002/2003 6.1 2003/2004 5.9 2004/2005 7.7 Table 3 reveals a progressive decrease in processed cocoa exported in Ghana between the years 2001 and 2004 even though the crop year 2001/2002 experience d a slight increase over the previous crop year. Th e crop year 2004/2005 saw an increase in the export of pro cessed cocoa by a margin of 1.8% over the previous year. In the same year (2005), the government through the Cocoa Processing Company (CPC) embarked on a project to increase the quantity of processed cocoa within Ghana. Presently, the initial capacity of 2 5000 tonnes has increased to 65000 tonnes . The Ghanaian government through partnerships with t he private sector targets processing 300000 tonnes of the current production quantity of 700000 tonnes of coc oa beginning 2008. However, the ultimate government target is 80% [7].This increase in processing capac ity would yield benefits for the private sector oth er than the primary producers. From table1, similar processing patterns to Ghana a re derived for Cameroon but the quantity of cocoa processed in West African between the years 2001 an d 2005 increased by 24% as a result of significant increase in processing quantities for Cote d’Ivoire and Nigeria, yet gave West Africa a substantially small share of 14% [8] globally. 2.0 The Cocoa Value Chain The global cocoa value chain has experienced signif icant growth over the past decades and this has gen erated competitiveness among players in the cocoa sub-sect or. However, in some countries, the competitiveness in the value chain is threatened by inconsistencies in quantity and quality in production of the raw coco a [9], while others lack infrastructure, logistics and goo d policies [10]. 245 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE This is true for the major players in the cocoa mar ket - Cote d’Ivoire, Ghana, Indonesia, Cameroon and Nigeria even though the level of dependency of thei r economies on cocoa vary as a result of high diversification [2]. The cocoa value chain in most of the West African c ountries is comparable. The similarities arise from the extent to which they process while the differences are as a result of the level of involvement of gove rnment and private sector along the chain. Fig. 3 shows al l the players along the value chain of cocoa in Gha na. The government is involved from the farmer’s level to t he end of the value chain. Apart from the involveme nt of smallholder producers, private sector involvement i s only allowed after quality assurance. Moving up the value chain basically involves the ph ysical transformation of raw materials into manufac tured goods as well as improvements in the quality of raw and manufactured goods. Other researchers like Por ter [11], have the value chain spanning the transformat ion processes to the final product as well as the i nteractions that exist among those processes and any other exte rnal entity [12]. Contrary to happenings in Ghana, the trading in coc oa in other countries is not monopolised by governm ent thereby introducing fierce competition among collec tors and traders with few barriers for entry into t he market. The cocoa market environment differentiates little for quality [9] and hence the cocoa produce rs have little incentives to upgrade to more costly product ion and post-harvest practices which improve the qu ality of the cocoa bean produced.However, processors and man ufacturers have clear incentives to establish close r supplier relationships in order to improve the qual ity and consistency of their raw materials [9]. The suppliers have the responsibility to ensure that the strict s tandards of quality needed for processing and manuf acturing are adhered to. Some of the determinants of the price of cocoa incl ude the forces of demand and supply but at the poin t where competitiveness plays a role in who gets access to the market, other factors come into play [9] such a s availability of supply, fat content and flavor. A study conducted by USAID on some cocoa producing countries to determine their strength in terms of v alue delivery to the customer, revealed Indonesia produc es cocoa which is inferior in quality and fat conte nt yet there is demand for the commodity (especially from China). 2.1 Challenges to Moving Up the Value Chain A solid financial base, access to appropriate techn ologies, legal and policy frameworks (certification , standards, taxation, and tariffs), knowledge and sk ills, Education, research and training are invariab ly pre- requisite to moving up the value chain. Lack of a sizeable domestic market for products mad e from cocoa is another challenge to moving up the value chain. Policies to curb the over dependence of dome stic markets on imports of cocoa products could pro mote moving up the value chain in West Africa. 246 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE The unavailability of other raw materials required to manufacture cocoa products impedes moving up the value chain. Cocoa processors and smallholder producers in Niger ia are incurring higher costs (in tariffs) in order to access the market in the European Union (EU) following the nations’ refusal to sign the Economic Partnership Agreement (EPA) [14]. These high tariffs imposed on raw cocoa and even higher tariffs imposed on intermediary exports into the EU have the tendency to impede moving up the value chain. Fig. 3. Cocoa value chain in Ghana [13] In Ghana, only a small quantity of cocoa is manufac tured into chocolate due to the lack of logistics a nd infrastructural support for transporting the finish ed product abroad. In view of this challenge, attem pts are being made to channel efforts into the processing o f intermediaries (liquor, cake, cocoa butter and co coa powder) which are relatively easy to transport. Foreign multinational companies are vertically inte grated and so prefer to source raw cocoa as opposed to processed cocoa. This is because of the high invest ment made in technology acquisition and research an d development. This is a challenge to moving up the v alue chain in West Africa as these multinational companies have the buying as well as the negotiatin g power to influence what they purchase as well as the power of choice to source from other countries. 3.0 The Fair Trade Initiatives Fair Trade has emerged as an alternative form of in ternational trade to promote payment of fair prices as well as labour and environmental standards in areas rela ted to the production of a wide variety of commodit ies. Fair Trade ensures eligibility to buy, process, and sell fair trade products through: 247 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE • Standards • Certification of institutions • Labeling of products (Max Havelaar , Fairtrade, UTZ certified and MADE BY) • Strict monitoring • Smallholder producer participation in a democratic organization • Harmonization of the Fair Trade message • Attempts at harmonising the Fair Trade certificatio n mark. The certification authorizes producers and manufact urers to use the designated Fair Trade label on the ir products and guarantees consumers of commodities me eting labour and environmental standards [15, 16]. The Fairtrade Labeling Organization International ( FLO) collaborates with other organizations known as Alternative Trade Organizations (ATOs) to develop t rade systems that promote sustainable production an d trade with the disadvantaged workers and producers around the world. 3.1 Products and Markets The quest to make Fairtrade part of mainstream trad e has led to the expansion of the range of products . Currently, over 3000 products have been licensed to carry the Fairtrade mark [17]. Amongst these are f lowers, rice and those manufactured from the primary range of commodities such as chocolate, clothing, footbal l, jewellery and other ethical gifts. Fig.2 is a typical representation of the market ava ilable for cocoa producers in the developed world. Processors and consumers would want to buy cocoa at the cheapest price possible but with education and awareness of ethical issues, the market for Fair Tr ade is growing significantly. Europe and United Sta tes of America (USA) hold the largest share. In spite of t he fast growing nature of the market, there are sti ll lingering doubts among critics of Fair Trade as to whether th e Fair Trade market could reach a size large enough to have substantial impact on the standards in developing n ations [18]. 3.2 Price Determination for Fairtrade Products The minimum Fairtrade price is set at the average p roducers’ costs of sustainable production (‘COSP’) per product [17]. A global or regional minimum price is normally set; however when this is not possible be cause of the large variation in the costs of production o f cocoa in the different regions, a national minimu m price is set. Fairtrade is said to create distortions in the mark et as it attempts to help producers elude the full weight of the demand supply forces [19] leading to the over produ ction of Fair Trade goods [20]. It is worth noting that the market for Fairtrade products is limited hence not all Fairtrade products get marketed under the Fairt rade scheme. Fair Trade is also said to discriminate bet ween suppliers with identical quality [21] and in d ifferent 248 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE countries. However, Mann [20] argues that the attri butes of Fairtrade products are incomparable to att ributes of products in the conventional line of trade. Mann [20] further argues that even though the price mod els used in determining price for products may generate pric e differences among producing countries, the Fairtr ade price is set above equilibrium [17] and hence Fair Trade rests on market forces as much as the convent ional trading does. 3.3 Awareness Programmes Commencing almost half a century ago, Fair Trade di d not receive as much attention and patronage as it is receiving presently. Public interest in the product s certified by the Fairtrade Foundation has improve d through awareness creation and campaign for changes in the conventional model of international trade champione d by the UK and USA.In UK alone, over 300 ‘fair trade to wns’ have been created with over 300 companies licensed to sell Fairtrade products [17]. 2006 saw a rise in sales by 49% [17]. Awareness of the Fairt rade mark amongst UK adult population grew to 57% in 2007[16] . This percentage signifies more avenues for growth and expansion. A Fairtrade fortnight has been set a side to boost awareness of Fairtrade and the sale o f Fairtrade products in the UK. 3.3.1 Role of Institutions in the Growth of the Fair Trade Initiative Many institutions are buying into the concept of et hical consumption. The willingness and cooperation of both smaller and larger companies to participate in the Fair Trade initiative is one of the reasons underpi nning the success of Fair Trade. The K-2 Unit of TransFair USA for instance has crea ted a Fair Trade Curriculum to educate students on “The Journey of the Cocoa Bean from Farm to Fair Trade C hocolate” [22] teaching the concept of global interdependence in order to make students take resp onsibility for the impacts of their actions on huma n communities and the environment. 3.3.2 Involvement of Major Super Markets Major supermarkets in the UK like Sainsbury, Waitro se, Tesco and Marks & Spencer (M&S) have joined the chains of supermarkets to promote the Fairtrade ini tiative. M&S for instance has refused to sell any t ea that does not bear the Fairtrade mark and all tea sold b y Sainsbury now bear the Fairtrade certification ma rk. Asked whether the supermarkets would relent in thei r efforts to support the new model of trade, representatives for the supermarkets, pledged their unflinching support to marketing Fairtrade product s. These examples cannot be discounted as merely the practic es of an obscure and irrelevant group of progressiv e producers and elite consumers [15]. 3.4 The Fair Trade Debate 249 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE In the past decade, there were concerns among consu mers and civil society groups whether the Fair Trad e model could be extended to factory manufacturing an d clothing. Presently, the range of products covere d is wider through the efforts of Fair Trade activists. There are still lingering doubts about the possibil ity of extending the Fair Trade model to cover all manufac tured goods and the availability of enough ethical consumers to sustain the initiative. The possibility of retailers of Fair Trade products charging above the mandated prices of the products raises concerns. Following this, the questions are whether governments should get involved to devise alternat ive ways of fixing the prices of Fair Trade products [2 0] through regulations. In response, Lamb [23] argu es that the involvement of governments would dilute the sta ndards: a voluntary scheme owned by producers and driven by the public with integrity and credibility is appropriate. She added that maintaining neutral ity and rigour is the strength of Fair Trade. While others also argue that only a quarter of the earnings reach the farmers due to corruption along the supply network, Ronchi [24] reports of producers having ac knowledged receipt of all proceeds in addition to h aving the prerogative to decide on the community project to invest premiums in Ghana. The debate is not without questioning the credibili ty of producers and manufacturers. Critiques assert that products passing for Fair Trade prices may not have necessarily complied with the standards [25] of Fa ir Trade consequently defeating the aims of Fair Trade proponents. Proponents of Free Trade are challenging Fair Trade activists to produce hard evidence of what Fair Tr ade claims to achieve instead of relying on anecdotes [ 26].The evidence of the impact of Fair Trade is evi dent in the attestations of producers and the increase in t he number of applicants wanting to be Fairtrade cer tified. One of the greatest challenges of Fair Trade is the att ribution of impacts to factors like government policy among other [24]. This is further complicated because cri tics are often looking for the translation of the i mpacts into money in the pocket of producers. However, Fairtrad e supports the producer organization as well as the producer, giving the prerogative to producers throu gh voting rights to determine areas of investment [ 23, 24]. 4.0 Role of Fair Trade in Moving Up the Value Chain FLO International, together with other members of t he Fair Trade Movement, aims for the highest impact possible on disadvantaged producers and workers in developing countries [17]. Global analysis on the cocoa market prospects to 20 10 suggests a continual expansion of cocoa producti on, processing and consumption [27].The anticipated inc rease in consumption is as a result of efforts to b oost domestic consumption in the producing countries and the scientifically backed nutritive and cancer fig hting capabilities of cocoa-rich chocolates [27] and coco a products. 250 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE Moving up the value chain requires a steady supply of cocoa without which the anticipated rise in dema nd of cocoa may not materialize. Challenges to a steady s upply of cocoa are cocoa diseases and pest infestat ion. This is a typical instance where Fairtrade may make a difference as the Fair Trade organization arrang es for the education and training of smallholder producers on how to handle the pest and diseases associated with lower yields. Under the conventional model of trade, the forces o f supply and demand determine the price of the comm odity however, Fairtrade guarantees a stable minimum pric e to producers enabling them to plan long term, inv est and develop technical support for the future [15]. The minimum price goes to the smallholder producer while the premiums are used for community projects to enh ance the livelihoods of smallholder farmers and the ir dependants. The prerogative of what premiums should be used for is decided by members of the producer organization and so premiums go to address the domi nant needs of the smallholder farmers. The producer organizations under the Fairtrade sche me may be able to attract loans to establish grindi ng and manufacturing plants which would empower producers to develop the capacity to compete in the global ma rket place as well as the moving up of the value chain. This possibility is not farfetched as the Ghanaian producer organization (Kuapa Kooko) has been able to secure a part ownership of the Devine Chocolate Company in the UK and USA which has recorded a first dividend share this year [28]. If the concept of Fairtrade is supported and promot ed, the future could see producers having a share i n supermarkets, manufacturing plants, shipping lines, and eventually, the whole supply chain would be underpinned by the Fair Trade concept. The involvement of private sector in moving up the value chain is appreciated however; efforts could b e invested into making the value chain processes equi table so as to reinforce the Fair Trade effort of i mproving the livelihoods of disadvantaged producers in the s outh. 5.0 Conclusions In conclusion, moving up the value chain requires a host of resources, changes in policies and attitud es linked to access to finance. Fair Trade is not the only an swer to moving up the value chain however, it provi des a means through which marginalised producers who may not even be able to afford products manufactured fr om commodities grown by them, the opportunity of parta king in the returns from their hard work. The Fair Trade concept also empowers consumers in the West who may previously have not been able to interfere in trad e negotiations, the opportunity to better the livelih oods of disadvantaged smallholder producers in deve loping nations. Even though there may be a lot of debate about the claims of Fair Trade, the concept has the capacity to initiate value addition to commodities. 251 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MOVING UP THE VALUE CHAIN OF COCOA: THE ROLE OF FAIR TRADE The journey for economic liberation and poverty all eviation in developing nations would have to begin from somewhere and it is definitely not in the continual giving of aid, and grants. A better model of trade is preferable. References [1] IBM Global Innovation Outlook Report, 2007. Interna tional Business Machines Corporation. [2] World Facts Book https://www.cia.gov/library/publications/the-world- factbook/geos/iv.html accessed [3] Buatsi, N. S. (2002). Financing non-traditional exp orts in Ghana. Journal of Business & Industrial Marketing. Volume 17 (6) pp.501-522. [4] UNCTAD < http://unctad.org/infocomm/anglais/cocoa/market.htm > [5] Global Trade Atlas, annual data compiled from repor ting countries official statistics. [6] Food and Agricultural Organization accessed on the 26 th of February, 2008. [7] Koranteng, A. (2008). Ghana set to process 300000 t onnes of cocoa in 2008. The statesman. http://www.thestatesmanonline.com/pages/archive_det ail.php?newsid> accessed on 4 th of March, 2008. [8] Annual Report, United States International Trade Co mmission (2007).Sub-Saharan Africa: Factors Affecti ng Trade Patterns of selected Industries. < http://hotdocs.usitc.gov/docs/Pubs/332/Pub3914.pdf > accessed 6 th of March, 2008. [9] United States Agency for International Development. Indonesia cocoa bean value chain case study. www.usaid.gov accessed on 25 th January, 2008. [10] Cocoa Processing Company. < http://www.foodprocessing-technology.com/projects/c ocoafoodprocessing/ >accessed on 25 February, 2008 [11] Porter, M. E. Competitive advantage: Creating and S ustaining Superior Performance. The free press, a d ivision of Simson & Schuster Inc., New York. [12] World Bank Group, Report 2008.Moving towards Compet itiveness: A Value-Chain Approach. [13] Osei, I.,(2007) Sustainable Practices in the Global Cocoa Economy : A Producers’ Perspective. The 4 th Indonesia International Conference. [14] International Center for Trade and Sustainable Deve lopment. EU Clamps Down on Nigerian Cocoa. Weekly T rade New Digest, Volume 12,(27)2008 [15] Raynolds, T. L, Murray, D and Wilkinson, J. Fair Tr ade: The challenges of transforming globalization. Published by Routledge, Taylor and Francis Group. [16] Transfair USA< http://transfairusa.org/content/about/certification .php > Accessed on 15 th of February, 2008. [17] Fairtrade Foundation (2008). Fairtrade Explained. < www.fairtrade.org.uk> [18] Hiscox, J. M. (2007). Fair Trade as an Approach to Managing Globalization. The conference on Europe an d the Management of Globalization. [19] Sidwell, M. (2008). Unfair Trade. Adams Smith Institute, London. [20] Mann, S. (2007). Analysing Fair Trade in Economic T erms. Journal of Socio-Economics. Doi:10.1016/jsocec.2007.11.002. [21] Yanchus, D, de Vanssay, X. (2003). The myth of fair prices: a graphical analysis. Journal of Economic Education 34(3), 235-240. [22] Around, E. (2007).Focus on Fair Trade Curriculum. A publication by TransFair USA. [23] Lamb, H. What is the future of Fair Trade? New Consumer Magazine. Issue 35 , February, 2008.page 18-22 [24] Ronchi, L (2002). Monitoring the Impact of Fairtrad e Initiatives: A Case Study of Kuapa Kokoo and the Day Chocolate Company. [25] Renard, M. C. (2003).Fair Trade: quality, market an d conversions. Journal of Rural Studies . 19, 87-96. [26] House, R. and Trebilcock, J. (1996). The Fair Trade -Free Trade debate: Trade, Labour and the environme nt. International review of Law and Economics, 16:61-79. [27] COPAL Cocoa Info (2007).A weekly newsletter of Coco a Producers Alliance. Issue 241. [28] Divine Chocolate Ltd, USA http://www.divinechocolateusa.com accessed on the 16th of March, 2008. 252 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PROBLEM SOLVING CAPABILITIES IN IMPLEMENTING LEAN PROCESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY A.P. Puvanasvaran 1 ,M.H.M.A. Megat 2 ,S.H.Tang 2 ,Rosnah M.Y 2 , Muhamad M.R 1 , A.M.S. Hamouda 3 1. Faculty of Manufacturing Engineering, University T echnical Malaysia, Karung Berkunci 120 0 , Ayer Keroh, 754 5 0 Melaka, Malaysia. punesh@utem.edu.my 2. Department of Manufacturing Engineering, University Putra Malaysia, 434 0 0 Serdang, Selangor, Malaysia 3. Mechanical and Industrial Systems Engineering, Qata r University, Doha, Qatar. Abstract Globalization and customers’ expectations for higher quality ser vices and products with competitive cost requires OEM’s (Original Equipment Manufactures) in Aerospace, Automotive and Electronics Industries face various challenges such as fast deliveries when needed. Higher cost of operations is puttin g heavy demands on supplies to provide solutions with higher quality and reliabi lity in order for them to be competitive and profitable. Supplies and solutions providers ar e challenged to improve their own operating cost and performance more effective and efficiently through strategic initiatives such as Lean Process Management ( LPM) to driven wastages and losses in their manufacturing, administration and e ntire supply chain processes. It can be successfully implemented in any organiz ation if be there is good commitment from top to bottom management. Human factor plays an imp ortant role in ensuring lean process management to be successful and pr ovides good proposition for the success of the organization in the long run. One of th e main elements of people is their problem solving capability (PSC) in identifying and eliminating wastages. So, this paper presents the conceptual framework and focuses on how this people development system can help organizations to enhance employees’ capability in identifying and eliminating wastages continuously and effectivel y. By means of a case study that discusses the people development system (PDS) framework, the strategy used by the aerospace company for implementing lean process man agement, and the significant benefits that were accrued in manufacturing ope rations and meeting 253 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY company goals by identifying and eliminating wastages. The data for th is study were obtained through actual monitoring of a pilot study in aerospace man ufacturing company. The pilot study was conducted to validate the people de velopment system framework; and the results of the assessment were satisfyi ng and many improvements have been done to achieve company goals by identifying and eliminati ng wastages. The obstacles, benefits, future work and improvement on impl ementing the people development system are also discussed. Keywords: Lean Process Management, Problem Solving Capabilities, People Development System. 1.0 Introduction In today’s competitive world, no company can afford to waste all form of resources. The most underutil ized resource of most manufacturing company is their peo ple. The number one asset of any organization is al so its people. In fact, people are one of the few apprecia ting assets an organization has second to products or services. The real advantages of employee’s involve ment are to focus a group of employee s with different perspective on a single objective that support the organization’s strategic focus. The companies that develop and leverage the capabilities of all their employee s will achieve continuously better performance than those that do not. The companies that fail to unlock the potential of their workforce will be forced to carr y more overhead, have more layers of management, and will be slower to react to changing business, social and political needs. Lean process management definitely becomes their arms to fight to achieve this goal w ith the employees’ problem solving capability in eliminatin g wastages. 2.0 Background of Study Many industries ranging from aerospace to service a nd health businesses have successfully adopted lean manufacturing practices and principles and made the ir operations more efficient. Most of these compani es are still not what can be considered fully lean organiz ations [1]. In order to become fully lean, a compan y must understand lean as a long-term philosophy where the right processes will produce the right results and value can be added to the organization by continuously de veloping people and partners, while continuously so lving problems to drive organizational learning [1]. Whil e there is no exact definition for a fully lean org anization, it is important that an organization must understand a nd apply all of the practices and principles. It is also important to understand that lean thinking, which a ffects the whole business model, is the key and not solely leaner production, where only parts of the whole le an philosophy are applied. The goal of becoming a fully lean organization can only be reached if the employees are well aligned w ith the new philosophy. Gagnon’s [2] work suggests, “produc tion employees who are not well aligned with a philosophy will exhibit lower levels of desired att itudes and behaviors”. Since lean thinking requires a great level of employee’s involvement and change in attit ude and behaviors [2], strategic employees alignmen t plays an important role in the quest to become lean. To e nsure employees alignment, it is particularly cruci al to have 254 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY open, honest communication, and delegation of autho rity. Spear [3] argues that these are necessary for a successful lean implementation. Emiliani [4] defines repeated mistakes as another p rimary type of waste and argues that a business tha t is unable to learn and change its behavior will, “no d oubt, risk the future existence of their entire ent erprise as currently governed”. In a lean company, learning co ntinues, since “lean is a continuum and not a stead y state” [5]. Although lean thinking is a buzzword, the lean philosophy, practices and principles offer industr y a potential mechanism for improving performance. 2.1 Objective of the Study A scientific approach is needed in order to engage people from Top to shop floor personnel to solve an d improve the problem solving its source. Every probl em is an opportunity to improve the process and to develop people. In this research it was shown that people is only key or primary factor to make decisi on to make changes to drive the strategic initiatives of the organizations to be world class or be the best. As such it is important to consider the best way to overcome t he interruptions and hiccups in the processes. Ther e is a need for a common approach to problem solving capab ilities across the organization and a common langua ge for communicating the diagnosis and the results. Th is includes for a policy deployment framework for a ligning and prioritizing problem solving activities in line with the business goals of the organization. The m anagers’ (e.g. CEO, GM, MD, Directors, Dept. Heads, Section Managers, and Supervisors of various departments) r ole to lead by developing the abilities of their staff with problem solving capabilities, at every level i n the organization and throughout their career is very cr ucial in lean process management and a key feature of Toyota Production System (Lean Process Management) is not to lay-off its employees. Therefore, the research conducted in this thesis ai ms to identify the following research problem on wh y lean process management is focusing on people as a key d river to obtain the optimum participation of employ ees in eliminating wastages and how can we enhance problem solving capability across the organization. Consequently, two research questions are derived an d will be answered in this thesis. What are the characteristics of lean process management which are focusing on people as a key driver to obt ain the participant of employees in eliminating wastages an d how to enhance problem solving capability across the organization. 3.0 The Need of Developing a New System to Enhance Problem Solving Capability Each of the three systems in framework has an own o bjective. The objective of the lean process managem ent system is to identify and eliminate wastages by rem oving non value added activities. People management systems need to provide the capability for rapid im provement and adoption to change. Here, again, we m ust accept the fact that change is inevitable and that the speed with which the necessary modification are made is the deciding factor in our survival. The objective of the business management system is to apply caref ully the organization’s limited resources, including capital and hard assets as well as time and human assets. 255 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY Three integration elements with total employee invo lvement from top to bottom play an important role f or sustaining problem solving among employees in pract icing lean concept. It is important to create peopl e development system (PDS) which consists of all thes e three elements with total involvement of people t o increase problem solving capability. People managem ent system, Business management system and Lean process management system are integrated by princip les that, in a sense, hold them together. These pri nciples are meant to provide a framework (Fig. 1) to focus the direction in enhancing problem solving capabili ty among employees by forming as people development sy stem (PDS) in lean process management. They are: Team Environment Self Directed Communication Mission Core Value Vision Objective Strategy Strategy Initiative Personal Objective Technical Requirements Cross Functionality Training Needs & Effectiveness Skill Achievement Lean Process Management System Respect For People Business Management System People Management System KPI People Development System Skill and Knowledge Fig. 1. PDS Framework for Enhance Problem Solving C apabilities among Employees [6] • Key performance indicator - KPI for every level suc h as company, department, section and individual levels which is link towards organization goal. • Respect for people – Respect for people which mainl y focuses on the lean behaviors that each employee in organization should build in their mind . • Skill and Knowledge – Skill and Knowledge for emplo yees will support them in practicing lean concept effectively and efficiently by utilizing th e lean tool and techniques. Another important element incorporated with this pe ople development system framework is teamwork of to p, middle and bottom management. The total commitment of all these three levels will enhance of problem solving capability in lean process management among employees. 3.1 Key Characteristic, Critical Success Factors (CSF) and Related Perfor mance Matrix 256 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY The following key characteristics, CSFs and related performance metrics are identified as crucial in p eople development system of lean process management as in Table 1 below. Table 1: An analytical framework for measuring prob lem solving capability in lean process management [ 6] Key characteristics of integration elements Critical success factors (CSF) of People Development System Performance Matrix KP I Mission Core Value Vision Objective Strategy Strategy Initiative Personal Objective Customer Satisfaction On Time Delivery Zero Defect Cost reduction Effective Operation Cost Achievements of KPI for each level versus goal/target. • Productivity • Customer complain • Scrap/Number of reject • Attendance/ Absenteeism • Tardiness (Schedule time) • Using QCDAC principles Respect for peo p l e Team Environment Self Directed Communication Top Management Commitment Team effectiveness/formation Ideas cost or value Continuous improvements Lean Behaviors Rewarding system • Number of ideas generated • Level of people involvement • Usage of lean tools • Total cost saving projects Measured by Likert-type scale on the following items: • Top Management Commitment • Lean behaviors • Achievement of Leanness level S k i l l and Knowled g e Technical Requirements Cross Functionality Training Needs & Effectiveness Skill Achievement Produce skilled, knowledgeable and innovative employees • Lean tools and techniques assessment • Employee skill metric • Audit by 3 rd party or customers on lean practice • KPI in lean process management determination throug h Mission, Core Value, Vision, Objective, Strategy, Strategy Initiative and Personal Objective for peop le development system is crucial. This will align overall workforce of the company to follow for one common goal. Each level has its own portion of contribution towards the target. The results are co mpared with the target or goal used to measure the success of KPI. The accumulation of success from ea ch portion will reflect the overall achievement of the company goal. • Respect for people in lean process management is an other crucial factor in developing the lean culture throughout organization. In order to measure the le an behaviors, top management commitment, leanness level of the company and perception of team member’ s capability, Likert-type scale is used to get the responses from respondent. For example, one can ask managers to rate the degree of support by top management on five-point scale from no support (1) to total support (5). Beside this, the problem solv ing capability also can be measured by counting the num ber of ideas generated, Level of people involved an d the total cost of the project. • Skill and Knowledge in lean process management is t he fundamental requirement for employees to equip themselves. Without this they can’t perform well in solving problem to identify and eliminate wastages . Lean tools and assessment techniques by using asses sment criteria to determine the level of implementation using spider web chart with rating o f 1 (beginning to introduce) to 5 (practice with 257 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY excellent). Another measurement on employee skill m etric will emphasize on employees skill and their cross functionality. 4.0 Methodology The method of data collection was by participatory action research at kitting department as Case Study Company. The researcher attached with pilot team th roughout the study conducted to implement the peopl e development system developed [6]. The performance m easurement model (Fig. 2) was developed to measure the effectiveness of People Development System fram ework developed. The performance measurement is divided into three phases, which are: • P hase I (Input) Phase I begin by generating the Value Stream Mappin g (VSM) of current state by value stream manager with the help of functional manager. The VS M will be create by using a simulation software which is PROMODEL, information such as run ratios, scrap rates, manpower, work hours & schedules, changeover times, tool change times, and inventory level will be key into PDS database, here value added and non-value added activities wil l be identified by propose the future value stream mapping. Wastage identified will be categorized und er the QCDAC principles. • P hase II (Peo p le Involvem ent) Phase 2 will incorporate ‘People Involvement’ which refers to the employee participation of the respective department. The identification of the 11 wastages in the previous phase 1, will lead to problem solving. Phase 2 will incorporates the invo lvement of the employee in the problem solving activities. PDS will measure the people involvement through few types of measurement mechanism that will cover each of the QCDAC principles which are Quality, Cost, Delivery, Accountability, and Continuous Improvement. • P hase III (Output) In Phase III lean metrics is link towards the compa ny KPI. Phase III lean metric is adopted from one of the case study company’s customer to show the le anness level of the company in practicing lean concept. By eliminating wastages activities take pl ace in Phase II through involvement of people Phase III lean metrics is generated. Beside this, t he results also will be use to benchmark with the company KPI achievements. 258 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY Fig. 2. PDS Performance Measurement Model 5.0 Results and Analysis The PDS implementation results achievement during p ilot study at kiting department for the duration of one year are presented here. They are only a summary ac cording to the main objectives achieved through the deployment of the PDS framework which has been enha ncing the people involvement in eliminating wastage s. The PDS performance methodology (Fig. 2) are use to monitor the activities of employees in eliminating wastages. Furthermore, PDS Database was developed u sing Microsoft excel and Microsoft assess was used to developed interface for employees to easily key in and generate visual indicator for their use. 5.1 Total Employee Involvement in Lean Process Management The employees’ involvements are categorized into th ree main levels which are top, middle and bottom management. The pie chart on Figure 3 below shows t he level of involvement of employees by generating ideas in eliminating wastages for the year 2007. Th e highest contribution comes from the b ottom level which is 27% and followed by middle level with 9%, while top level is 1%. This indicates that the bottom and middle level involvement which can highly contribute to el iminating wastages activities. Beside this, there a re also combination level involvements in waste elimination . Bottom-middle level with 37%, middle-top with 25% and bottom-top 1%, which shows the important role o f middle management in lean process management implementation to become the mediator to create co mmunication from top to bottom and from bottom to t op 259 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY level management. Furthermore Figure 3 also consist s of pie charts on distribution of type waste elimi nated and type of lean tools used in eliminating wastages . BOTTOM 27% MIDDLE 9% TOP 1% BOTTOM- MIDDLE 37% BOTTOM- TOP 1% MIDDLE- TOP 25% LEVEL OF EMPLOYEES INVOLVEMENT FOR YEAR 2007 5S 29% TPM 33% Kaizen 14% Std. Work 12% VSM 2% TQA 4% SMT 6% Visual Indicator 7% JIT 1% LEAN TOOLS & TECHNIQUES USE FOR YEAR 2007 Material 12% Inventory 2% Overproduction 2% Labor 13% Complexity 10 % Energy 7% Space 16 % Defects 12 % Transportation 2% Time 21% Unnecessary Motion 3% TYPE OF WASTAGE ELIMINATED FOR YEAR 2007 Fig. 3. Pie charts Illustrate on people commitment on LPM implementation 5.2 Top Management Commitment, Leanness Level and Lean Behavior thr ough Questionnaire Surveys Degree of leanness (DOL) was measured as the averag e of the actual changes taking place as measured by the nine principles of lean manufacturing [7]. Degree o f management commitment (DOC) was measured by the level of investment in supporting manufacturing inf rastructures, as measured by WEMP, TRAIN, GROUP and QLEAD [7]. The degree of lean behavior (DOLB) is me asured using the 30 lean practices [8]. The table 2 indicates the mean and index value of DOL, DOC and DOLB. The initial results of January 2007 indicate the de gree of leanness of the company is low with mean va lue is 2.90±0.20. However, at the end of the year which is on December 2007, the mean value increased to 3.87±0.47, increments is 33.4%, and become moderate to high level. Meanwhile, there is a significant increment in the index value of lean behavior, as i t raised from 0.691 to 0.7164, though there is a mi nimal gap to meet the lean standard (0.800). Furthermore the degree of management commitment is moderate at Janu ary 2007 with mean value is 3.32±0.10.but, for December 2007 it also increased to 3.85±0.7, an increment o f 16%, which is almost close to the high level. 260 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY In conclusion, we can say that when the degree of m anagement commitment is increased, the degree of leanness and lean behavior also increased. So, the level of management supporting manufacturing infrastructures have made the company to be more le an and increase level of lean behavior among emplo yees. 5.3 KPI Achievement KPI is an important element that enables the achiev ement of vision, mission, core value, strategy, and personnel objective for people development is cruci al. Achievement of KPI shows the evidence of people involvement to drive high performance so that stake holders and customer will be satisfied. Monitoring on each performance measurement and countermeasure taken to solve any problem occur have contributed to the achievement of KPI. Without classification of any w astage into performance measurement, no monitoring can be made and no problem solving can be done to reduc e the wastes, which have directly caused the failur e of KPI achievement. Kitting department of the Company have gain benefits from many element that are not b een monitored before and have been monitored after PDS is been implemented. Wastages have been reduced dramatically. Principles Matrix Unit Goal/Limit 200 7 Achievement Scrap MQ% 2 . 6 0 % 1.9 7 % NCR % 7 . 8 0 % 0 Snag Sheet Control Limit% 2 0 % 0 Quality Audit # of CAR Zero 1 Overtime Total Monthly man hours% 1 2 % 10 . 5 0 % Downtime % 1 0 % DCS 1 % 1 0 % 9.1 4 % DCS 2 % 1 0 % 8.8 0 % DCS 3 % 1 0 % 7.6 5 % Cost S91 % 1 0 % 10 . 3 4 % Delivery Output % 9 7 % 10 0 % Attendance % 9 2 % 90 . 7 % Training Hours 1 8 8hrs 231 4hrs Staff/trg hours % 4 7 staff 100 % Major Accidents Qty accidents Zero 0 Accountability Accident Free Days # of days 9 0days 365 Kaizen RM 1 5 0 K 15 6 K SMT Level Level 4 L4 Continuous Improvement 5S Level Level 4 L4 Table 2 shows the main QCDAC principles with the Pe rformance Measurement metric. Each one has its own targets and limits. The impacts of PDS implementati on have pushed scrap percentage below the limit val ue which has the value of 1.97, that can be considered as achievement of leanness. The reason why the val ue is achieved is because PDS have solved many scrap issu es, such as material dry and ply damage for the who le year. Beside these, complains on product produced f rom internal and external customer shows null. But still Table 2: Kitting Department KPI Achievement for the year 200 7 261 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY there is one finding from quality audit by external auditor on the freezer temperature issue which was due to problem on the temperature sensor was faulty. As for Cost, there are two elements that are compar able, which are Overtime and Downtime. As for Overt ime, kitting department has set the limit to below 12% f or the year 07, during PDS is being implemented, ov ertime are controllable all the time, not even a month exc eed the limit of overtime, which eventually give a value of 11.5% for the whole year. The value of down time fo r each machine that DCS 1, DCS 2, DCS 3, and S91 ar e identified and then been converted into percentage value. As a result, all three (3) machines, DCS 1, DCS 2, DCS 3; downtime value are 9.14, 8.80 and 7.65 where by is below the KPI value of 10%. Machine S91 shows a slight overshoot of KPI value with a value of 10. 34%. It can be said that downtime of machines avail able in kitting department are controllable to below 10% wh ich are considered as an achievement of leanness. Under Accountability, attendance KPI for kitting de partment has been set to above 92% for the whole ye ar. Problems such as emergency leave, annual leave and unpaid leave have cause PDS attendance failed to achieve its target. It only managed to achieve 90.7 % below the minimum requirement of 92%. A solution to overcome this problem has been proposed by kitting department and some of the action items have been executed, for example implementing in-house clinic for employees for those required treatment due to sickness. Beside this the top management has decide d to give rewards for those giving the best commitm ent. Focus on the training and development of employees at kitting department shows achievement of KPI in t erm of training hours and total training program conduc ted. Employees’ involvement in any accidents throug hout the year is zero. In term of continuous improvement, the Kitting depa rtment manage to save cost around RM147,000 for th e year 2007 with projects such as JIT production syst em implementation, There is a reduce in set up time for S91 machine, implementing tools trolley usage, Spl it and batch paperwork implementation, eliminate dr y and resin rich issue. The ideas for the kaizen projects are inspired from the observation on visual indica tors (Phase 2) that was created through the PDS implementation. Each activity also shows the involvement of employ ees from various levels to make the project success. Fi nally, Kitting department also successfully achieve d level 4 for the practice of lean tools on 5S and SMT (Self Management Team) which was audited by third parties on lean practices. 6.0 Conclusion Based on a method point of view, the lean practices at kitting department able to classify and monitor the performance measurement of PSC of employees in elim inating wastages. By analyzing the Phase 2 portion, which will become the Visual Indicators for employe es to notice any abnormalities, PDS also enables th e improvement on capabilities of employees in elimina ting wastages to achieve company KPI. From a lean process point of view, PDS can be implemented in th e Manufacturing Environment to eliminate wastages. Among the benefits of PDS are: • To enhance problem solving capability among employe es at all levels. • To get total commitment of employees from top to bo ttom. 262 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 P EO P LE DEVELOPMENT SYSTEM TO ENHANCE PRO B LEM SOLVIN G CAPABILITIES IN IMPLEMENTING LEAN PRO CESS MANAGEMENT: A CASE STUDY IN AEROSPACE COMPANY • To create lean behaviors among employees and become change agent with the lean culture. • To produce workers with skills and knowledge in usi ng lean tools and techniques. • To increase CI (Kaizen) Activities. • To work towards Vision and Mission of company and I ntegrate LPM in company strategy to work towards achieving business goals and be cost compet itive. The development of PDS takes a different attitude t owards total commitment of employees in lean. It pr omotes the proactive employees in solving problems to elim inate wastages. By using PDS, Problem solving capabilities among employees increase. PDS was succ essfully implemented and it helps employees under p ilot study to enhance problem solving capability in elim inating wastages. It also helps the pilot study to achieve its KPI. For future work the author would like to proli ferate the implementation to other sections which a re different in nature of business such as, area which is not involving machinery and service department to get generalization for the PDS framework to adopt and b enefit from it. Furthermore, the author also intend to develop the PDS software which can help the practit ioner to be more efficient and user friendly with the system. Acknowledgement The authors wish to acknowledge the assistance and support of the collaborative company to allow for t esting the model in real life. Beside that, the author wou ld like to acknowledge the University Technical Mal aysia Melaka for the scholarship granted for his doctoral study. References [1] Liker, J.K., 2004. The Toyota Way: 14 Management Pr inciples from the World’s Greatest Manufacturer. Ne w York: McGraw-Hill: 330 p. [2] Gagnon, M.A., Michael, J.H., 2003. Employee Strateg ic Alignment at a Wood Manufacturer: An Exploratory Analysis Using Lean Manufacturing. Forest Products Journal 53(10):24-29. [3] Spear, S.J., Bowen, H.K., 1999. Decoding the DNA of the Toyota P roduction System. Harvard Business Review 77(5):96-106. [4] Emiliani, M.L., 1998. Lean Behaviors. Management De cision 36(9):615-631. [5] Liker, J.K., 1998. Becoming Lean: Inside Stories of U.S. Manufacturers. Portland, Or. : Productivity P ress: 535 p. [6] A.P. Puvanasvaran, M.H.M.A. Megat, Tang S.H, Muhama d, M.R, A.M.S. Hamouda, (2008a). “A Review of Problem Solving Capabilities in Lean Process Manage ment” American Journal of Applied Sciences Vol.5 No. 5, 2008, ISSN 1546-9239 © 2008 Science Publications, pp.504-5 11. [7] A.P. Puvanasvaran, B.H. Tan, M.H.M.A. Megat, Tang S .H, Muhamad, M.R, A.M.S. Hamouda, (2007a). “Degre e Of Leanness And Managerial Commitment In An Aerospa ce Company” Proceedings of International Conference on Enginering and ICT, ISBN 978 - 983 - 2948 - 22 - 3 ©2007 Universiti Tekn ikal Malaysia Melaka , pp.33-40. [8] A.P. Puvanasvaran, Ooi, M.H.M.A. Megat, Tang S.H, M uhamad, M.R, A.M.S. Hamouda, (2008b). “Lean Behavi or Among Employees In Aerospace Company” Journal of Productivity , National Productivity Cen tre, Malaysia. Vol. 25 pp.29-44. [9] A.P. Puvanasvaran, Rohana Abdullah, M.H.M.A. Megat, Tang S.H, Muhamad, M.R, A.M.S. Hamouda, (2007b). “Lean Process Management as a Company Strategy to b e a Cost Competitive: A Case Study in a Lean Aerosp ace Company” Proceedings of International Conference on Enginer ing and ICT, ISBN 978 - 983 - 2948 - 22 - 3 ©2007 Universiti Teknikal Malaysia Melaka , pp.27-31. 263 264 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS MODEL DRIVEN AND CHANGE CAPABLE PRODUCTION SYSTEMS R H Weston, A Rahimifard, J Ajaefobi, Z Cui MSI Research Institute and Wolfson School of Mechani cal and Manufacturing Engineering, Loughborough University, Loughborough, Leics. LE1 1 3TU Abstract The next generation of production systems will need to be su fficiently programmable and re-configurable to realise multiple value streams with a common and finite set of human and machine resources. It follows that new forms of engi neering environment are required, capable of enabling the full life cycle of comp onent based production systems. Such systems will likely need to cater for a growing product dynamic. This paper reports on a new integrated approach to production systems engineering, which is based on the use of new modelling formalisms, virtual engin eering tools, infrastructure services and reusable production system comp onents. The approach deploys reusable models of people, product, process and plant ( ip4) such that large scale production systems design and on-going change can be achieve d with reduced engineering effort and/or within reduced timescales. The p resentation will illustrate a use of the integrated approach in an automotive industry case study. Keywords: Production Systems Engineering, Complex S ystems Modeling, Virtual Engineering, Virtual Engineering, Component Based Systems. 1.0 Introduction The business environment world-wide has become more dynamic and uncertain. Product related aspects of that dynamic can typically take the following forms : (i) increased product variance during normal oper ation of production systems, (ii) more varied production vol umes, again during normal operation, (iii) more var ied production mix, during normal operation, (iv) more frequent significant change requirements, when new products are engineered and introduced into product ion. A consequent outcome is that the average economic l ife of production systems will reduce unless they a re ‘change capable’; i.e. can be readily and effective ly engineered to cope with impacts arising from 1) to 4). 265 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS Various manufacturing philosophies have emerged tha t can guide industry toward increased responsivenes s such as Group Technology [1], Cellular Manufacturin g [2], Reconfigurable Manufacturing Systems [3], Ag ile Manufacture [4], Mass Customization and Postponement [5], Economy of Scope Production [6] and Holonic Manufacture [7]. But there is significant overlap in their unde rlying concepts and related emergent methods and tools. However the process of choosing a suitab le guiding manufacturing philosophy is not a trivia l one. What is more, if a bad philosophy is chosen, such t hat production systems developed and deployed do no t well match their engineering and business environment, t he likelihood is that the Manufacturing Enterprise (ME) will fail sooner or later. Further, as their enviro nments change over time the best choice of philosop hy will change; as might that best choice with respect to d ifferent product groups made by a ME It follows that design of any given ME must recogni ze key characteristics of its business environment as well as understand impacts arising from associated engin eering and operational constraints. This requiremen t is illustrated conceptually by Figure 1. If these char acters and their impacts are well known, in princip le the said ME can select a manufacturing philosophy which suit s its specific circumstance; so that significant competitive advantage can be achieved relative to o ther MEs. But the reality is very complex; especial ly if a given ME needs to diversify its product range. In s uch cases typically MEs must use a finite human, IT and machine resource to achieve so called economies of scope, as well as economies of scale. Hence, a new opportunity exists for ME’s to realize a step change in production systems delivery and c hange, leading to world class high value, small to medium volume production. This paper reports on developmen ts that promise such a step change, initially in auto, aero and construction equipment industries with ro ll out to other sectors. It describes how integrated models o f people, product, process and plant (ip4) can be d eployed, using virtual engineering software tools, along wit h innovative forms of reusable assembly system components, to full life cycle engineer large scale assembly systems. business environment engineerin g environment production processes production resources P roduction System production demands production materials products: of increased value configure program schedule monitor controlplan manage customers suppliers competitors stakeholders governmental impacts legislation market trends & values product dynamic market structures supplier chain configurations b s i n s s n ir n nt ng in r in g n i r n nt P r d c ti n S st b s i n s s n ir n nt ng in r in g n i r n nt P r d c ti n S st ng in r in g n i r n nt P r d c ti n S stP r d c ti n S st & pr d ct d na ic Fig. 1. Next generation production systems must con tinue to fit their changing environments 266 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS pitch local controller local controller zone controller test station manual tool manual tool local controller local controller zone controller fixed stops pitch OEM supplied robotic system OEM supplied dedicated machine M any work st ations hand tools rework (pull-off spur) rework (pull-off spur) test station engine flow Takt time typically 40 s RF tagRF tagRF tagRF tag work flow M an r k st ati ns ll l l tr ll r l l tr ll r t ll t t t ti l t l l t l l l tr ll r l l tr ll r tr ll r i li r ti t li i t i M an r k st ati ns t l ( ll- ff r) ( - ff r) t t t ti i l t t t t f i i ll l 2.0 Full Life Cycle Engineering Requirements Defined Here automotive industry best practice is outlined when engineering a large scale assembly system 9 which is required to make car engines in significant volumes . The purpose of so doing is to illustrate how virt ual environments have potential to realize a step chang e in best practice industry wide; such that ‘full l ife cycle engineering’ of next generation manufacturing syste ms can lead to business benefits. Figure 2 illustra tes elements of a typical automotive engine assembly li ne. Fig. 2. Illustration of current best practice autom otive engine assembly system Often such a line comprises some 40 to 70 workstati ons that range from being largely automated to prim arily manual; dependent on specific assembly operations r equired at each stage of value generation. Common practice is to ‘pull’ engines through work stations at a specific Takt time; which determines the prod uction-rate at which engine products are output from the line. This requires complex component feeding operations and effective synchronization of operations and work fl ows. Also distributed controls are used to achieve information support and process synchronization. Current auto industry assembly systems engineering practice is world class and is conceptualized by Fi gure 3. But significant constraints remain with respect to making custom products. Inherent complexity levels necessitate formation of multidisciplinary teams, s ome affiliated to end user manufacturers, others to original equipment manufacturers and technology vendors. Tho se teams can be distributed globally and typically comprise 10 to 100 persons. The perspectives of tea m members on ‘what is required’, ‘possible conceptu al and detailed solutions’ and on ‘runtime operations’ and ‘support services’ are different but complemen tary. 267 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS But they are all concerned with people, product, pr ocess and plant (p4) issues, which themselves have complex interdependencies. What team members do and when, i s structured by proven methods and enabled by many kinds of computer tool and information support syst em. Methods used are generally Manufacturing Enterp rise specific but build upon widely known approaches to systems engineering, software and database engineer ing, control systems engineering, waste reduction, proce ss synchronization, etc. Life-Cycle Engineering Requirements of Assembly Line Conceptual Design of Assembly System & its Elements Detailed Design of Assembly System & its Elements Interoperable Elements of the Runtime Assembly System Support Information Services & Structures General Engineering Methods -centred on CE, Lean, Agile, postponement & mass customisation concepts & tools Life-Cycle Engineering Processes -conceived & deployed by Multi- National Manufacturing Partners & & & fr a gm e n te d, m u lti - pe rs pe ct iv e , m u lti pu rp o se , m u lti - ty pe ex pl ic it as s em bl y s ys te m m o de ls - during the vario u s life phases o f su ch syste m s fr ag m e n te d, m u lti - pe rs pe ct iv e , m u lti pu rp o se , ta c it kn o w le dg e ab o u t as se m bl y s ys te m m o de ls - during the vario u s life phases o f su ch syste m s process engineering team production engineering team systems engineering team ergonomics engineering team OEM engineering teams Project engineering CAE software engineering control systems engineering plant engineering information management engineering tools fra gm e n te d, m u lti - pe rs pe ct iv e , m u lti pu rp o se , ta c it kn o w le dg e ab o u t as se m bl y s ys te m m o de ls - during the vario u s life phases o f su ch syste m s j i i Fig. 3. Conceptualization of Current Best Practice Assembly Systems Engineering One major constraint of current best practice is th at necessary requirements to reprogram and reconfig ure (collective and individual) operations of workstati ons (that typically comprise a large scale auto ass embly system or ‘line’) need largely to be determined dur ing first off systems engineering. This requires si gnificant foresight about needed processing routes and operat ions, for all product variants, that must be realiz ed by the assembly system during its intended lifetime. To so me extent remaining uncertainties can be mitigated by embedding redundant capabilities into assembly syst ems, but generally such an approach cannot lead to economic and timely assembly of multiple product ty pes (with their different ramp up and down profiles ) through extended time periods. It follows that curr ent best practice first off engineering is very cos tly. Also as current practice only facilitates limited externali zation and integrated reuse of p4 knowledge and dat a then subsequent projects (e.g. to create a new large sca le assembly system or to make a major change to an existing one) may be equally costly. Even relatively minor u nforeseen changes may not be catered for without ve ry significant re-engineering implications. Consequently, the useful lifetime of conventionally engineered large scale assembly systems will in ge neral decrease as product lifetimes decrease. Other major constraints arising from current best p ractice include: lack of multi-perspective project quantification and decision support; lack of well d esigned and explicitly specified ‘interfaces’ betwe en system 268 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS elements (e.g. modules or ‘assembly system componen ts’); ad hoc use of systems integration technologie s and services; and locking into specific technology or O EM that constrains later change. suppliers customers competitors product dynamic stakeholders technologies government used to experiment, understand, predict & inform policy & production systems re-design & re- engineering Real World Ideal World Model of production system & its business environment Enterprise & business environment modelling Process & value stream modelling Workstation modelling IT systems modelling Integrated information modelling People modelling Control systems modelling Machine modelling Product design (CAD) Product engineering (CAD/CAM) & & & Fig. 4. Virtual engineering of responsive productio n systems Current work of authors and their research colleagu es in the MSI Research Institute and Centre of Exce llence for Customized Assembly at Loughborough University seeks to deliver a step change in current best prac tice leading to ‘full life-cycle engineering’ of large s cale assembly systems; this paper and conference pr esentation explains in outline how new forms of ‘virtual engi neering environment’ and ‘reusable assembly system components’ are being developed that are suitable f or cross industry sector deployment. Figure 4 shows in concept how CECA project engineers are deploying a set of virtual engineering tools to create an integ rated set of multi-perspective models of assembly systems and their dynamic behaviors. This toolset brings t ogether models of products, processes, people and machines (plant) related to the design, implementation and o ngoing programming and reconfiguration of production syste ms capable of coping with predictable and unpredict able product dynamics of types 1) to 4). As illustrated conceptually by Figure 4, three prime keys to the u se of this virtual engineering toolset are: 1) Use of enterprise modeling to provide modeling form alisms that are needed to cope with the high levels of complexity involved when capturing enduring charact eristics of the business and engineering environments in which a given reconfigurable produc tion system will be used; thereby providing a well defined context for production systems modeling 2) Use of a Product Lifecycle Management (PLM) softwar e platform to manage coherent access to actual product, process, people and machine data and model s pertaining to a given ME and its production systems, and 269 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS 3) Use of various simulation tools including discrete event and continuous simulators that computer execu te and animate models of workloaded production lines a nd production workstations and their related machine and people systems; thereby enabling the se lection and matching of suitable manufacturing philosophies. The set of modeling viewpoints shown conceptually i n Figure 4 is being used to create an integrated se t of graphical and computer executable models that repre sent the example engine assembly production line sh own in Figure 2. The approach taken is illustrated in F igure 5. build & test assembly line& Computer Executable Model of the Assembly Line specify assembly requirements negotiate with & select OEM suppliers conceptually design assembly line in-house & OEM detailed design & planning train operators run assembly Line change assembly line Actual ‘Project Engineering’ Process K rauser ABB Siemens others? OEM Rolesp rocess engi neerin g Roles cont rol sys tem engineerin g Roles production engineerin g Roles ergonomic engineerin g Roles Ford ‘as is’ Roles unified multi-perspective, multipurpose, flexibly integrated set of mixed reality models that externalise & make re-usable key understandings, knowledge and data needed to project engineer the engine assembly line if l i t ti t it & l li t ll i l li i - & t il i & l i t i t l i l li ‘ ’ r r & t r & & r r r r l ‘ j i i ’ K ra s r A BB Si n s th r s? p r c ss n g i n r in g R l s c n t r l sys t n gi n r in g R l s p r d cti n n g in ri n g R l s rg n i c n g in ri n g R l s ‘ ’ K ra s r A BB Si n s th r s? K ra s r A BB Si n s th r s? lp r c ss n g i n r in g R s c n t r s s t n gi n r in g R s p r d cti n n g in ri n g R s rg n i c n g in ri n g R s ‘ i ’ l i i l i i l i l i l i i li l li & l i l j i i l li Fig. 5. Illustrative Use of Multi-Perspective Assem bly System Models by Engineering Teams 3.0 Dynamic Producer Unit (DPU) Concepts; enablers of iP4 model Integration It was observed that new modeling concepts are requ ired to facilitate (I) the required integration of modeling viewpoints illustrated by Figures 4 and 5 and (II) realize effective modeling of production system com ponents. This led the authors and their colleagues in the MS I Research Institute to conceive and develop so cal led Dynamic Producer Unit (DPU) modeling concepts. The prime purposes of DPUs is to enable human, machine and IT resource systems to be described coherently and explicitly as ‘reusable’, ‘change capable’ ‘components’ of manufacturing enterprises. DPU char acterization is designed to facilitate: (1) graphical representation of resource systems (2) explicit specification of resource systems and (3) implementation descriptions of resource systems that can be computer executed within simulation modeling environments. Also developed has been a methodological use of the DPU modeling constructs, in which their use complements that of ISO Enterprise Modeling and pro prietary (discrete event and continuous) simulation software. By so doing the modeling of responsive pr oduction systems is enabled; where such systems com prise 270 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS user defined configurations of process networks, re source systems and time dependent flows of units of work. This modeling method enables decomposition and sema ntically rich representation of complex systems composed from interoperating DPUs that can be compu ter exercised within specific organizational contex ts. It follows that a DPU is defined as being ‘an organ izational unit comprising people, machines and/or c omputer systems that form a configurable, re-usable and int eroperable component of a more complex production system’. Figure 6 illustrates key properties of DPU s that are modeled. DPUs need to function (a) indiv idually, as a holder of one or more assigned roles and (b) c ollectively, by interoperating with other DPUs to r ealize higher level roles (i.e. some configuration of role s to which the interoperating DPUs are assigned). Dynamic character sets are used to describe and qua ntify inherited and acquired behavior traits of DPU s. The generic attributes defined for this purpose belong to three classes: (1) productivity characters, (2) change capability characters and (3) self characters. In g eneral it assumed that all DPUs behave in ways rela ted to these traits, but when a given configuration of DPU s is assigned to a specific role set it is understo od that not all character sets are of equal importance to diffe rent users of manufacturing system models (e.g. to product, process, automation, IT systems and ergonomic engin eers or to business and manufacturing managers). Th e conference presentation will illustrate a case stud y use of DPU concepts to: (i) conceptually model al ternative manufacturing system configurations, (ii) to match these alternative DPU configurations to work loaded roles and (iii) to predict individual and collective DPU behaviors when subject to different forms of produc t dynamic. has productivity characters cost utilisation efficiency generated values output rate timeliness has changeability characters longevity pro-activity reactivity programmability mobility configurability culture has self & characters inter-personal ability motivation stressors & stresses personality D PU p o d i i p h n ge il i p - p b b b el f & -p b& p D PU Fig. 6. Dynamic Producer Unit Concept 4.0 Conclusion 271 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODEL DRIVEN, CHANGE CAPABLE PRO DU CTION SYSTEMS This paper has described in outline how virtual eng ineering tools can be used in a unified fashion to model people, process, product and plant (machine) aspect s of responsive production systems. The multi-persp ective models developed can be captured, managed and updat ed and used by various members of large-scale production systems engineering teams to design and support ongoing engineering of actual production li nes. They achieve this by computer executing and experim enting with the various modeling viewpoints afforde d to them. The approach described is particularly novel in the way that it uses a combination of (i) enterp rise modeling concepts, to structure the design of the m ulti-perspective models, (ii) extended product life cycle management software, to provide an integrated platf orm of reusable p4 information, and (iii) DPU compo nent based modeling concepts, that bridge a current gap in proprietary virtual engineering technology provi sion by providing a generalized abstraction of any form of component building block used in production systems . The overall approach is being case tested in automo tive, aerospace and furniture industries. The confe rence presentation will illustrate in outline how signifi cant benefit is promised to industrial collaborator s in terms of: extended production system lifetime, and better, fa ster, more responsive and leaner production. Acknowledgement The authors wish to acknowledge the assistance and support of Keith Case and Robert Harrison, CECA@Loughborough project engineers and other resea rchers in the MSI Research Institute at Loughborough University. References [1] Suresh, N.C. and Kay, .J, Group Technology and Cell ular Manufacturing: Updated Perspectives, Group Tec hnology and Cellular Manufacturing- A state-of-the-art synt hesis of research and practice, A, 1998, 1-14. [2] Angra, S. et al., Cellular manufacturing - a time b ased analysis to the layout problem . Int. J. Production Economies 112 (2008) 427-438. [3] Koren, Y., et al., “Reconfigurable Manufacturing Sy stems”, Annal of the CIRP Vol. 48/2 1999. [4] Ian, C. et al. “Agile manufacturing transitional st rategies, manufacturing information systems”: Proce edings of the Fourth SME International Conference. [5] Loe N., Postponement for mass customisation, Chapte r 5, in Gattorn J; Strategic Supply Chain Alignment , Gower, 1998. [6] Cui, Z and Weston, R.H., “Model driven engineering of economy of scope systems”, International Confere nce on Advanced Design And Manufacture, Jan 2008, Haikou/S anya, China. [7] Suda, H., “Future Factory System Formulated in Japa n,” Parts 1 and 2, Techno Japan, Part 1: Vol. 22, N o. 10, October 1989, pp. 15-25; Part 2: Vol. 23, No. 3, Ma rch 1990, pp. 51-61. 272 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS AN INTEGRATED METHODOLO G Y FOR STRATEGIC SELECTION PROBLEMS S.M.Ali Khatami Firouzabadi Allameh Tabatabaie Business School, University of A llameh Tabatabaie, Tehran, Iran Tel No. 009 8 (0 ) 2 1 88 7 7 00 1 1 , Fax No. 009 8 (0 ) 2 1 88 7 7 00 1 7 smakhf@ma-atu.ir Abstract The strategic selection problem is a multi criteria proble m which includes conflicting tangible and intangible criteria. In order to select the most appropriate strategic alternative, it is necessary to make tradeoffs between these criteria, as well as taking into account the resource limitations which may exist. Available methods neglect the distance concept which exists between the alternatives’ w eights with regard to a single criterion and its target value. In this paper in addition to ap plying the Analytical Hierarchy Process (AHP) as a stand-alone methodology, an integration of the AHP and Zero-One Goal Programming (ZOGP) is proposed. In this inte gration, each single criterion is viewed as a constraint in a ZOGP model which e nable the model to take into account not only the distances but also to consider the r eal resource limitation for tangible criteria. In order to justify the methodology, it is app lied to selection of Advanced Manufacturing System (AMS). Keywords: Strategic, Selection, AHP, ZOGP , AMS 1.0 Introduction The strategic selection problem is an important tas k for companies. Choosing the best manufacturing pr ocess, choosing between different advanced manufacturing t echnologies and choosing between different supplier s are some examples of this situation. In these situation s, those alternatives should be selected which best consistent with the company’s strategic objectives. The select ion process is the process of narrowing the set of alternatives under consideration [1,4]. A large set of alternatives should be initially screened down to a smaller set because some are clearly not feasible for obvio us reasons, such as infeasibility for manufacturing or the cost of producing [7]. Using rational decision maki ng techniques which compare the remaining alternati ves, a dominant alternative can be chosen [11]. The strategic selection process is a complex task b ecause of the conflicting tangible (such as cost) a nd intangible criteria (such as flexibility), sub-crit eria, different stakeholders and real constraints. An alternative which is best from the point of view of, for exampl e, the manufacturing department, may not be the bes t from 273 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS another department, for instance, marketing, becaus e each individual department has its own perception , viewpoint and criteria. In general, this sort of pr oblem should be investigated at the Multiple Criter ia Decision Making (MCDM) environment [5]. In this situation, t he decision maker(s) is unable to decide between a numbers of alternatives not only because of the pre sence of different objectives but also because ther e are often multiple conflicting criteria. To make a decision, decision maker(s) should make tradeoffs between the conflicting criteria in order to prioritise the goa ls and criteria for selecting one alternative. In t his situation, decision makers should decide which criteria have m ost effect in the decision [3]. This paper suggests a methodology based on integrat ion of AHP and Zero-One Goal Programming (ZOGP) for selecting the most appropriate alternative amon g available conventional and strategic alternatives , considering the distance concept. 2.0 Methodology The basic idea in the methodology is to use problem decomposition and explicit value or preference tra deoffs from the point of view of each criterion or sub-cri terion. When there are multiple criteria for evalua ting alternatives, the best alternative from point of vi ew of each criterion is different [5]. Therefore, t here are distances between the final selected alternative an d the best alternative from point of view of each s ingle criterion. Goal Programming (GP) is a procedure for handling m ultiple-objective situations within the general framework of linear programming. Each objective is viewed as a goal. Then, given the usual resource limitations or constraints, the decision-maker atte mpts to develop decisions that provide the “best” s olution in terms of coming as close as possible to reaching al l goals. The analytic hierarchy process (AHP) has b een proposed as a means of reconciling initial decision -maker’s expression of preference, as well as means of identifying the consistency of that expression. It provides an estimate of additive utility weight tha t best matches the initial information provided by the dec ision-maker. Moreover, when the AHP is used to obta in an initial estimate of the priorities, the initial poi nts are selected on the bases of pairwise compariso n of alternatives. The AHP addresses how to determine th e relative importance of a set of activities in a m ulti- criteria decision problem. The process makes it pos sible to incorporate judgements on intangible quali tative criteria alongside tangible quantitative criteria. The method utilises pairwise comparisons of alterna tives as well as pairwise comparisons of the multiple criter ia. The use of such pairwise comparisons to collect data from the decision-maker offers significant advantag es. It allows the decision-maker to focus on the co mparison of just two objects, which makes the observation as free as possible from extraneous influences. Addit ionally, pairwise comparisons generate meaningful informatio n about the decision problem, improving consistency in the decision-making process. AHP-ZOGP objective fun ction seeks to minimise deviation from desired targ ets for limited resources. The goal constraints represe nt the availability of limited resources. The right -hand side of each equation reflects the targeted or desired l evel of the resources utilisation. In the combine m odel, the objective function seeks to minimise deviations fro m desired levels. The most obvious advantage of usi ng the AHP model for deriving weights for the problem is t hat it provides for consistent decision-making. The combined AHP-GP method offers a systematic, easy-to -use approach to the service quality control instru ments selection decision problem. 274 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS The methodology tries to select that alternative wh ich has minimum total distances using the ZOGP mode l. The distances are minimised with their associated w eights, which are the relative importance of each s ingle criterion, obtained by global weights of the AHP. In the methodology, each individual criterion or su b-criterion has a constraint in the ZOGP model. The global weights of criteria become the objective function c oefficients of the ZOGP model. These coefficients associated with the distance from the left hand sid e (coefficients of each individual alternative) can be obtained by AHP (if they are related to intangible criteria), and a normalization process (if they are related to tangible criteria). When alternatives are compared with regard to a single criterion, they will have w eights that show the relative importance of the alternatives re gard to that criterion. These weights become the co efficients of the zero–one variable (non-selection and selecti on) of an alternative, respectively. The right hand side (the target value) of each constraint is in fact, the be st coefficients of left hand sides’ coefficients fo r intangible criteria and normalized target values for tangible criteria. There is also a distance between the lef t hand side and target value of a constraint. The distances are the slack (surplus) variables of each constraint. If the best alternative from point of view of a criterion is id entical with the final selected alternative, then t he distance is zero. Otherwise, there is a distance which impacts on the value of objective function. The model minim isation iterative process tries to eliminate those alternat ives where the related criterion coefficients in th e objective function are more than others. The nature of zero–o ne variables will select one alternative which has the smallest distance. 3.0 Case Study To implement the methodology, the problem of AMS se lection is considered which has been previously sol ved by AHP method [2]. The AMS selection is a strategic decision because of its long term intangible benef its (such as improving flexibility and the standards), non-repetitively, having many conflicting criteria, and non- existience of the status quo alternative [8,9,10]. These factors necessitate applying the MCDM approac h because traditional financial methods are not able to include intangible benefits associated with AMS. A comparison is then made between using AHP and the p roposed methodology. This problem is selected because all the elements of the methodology such as hierarchy construction, determining criteria, pair wise comparisons, AHP solution, have been identified. Th e problem involves the selection of one alternative among Flexible Manufacturing System (FMS), Transfer Line (TL), Flexible Manufacturing Cell (FMC), Flexible Manufacturing Module (FMM), and Job Shop (JS) for d eveloping a company. 3.1 Solving the Problem using AHP alone This problem has been solved using AHP [2]. All the criteria and pairwise comparisons have satisfied t heir associated rules. The hierarchy, criteria, alternat ives, relative importance of criteria, and final we ights of alternatives have been depicted in Figure 1. Applyi ng AHP indicates that the ranking of alternatives a re FMS, TL, FMC, FMM, and JS, respectively. Interested read ers are referred to the reference for a full descri ption of the criteria and alternatives. 3.2 Integration of AHP and ZOGP 275 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS To solve the problem using an integration of AHP an d ZOGP, it is necessary to construct the constraint s and objective function of the ZOGP model. The alternati ves should be evaluated by pairwise comparison agai nst a criterion in order to obtain the partial weights wh ich will form the parameters of constraints in the ZOGP model. In other words, the winner alternative can b e determined from the point of view of each criteri on by the relative importance of available alternatives. Thes e partial weights can be found in the original pape r [2]. 3.2.1 ZOGP Model The ZOGP model of this problem based on minimisatio n sum of individual relative importance of criteria or sub-criteria and using partial weights for paramete rs of constraints is as follow: − + − + − + − + − + + + + + − + − + − + − + − 1 2d017.01 1d017.01 0d123.00 9d014.0 0 8d110.00 7d129.00 6d068.00 5d038.00 4d042.00 3d026.00 2d214.00 1d202.0Min (1) Subject to 4 9 9.00 1d01dJS077.0FMM149.0FMC239.0FMS499.0TL036.0 = + − −+++++ (2) 4 9 1.00 2d02dJS039.0FMM086.0FMC243.0FMS491.0TL141.0 = + − −+++++ (3) 4 8 3.00 3d03dJS044.0FMM107.0FMC218.0FMS483.0TL149.0 = + − −+++++ (4) 5 3 7.00 4d04dJS043.0FMM081.0FMC193.0FMS537.0TL146.0 = + − −+++++ (5) 5 0 6.00 5d05dJS042.0FMM091.0FMC210.0FMS506.0TL150.0 = + − −+++++ (6) 0 3 6.00 6d06dJS036.0FMM098.0FMC242.0FMS493.0TL130.0 = + − −+++++ (7) 0 4 1.00 7d07dJS041.0FMM079.0FMC151.0FMS284.0TL445.0 = + − −+++++ (8) 4 6 2.00 8d08dJS037.0FMM079.0FMC126.0FMS297.0TL462.0 = + − −+++++ (9) 5 0 3.00 0d09dJS039.0FMM087.0FMC252.0FMS503.0TL118.0 = + − −+++++ (10) 5 0 5.01 0d10dJS035.0FMM067.0FMC132.0FMS261.0TL505.0 = + − −+++++ (11) 5 0 2.01 1d11dJS502.0FMM266.0FMC136.0FMS069.0TL027.0 = + − −+++++ (12) 4 8 2.01 2d12dJS042.0FMM100.0FMC261.0FMS482.0TL115.0 = + − −+++++ (13) 1JSFMMFMCFMSTL =++++ (14) 1or0JS,FMM,FMC,FMS,TL = (15) 0/jdAll ≥ −+ (16) For example, 0.036 in first constraint (2) is the r elative importance of TL alternative when all the a lternatives are compared against flexibility criterion. Objecti ve function coefficients are the relative importanc e of each criterion which has been shown in Figure 1 (level 2 ). The target values of constraints are the best pa rameters of relative constraints, so that when choosing the best alternative from a special criterion, there is no distance or difference between the target value and the chos en alternative. However, when a non-optimal alterna tive is chosen in respect of that criterion, there is a dis tance which is considered with its global weight th at is now the 276 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS coefficient of objective function. It is worth noti ng that the nature of objective function minimises the undesirable distances. The result of the AHP and AH P-ZOGP approach are shown in Table 1. As the table shows, when more than one alternative should be sel ected (for instance two of them), then the solution of new methodology is differed from AHP alone. The AHP alo ne selects the FMS and TL, while the new methodology chooses FMS and FMC. Selection of AMS Flexibility (0.202) Quality & Reliability (0.214) T echnical Feasibility (0.026 ) Market Position (0.042) T echnology Position (0.038) Investment (0.110) T hroughput (0.129) Inventory (0.110) In formation (0.014) Capacity Utilisation (0.123) Employee Relations (0.017) Human Factors (0.017) T ransfer Line (T L) (0.236) Flexible Manu facturing System (FMS) (0.412) Flexible Manu facturing Cell (FMC) (0.198) Flexible Manu facturing Module (FMM) (0.099) Job Shop (JS) (0.054) Level 1 (O verall Goal) Level 2 (C ri teria) Level 3 (Alternatives ) Table 1: Result of AHP and AHP-ZOGP Alternatives Orders using AHP Orders Using AHP-ZOGP TL FMS FMC FMM JS Second First Third Fourth Fifth Third First Second Fourth Fifth Fig. 1. Hierarchy, criteria, and alternatives of AM S problem 277 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN INTEGRATED METHODO LO GY FO R STRATEGIC SELECTION P RO B LEMS 4.0 Conclusion This paper indicated that when making a strategic s election decision involving diverse range of confli cting criteria, integration of AHP and ZOGP could help th e decision maker(s) to make a sound decision. AMT selection was applied to justify the methodology be cause of its strategic nature. The case study demon strated that the proposed approach can give the decision ma ker(s) some useful aids in order to make a final decision. These aids include the deviations from each target value which means with selecting a specified altern ative, what criteria are satisfied exactly and how much at tainability has been occurred for other ones. This methodology introduced resource limitations for tan gible criteria (such as budget limitation) and crit eria constraints in order to remove the drawbacks of AHP when it is used alone. The methodology can be app lied for identical selection decisions such as selection of design alternatives in the early stages of desi gn process, selection of the contractors for manufacturing the special parts for production, selection of supplier s, and so on. It is obvious that for these applications, the problems will have the special criteria and limitat ions which will be included in the selection process. Acknowledgement The author wishes to acknowledge the assistance and support of Allameh Tabatabaie University. References [1] Chen, L.C., and Lin, L., 2002, Optimisation of prod uct configuration design using functional requireme nts and constraints, Research in engineering design , Vol 13, 167-182. [2] Datta, V., Sambasivarao, K.V., Kodali, R., and Desh mukh, S.G., 1992, Milti-attribute decision model us ing the analytic hierarchy process for justification of man ufacturing systems, International journal of production economics , Vol 28(2), 227-234. [3] Galotti, K.M., 2002, Making decisions that matter, Lawrence Erlbaum Associates . New Jersey. [4] Green, G., 2000, Towards integrated evaluation: val idation of models, Journal of engineering design, Vol 11(2): 121-132. [5] Khatami Firouzabadi, S.M.A., and Henson, B.W., 2004 , An aggregation method for multiple stakeholders’ in design selection decisions, Proceedings of the second international conference on manufacturing research, Sheffield. [6] Khatami Firouzabadi, S.M.A., and Henson, B.W., 2006 , An alternative method for evaluation of design alte rnatives in the early stage of design process, Proceedings o f the fourth international conference on manufactur ing research, Liverpool. [7] Lovatt, A.M., and Shercliff, H.R., 1998, Manufactur ing process selection in engineering design, Part 2 : an approach for creating task-based process selection procedure s, Journal of material and design, 19: 217-230. [8] MacDougall, S.L., and Pike, R.H., 2003, Consider yo ur options: changes to strategic value during imple mentation of advanced manufacturing technology, OMEGA, The International Journal of Management Science, Vol; 31: 1-15. [9] Noble, J.L., 1990, A new approach for justifying co mputer-integrated manufacturing, Cost management , Vol 3(4) 14-19. [10] Ordoobadi, S.M., and Mulvaney, N.J., 2001, Developm ent of a justification tool for advanced manufactur ing technologies: system-wide benefits value analysis, Journal of Engineering and Technology Management, Vol 18: 157-184. [11] von Winterfeldt, D., and Edwards, W., 1986, Decisio n analysis and behavioral research, Cambridge University Press, Cambridge. [12] Zeleny, M., 1982, Multiple criteria decision making , McGraw Hill , New York. 278 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN AN IT PERSP ECTIVE OF LEAN Zeynel Badak 1 , S Ahmed Abbas 1 , Colin Herron 2 1. Teesside Manufacturing Centre, School of Science a nd Technology, University of Teesside, Middlesbrough, Tees Valley, TS1 3BA, UK 2. One NorthEast – Regional Development Agency Goldcr est Way, Newburn Riverside, Newcastle Upon Tyne, NE1 5 8NY, UK Abstract It is generally accepted that Lean improves manufacturing proce sses with recognised tools and techniques, and equally assumed that present day IT systems, particularly enterprise-wide systems, are essential for companies seeki ng efficiencies through organisational integration. Addressing the question “can Lean and I T co-exist or even should they?”, it may be observed that both Lean and IT are tools wi th respective uses and strengths. Hence, a potential exists to maximise impact for a business if they are used concurrently. The authors of this paper bring a collecti ve insight into this debate based on practical experiences in delivering Enterprise In tegration and Lean respectively, and the development of a methodology that is constantly evolving. The central theme of this paper it to evaluate how a Lean progr amme should be supported and enhanced by a complementary IT infastructure an d tools. The proposition is that a well configured IT system acts as the “i nformation spinal cord” and “nervous system” reaching out and connecting departments as w ell as the processes taking place. Two case studies are used to illustr ate the coupling between Lean and IT being investigated in these long term studies c onducted jointly by two teams in the North East region of England. Keywords: Lean, IT, Enterprise Integration, ERP, Co ntinuous Improvements. 1.0 Introduction Lean Manufacturing has become an established and pr oven mechanism to improve competitiveness within a manufacturing organisation given the success of Toy ota Corporation [14]. With regard to IT based syste ms intended to provide business-wide integration and e fficiency, these have been evolving since the early 1950s through MRP and MRPII to the current ERP (Enterpris e Resource Planning) and ERPII. Nevertheless, chang e 279 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN programmes and implementations of ERP systems have been adopted with varying degrees of success and failure [6] and various strategies and approaches e xist for implementation of these systems in an orga nisation [10]. The research of Bhasin and Burcher [5] who, q uoting Mora [17]; Sohal and Eggerston [22]; Baker [ 3] and O’Corrbui and Corboy [19], report failure rates at 90%. A fundamental question is which technology should a business implement first or should there b e a parallel approach. The authors of this paper ha ve encountered this quandary many times in their consu ltancy and training activities over recent years. H ence, this paper aims to addresses the question of compat ibility between computer based manufacturing suppor t systems and what is recognised as the Japanese appr oach to productivity improvement and enters the chi cken- and-egg discussion as to which should be commenced first: an integrated IT driven enterprise or a Lean programme. Two case studies illustrate a developed methodology, where two support programmes are worki ng together to achieve overall business improvement wi thin a company. 2.0 Holistic Approaches in Organisational and Systems Change The origins and analysis of communication networks go back to 1940-50s in the work by Leavitt et al [1 3]. In their findings, “all-channel” networks were found t he most suitable for organisations for corrective f eedback possibilities, which are important for an organisat ion to take rapid decisions and deploy actions. Nev ertheless, as more and more parties are involved in a communic ation network, the complexity increases. For exampl e if 16 individuals or departments are represented, 120 bi-directional channels are required (combination r ule in Math) for effective cross-functional collaboration in carrying out a task. Therefore, a more holistic and integrating tool is proposed in current environment to cope with the increasing complexity, manage information flow and ultimately increase the levels of integration. An appropriate and fit-for-purpose tool will assist people to manage information, have a greater perception about targets, responsibilities, proble ms, actions required to tackle constraints primarily in their own areas and then in the organisation as a whole. If we represent an organisation as a pyramid with t hree distinct layers, conceptualised as Strategic, Tactical and Operational, we could be looking to implement a n IT system such that it would help integrate these management tiers in an effective manner, managing b oth the information and material flow between departments or functions so that gaps between the f unctions do not occur. Fig. 1 depicts how the verti cal tiers and horizontal silos combine to form islands in an archipelago. Organisational Tiers Functional Silos Archipelago Fig. 1. Evolution of an archipelago Strategic Tactical Operational + = 280 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN The use of integrated systems and ERP systems in pa rticular, attempts to bridge these islands within t he archipelago. Only if the organisation and its proce sses are harmonised with the systems capabilities w ill the islands move from the archipelago to a truly integr ated “continent” as proposed by Bessant and Haywood [4]. From research [1]-[2] and direct experience of the authors, this integration also benefits company mis sion and strategy, which can be divided into Critical Succes s Factors (CSFs) or sub-goals; Key Business Process es (KBPs) to achieve the CSFs and Key Performance Indi cators (KPIs) to monitor the performance and the effectiveness of KBPs. Furthermore, every departmen t or function benefits as it has its own unique pro cesses. Some examples of these are given below: Table I: Examples of departmental processes − People Training, Development and Teamwork − Financial Performance − Customer Relationship Management − Marketing & Business Development − Product and/or Process Quality − Organisation Structure and Culture − Supply Chain Management − Sales Analysis & Forecasting − Production Planning and Scheduling − Data Collection − Financial Reporting − Time and Attendance − Business Communication − Recruitment, Reward and Retention − Lessons learnt & Continuous Improvement − KPI Measurement 3.0 Lean Manufacturing vs. IT The demands on a company that sets out to implement a Toyota production System (TPS) type methodology are far reaching, and may require a total revision of the organisation. There is an acknowledgement f rom some researchers of the difficulty of introducing a para digm shift such as Lean Manufacturing. The work of Papadopoulou and Özbayrack [20] confirms this by re porting: “The transformation process to a Lean Manufacturing production system requires a lot of e ffort, participation of all levels in the hierarchy , introduction of new principles not only in the shop -floor level but also in the company culture and organisational structure.” Motwani [18] also highl ighted the commitment required. A vision of a Lean company by James-Moore and Gibbons [11] should have the following five characteristics, which lend themselves to measurement: flexibility, process con trol, optimisation, waste elimination, and people utilisation. These parameters clearly highlight the need for a feedback and control mechanism in order to implement and monitor them. Moreover, the subject o f what is a Lean company is important. Jayaram et a l [12] and James-Moore and Gibbons [11] suggest that a Lean company is much more than a Lean manufacturer. They also emphasised the fact that a whole organisation is involved across all tiers and layers; e.g. not just the production department. This is wh y the authors have laid considerable emphasis on en terprise tools, notably ERP. To introduce a paradigm shift s uch as Enterprise Integration and Lean Manufacturin g specifically, there must be consensus within all th e management as to how a company would like to set out a strategy and operate to fulfil that strategy [15]. Perez [21] suggests: “When the full constellation of a new techno-economic paradigm tends to take over the bul k of production within a society, it will not yield its full growth until the socio-economic framework is transf ormed to adapt to its requirements”. There is no re ason why the organisational structure of a manufacturing company cannot be considered as the same as a soci o- economic framework in miniature. TMC’s CHI methodol ogy [2] is based firmly on this premise, such that without the consensus and commitment from the top m anagement team, no further step or work is initiate d. 281 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN 3.1 North East Productivity Alliance (NEPA) Approach for Lean Manufacturin g NEPA is a public/private advisory programme, which was created to improve the competitiveness of North East England regional manufacturing [8]. The progra mme has two elements: The Digital Factory Programme , which was created both to introduce engineers to di gital manufacturing technology and up-skill enginee rs in the latest software and Best Practice Dissemination Team (BPDT), which has the role of disseminating “world’s best” manufacturing practice into regional manufacturing [9]. 3.2 Enterprise Integration through TIE Approach TMC (Teesside Manufacturing Centre) is based at the University of Teesside, specialising in consultanc y and tailored training services in order to assist manuf acturing and engineering companies to become better integrated from an organisational, cultural and sys tems perspective, by using its own developed method ology [1]. The aim is to utilise carefully selected integ rated electronic-based management systems to benefi t a whole organisation through changes in associated processe s and procedures, and also people skills and attitu des. As Enterprise Integration can be a problematical and l engthy task and it involves large and diverse numbe rs of staff, associated processes and suitable systems, a three-phase methodology known as CHI [2] has been developed by TMC in order to “embed” its TIE (Total ly Integrated Enterprise) concept. The three steps consist of: Phase 1: C onsensus to be achieved by the top management team on the strategic and operational needs for enterprise-wide integration and improveme nt; Phase 2: H oming-in to the requirements by assessing the processes and procedures at present to identify potential solutions; Phase 3: Implementation of systems and new procedures across the company. 3.3 Harmonising Systems and Operational Processes All processes operate better if outputs or outcomes of a process are predictable within defined limits ; indeed, this is the prime objective of the tools and techni ques of Lean. Integrated systems also operate bette r with an element of predictability. It is the concept of sus tainability/predictability which may have an influe nce on the chicken or egg question (Lean or IT first). An out of control process requires constant and accurate u pdates. Therefore, it is suggested that parallel activities are commenced within a company to introduce stabil ity, which will introduce predictability with regard to output s and introduce a system, which will report on and drive the process. The strength of an integrated approach (su ch as the CHI process) forces the management to thi nk about what their requirements are. The Lean process gets on with the removal of waste to introduce the stability. Figure 2 depicts an illustrative situati on in a company, where the two types of interventio ns, from NEPA and TMC, are occurring more or less concurrent ly. 282 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN Fig. 2. Joint NEPA and TMC roles in organisational development A common characteristic (in both TMC and NEPA appro aches) is the involvement of people from the outset , then addressing the processes involved and finally setting the scene for systems. Any improvement init iative has to involve people and has to tackle processes t hat are people familiar with. This enables the sign -on being achieved and builds the foundations for good commun ication between all parties concerned. This people focused methodology then instigates analysis of the processes and encourages an enterprise-wide thinki ng acknowledging that processes ripple through the who le organisation. Capturing the “as is” state and appreciation of how processes interact with departm ents and individuals within the organisation or the whole picture is crucial to engineer the “to be” state, f acilitating introduction of appropriate integrated systems and, most importantly, for successful Enterprise Integra tion. 4.0 Case Studies The first case study organisation is a small compan y; a family owned business employing around 70 peop le. It is a manufacturer of accurate machined products, ma inly consisting of steel rods and tubular component s. The bulk of its customers are in the automotive industr y. Through interchanges with TMC, the company reali sed that the most effective way to improve the company’ s overall manufacturing performance was through systems. Senior management realised that in order t o achieve this goal the company would have to under take a comprehensive programme of change. In the joint exp erience of both TMC and NEPA, these situations are particularly fundamental and prevalent in tradition al family run businesses. Some problems did arise i n this company, with conflicts within the family hierarchy as to the need for change and the nature of the mechanisms being implemented. Since this was happen ing during the period that TMC and NEPA were engaging with the company, the consequences for the day to day to operations had to be dealt with by t he two teams, alongside the implementation of systems and working practices. Strategic Priorities Change awareness and preparation Systems implementation – putting into practice Using the tools and techniques - QFD, Kaizen, 5S, PLM, FMEA, Poka Yoke, PDCA, Value Stream Mapping Mainly TMC’s domain Mainly NEPA’s domain Organisational Development Time 283 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN Phase 1 – Consensus - was carried out by running st rategic management workshops with the management te am to review the company’s position compared to the vi sion and strategic priorities. The team participati ng in this phase included some people, who also had operationa l roles, and this had the great merit of opening up the wider picture. The process facilitated the manageme nt team to develop Critical Success Factors and Key Business Processes, and set the scene for an Enterp rise Resource Planning (ERP) system. Towards the en d of Phase 2 – Homing-in, a KTP - “Knowledge Transfer Pa rtnership” (partly funded by the UK Department of Trade and Industry), was initiated in order to depl oy a postgraduate “KTP Associate”, who then became the dedicated project coordinator and leader. An ERP sy stem was specified, selected and acquired. Addition al departmental interviews and exercises were carried out to identify improvement areas and identify oper ational requirements, training, etc. Soon, it was realised that shop floor efficiency and associated practices were proving problematic. Hence, the NEPA team was calle d in to address the issues, and subsequently both Enterprise Integration and Lean initiatives ran in a parallel and collaborative manner. The second case study organisation is a medium size d company in the Oil & Gas Sector that employs arou nd 650 people, operating in a niche market segment wit h technically complex and sophisticated products. The company has been acquired and sold several times ov er the past decade, which prevented the organisatio n from properly addressing and streamlining its busin ess processes. This resulted in outdated systems, m anual interventions, duplications and most importantly th e archipelagos as depicted previously in Figure 1. The first contact was made by NEPA. Following a dia gnostic assignment [8], commenced a programme of disseminating selected Lean tools and techniques ov er an initial 12 month period to improve quality, c ost and delivery performance. Soon it became clear that the principal problem was the ability of the company t o order and deliver the correct components to the point of fit when required. As a result TMC was asked to als o engage with the company. The company’s senior mana gement quickly acknowledged that Enterprise Integration was a strategic undertaking, and the CH I methodology was put into action. Phase 1, the str ategic management workshops, examined the company’s curren t position compared to the vision and strategic priorities Critical Success Factors (CSFs) and Key Business Processes (KBPs) prompted senior managemen t to order an internal review of its top-level thinki ng, which were later used as the basis of an organi sational re- structuring exercise. Hence, Phase 1 activity is de scribed as a “top down” approach. Phase 2 – Homing- in - was carried out immediately following the organisat ional restructuring and consisted of departmental interviews to capture the ”as is” state of existing processes, systems in use, gaps between the depart ments, linkages required in order to communicate with othe r departments and cultural issues inherent within t he organisation. Hence, this phase can be described as a “bottom up" approach. The findings of this phase were then communicated back to the senior management tea m. Now that the inefficiencies and problems at the tactical and operational levels could be identified , these problems were prioritised and the decision was made to tackle them before introducing any systems. This resulted in the re-engaging of the NEPA team to ad dress the manufacturing processes and quality issues. As a result the future strategy is now one of twin tra ck as more and more people in the company began to understand that an enterprise is the whole organisation in its totality, and that a holistic approach was going to be necess ary to implement change. TMC then used the information gathered through the Phase 1 and Phase 2 activities to specify and selec t a comprehensive ERP system to operate across the ente rprise as a major facilitator for change and improv ement. During this process of selection the team continued a parallel activity of enterprise-wide awareness o f the 284 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN anticipated systems and process reappraisals. Once again, a Knowledge Transfer Partnership (KTP) has been awarded by UK Government to facilitate the major ch anges being implemented in systems, processes, work ing practices and organisational culture. In this insta nce there are two KTP Associates, one focusing on t he front end activities (supply chain, customer interfaces, etc) and the other focusing on back end activities (manufacturing, planning, scheduling etc). In addit ion TMC has provided a project manager and extra pe ople resources for the implementation in the company ove r a one to two year period, to sustain the new syst ems and the associated working practices in the organizatio n so that the changes are not just implemented but embedded for the long term. The company is now grow ing rapidly, with manufacturing operations overseas , and has been floated on the UK Stock Exchange. The joint work of NEPA and TMC will continue to be an important factor as it strives to handle this expan sion with integrated systems and processes. 5.0 Conclusion Our observation so far in the industry is that impr ovement schemes around Lean and integrated IT Syste ms, usually in the form of ERP, to address company requ irements, have so far been running in a disjointed manner and the literature has been unfairly dismissing any collaboration initiative. There is also the proble m of the consultants working in a separate manner in the are as concerned, mainly because they lack the expertis e across both the disciplines. As our partnerships in both c ase studies have shown, organisations will reap the benefits through the combination of Enterprise Integration a nd Lean if the activity is undertaken in a cohesive manner. The depiction in Figure 3 represents the TMC and NE PA partnership and working model. It attempts to illustrate how both short term operational and long term strategic aspects are brought into play in th is joint approach. The overall objective is organisational d evelopment. The place of Phase 3 of TMC’s methodolo gy is not indicated explicitly but is embedded within the “operations” box, which is influenced both by t he Computer Integrated System and by the NEPA Lean app roach. OPERATIONS NEPALean Develop Company Operate Company COMPUTER INTEGRATED SYSTEM Short term intents Long term intents TMC Phase 1 & 2 Proactive (new objectives) Reactive (Problems) Information QCD Learning QCD Drives Through removing waste: - Enable predictability - Increase reliability - Optimise efficiency Through workshops & interviews: - Agree on strategic priorities - Identify improvement areas - Produce operational requirements Fig. 3. TMC/NEPA Collaborative Working Model 285 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN Our experiences have shown us that such a collabora tive model ties the improvement initiatives across the whole organisation covering strategic, tactical and operational intents rather than offering a point s olution in a specific area. TMC’s approach enables the company t o see the bigger picture, covering the long term an d short term intents in the existing operations. The NEPA a pproach removes waste within processes so that oper ations fulfill the remits of quality, cost and delivery ex pectations through “quick wins” or “low hanging fru its” after initial reactive responses. As the company knows t he strategic direction and intents from the TMC’s a pproach, any improvements made could be injected back into t he learning loop and decisions could be made proact ively to develop the company forward on the foundations. An additional point is that IT or an integrated sys tem cannot change a company on its own. It needs activi ties on the ground to respond to data and take acti on. Those who take the action have to be skilled enough to open the tool kit of Lean tools and techniques and apply them. The application must be directed by the system and generate its own data for the system. T he cycle of continuous improvement should be driven by data, as there is a clear cause and effect flow, w hich will direct the next lean activity. In this ever-changing and ever-competitive business environment, it is our view that Integrated System s thinking, i.e. ERP, is increasingly dependent on on e of the most effective continuous improvement init iative - Lean. On the other hand, Lean should also be operat ing in a subservient manner in the TIE up of the tw o ends. Both TMC and NEPA approaches should be adopted and monitored in a longitudinal manner to see the long- term impact and quantifiable benefits, and plans ar e being put in place for this. The next stage of ou r work is to follow up the results of our interventions with selected companies over a long timescale, monitorin g the progress, measuring the improvements, and analyzing the problems encountered. This will be reported in future papers. References [1] Abbas, A., Pinedo-Cuenca, R., Ahmad, M. and Morris, A. (2005). Developing a methodology for ERP implementation – through theory and practice. 3rd I nternational Conference on Manufacturing Research, September 2005, Cranfield University, Cranfield, UK. [2] Abbas, A., Pinedo-Cuenca, R., Badak, Z. and Ahmad, M. (2006). A Three-Step Methodology for ERP. 17th Production and Operations Management Society Confer ence, June 2006, Shanghai, China. [3] Baker, P. (2002). Why is lean manufacturing so far off? Works Management. October. pp. 1-4. [4] Bessant, J. and Haywood, A. (1988) Islands, Archipe lagos and Continents: Progress on the Road to Compu ter- Integrated Manufacture. Research Policy, Vol. 17. [5] Bhasin, S., Burcher, P. (2006). Lean manufacturing viewed as a philosophy. Journal of Manufacturing Te chnology Management. Vol. 17, no. 1, pp. 56-72. [6] Deutsch, C. (1998). Software that can make a grown company cry. New York Times 1998, No: 51. p 1-13. [7] Herron, C. (2006) A methodology to disseminate sel ected lean manufacturing tools into general manufac turing. Thesis accepted for the degree of Doctor of Philoso phy. The University of Newcastle upon Tyne, England . [8] Herron, C., Braiden, P.M. (2006). A methodology for developing sustainable quantifiable productivity i mprovement in manufacturing companies. International Journal o f Production Economics. Vol.104, Issue 1, pp. 143-1 53. doi:10.1016/j.ipe.2005.10.004 [9] Herron, C., Hicks, C. (2005). Utilising technology transfer and change agents to introduce selected le an manufacturing techniques into general manufacturing . Proceedings of the 3rd International Conference o n Manufacturing Research (incorporating the 21st Nati onal NCMR). Cranfield University. 6-8th September. [10] Holland, P. and Light, B. (1999). A critical succes s factors model for ERP implementation. 33rd Hawaii International Conference of Systems Science. [11] James-Moore, S.M., A. Gibbons. (1997). Is lean manu facture universally relevant? An investigative meth odology. Warwick Manufacturing Group, University of Warwick, Coventry, UK. International Journal of Operations & Production Management Vol. 17, no.9, pp. 899-911. 286 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 AN IT PERSP ECTIVE OF LEAN [12] Jayaram, J. S.K. Vickery, and Droge. C. (1999). A n empirical study of time-based competition in the North American Automotive supplier industry. Internatio nal Journal of Operations & Production Management. Vol.19, no.10, pp.1010-1033. [13] Leavitt, H.J. (1951). Some Effects of Certain Commu nication Patterns on Group Performance. Journal of Abnormal and Social Psychology. [14] Liker, J. (2004). The Toyota Way. McGraw-Hill. ISBN 0-07-139231-9 [15] Lillrank, P., Shani, A.B., and Linberg, P. (2001). Continuous improvement: Exploring alternative organ isational designs. Total Quality Management. Vol. 12, no.1, p p.41-55. [16] McAfee, A. (2006). Mastering the Three Worlds of In formation Technology. Harvard Business Review, Nov 2006, Vol. 84 Issue 11. [17] Mora, E. (1999). Management initiatives. Available at: http://www.m.e.umist.ac.uk. [18] Motwani, J. (2003). A business process change frame work for examining lean manufacturing: a case study . Industrial management and data Systems. Vol. 103, no. 5, pp. 3 39-346. [19] O’Corrbui, D., Corboy, M. (1999). The seven deadly sins of strategy. Management Accounting. No. 10, pp . 1-5. [20] Papadopoulou, T.C., and Özbayrack, T. (2005) Leanne ss: experiences from the journey to date. Journal o f Manufacturing Technology Management. Vol. 16, no. 7 , pp. 784-807. [21] Perez, C. (1985). Microelectronics, long waves and world structural change, Word Development. Vol. 13, no 3. [22] Sohal, A., Eggleston, A. (1994). Lean manufacturing production: experience among Australian organisati ons. International Journal of Operations and Production management. Vol. 14, pp. 1-17. [23] Shahnawazuddin, M., Pinedo-Cuenca, R., Abbas, S.A. and Malcolm, D. (2006). A case study of ERP impleme ntation within a small manufacturing enterprise, Internatio nal conference on Technology and Operations Managem ent 2006 (ICTOM’06), Faculty of Technology Management, Unive rsiti Utara Malaysia and Institute Technology Bandu ng, held in Bandung, Dec 2006. 287 288 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT APPLYING MODULARITY-IN-PRO DUCTION AND LEAN PRODU CTION IN A TERRY WEAVING MACHINE ASSEMBLY ENVIRONMENT A. Engelschenschilt 1 , S. Loke 2 1. Science & Technology Research Institute, Universit y of Hertfordshire, Hatfield Campus, College Lane, Hatfield, Herts AL10 9AB, UK 2. Faculty of Engineering and Information Sciences, Un iversity of Hertfordshire, Hatfield Campus, College Lane, Hatfield, Herts AL10 9AB, UK Abstract Lean manufacturing techniques can be found today in many indu strial environments. There is a growing trend for higher product complexity at low er cost, modularity-in- production and shorter delivery times. Textile machines th at produce high quality woven terry fabric are niche products and are assembled in c ellular configuration, without a well-defined internal logistical flow, with a lack of standardised work procedures and with poorly defined quality standards. The problems encountered in the implementation of lean man ufacturing techniques to the assembly of a niche market textile machine are discuss ed in this paper. Design criteria for factory design and assembly process are defined. T he basic design of cellular work cells producing modular pre-assemblies and delivering t o a short assembly line are described and critically analysed. Contribution of this paper is the research on modularity-i n-production in combination with lean production in a terry weaving machine assembly envir onment. This manufacturing approach is unique in the market of terry weaving machines. Opportunities and threats are stated clearly. An improved asse mbly methodology was developed and implemented to mount pre-assemblies on to a basic machine platform. The findings raise new questions about future trends in the assembly of niche market textile machines. Keywords: modular assembly, lean production, niche market, terry cloth 289 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT 1.0 Adoption of Lean Manufacturing 1.1 Fordism and TPS At Highland Park, Michigan, USA, 1913, Henry Ford m arried consistently interchangeable parts with stan dard work and moving conveyance to create what he called flow production. [3]. As Kiichiro Toyoda, Taiichi Ohno and others at Toyota looked at the Ford system just after World War II, it occurred to them that a ser ies of simple innovations might make it more possible to p rovide both continuity in process flow and a wide v ariety in product offerings. They revisited Ford’s original t hinking, and invented the Toyota Production System. [3] The Toyota Production System (TPS) [9] was developed in Japan by Ohno and Shingo and forms the basis of le an manufacturing. John Krafcik labelled the Toyota Sys tem as ‘Lean Production’. He said that Lean Product ion was ‘a system that uses less human effort and less capital to design products faster and with fewer de fects.’ [8] Toyota could not afford the capital-intensive mass production systems used in the USA so instead focus ed upon minimising waste in all aspects of its operations. [1] 1.2 Lean Used in Non-Japanese, Non-automotive, Situations The superior performance of (automotive) lean manuf acturing systems has encouraged the idea of transfe rring lean manufacturing to non-Japanese and non-automoti ve situations. [1] This is based upon the premise t hat manufacturing problems and solutions are universal. However, in practice, Western (non-automotive) manufacturers are often able to emulate the structu ral parts of lean, but have found it difficult to a dopt the required organisational culture and mindset. The im pact is often localised and falls short of the desi red improvements in the overall system. [2] 1.3 Lean Thinking As lean thinking continues to spread to every count ry in the world, companies are also adapting the to ols and principles beyond manufacturing, to logistics and d istribution, services, retail, healthcare, construc tion, maintenance, and even government agencies and non p rofit organizations. [3] Automotive leaning manufacturing techniques can be found today in many industrial environments. With the growing demand for producing product with higher complexity but at low er costs and shorter delivery times, lean technique s have found their way to the niche market of design and m anufacturing of terry cloth weaving machines. (figu re 1) Fig. 1. Visualization of market size of niche produ cts versus one-offs and mass production. 2.0 Problems Definition 290 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT 2.1 Overview of Terry Weaving Machine Assembly Area before Application of Le an. Fig. 2. Basic layout of non-lean assembly area of niche ma rket, low volume, weaving machines. 2.2 Manufacturing Systems in 2004, before Application of Lean. The low volume assembly facility for production of terry weaving machines, before application of lean, was set up in 2004 and measured 1080m². The facility was pl anned to produce 130 weaving machines on a yearly b asis. Total assembly time of a terry weaving machine was 103 hours. Assembly was divided into five sub-areas: 1. pre-as sembly of gearboxes (or direct machine drives), 2. assembly of small parts, 3. storage of small and long parts, 4. machine assembly, 5. shipping area. (see figure 2) The 2004 design process was a concept-driven proces s. Terry weaving machines were completely assembled in assembly cells from frame building up to final test ing (area 3, figure 2). Direct machine drives, moun ted on both sides of the machine, were modularly assembled from dedicated small pre-assemblies (area 1, figure 2) Modules were manually delivered to the assembly cel l with a dedicated cart. Other small and long parts , which required no pre-assembly, were stored at area 3 (fi g. 2). When the assembly and testing was completed, the completed machine was moved to area 5, shipping are a. Introduction of modularity in the Terry assembly in 2004 was a cost-efficient solution. ‘It supports s tep-by-step investment, where later upgrades and modifications to the system are easier’, says Heilala and Montone n. [10] A kitting system was designed to present material t o storage area 3 (small and long parts). Basis for this decision were component part obsolescence, reduced inventory cost and increased process control. 2.3 Problems Areas in the 2004 Layout, before Application of Lean. 291 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT In the assembly area, terry weaving machines were l aunched in batches of 6 machines; spread over 21 me tres of working area. With a maximum capacity of 18 cells, machines were spread over a length of 63 metres. In itially, operators spent 25% of their working time searching for parts in the small and long parts storage (are a 3, fig. 2). Due to the use of lift trucks to move long and heav y parts from storage area 3 to assembly cell area 5 , waiting times were incurred and added to the delay in deliv eries. For all working areas there was a lack of standardi sed work, visualisation, quality standards and inte rnal logistical flow. These items were not well develope d due to the lack of assembly knowledge of the oper ations manager and engineering team. In assembly area 5, t here was no focus on total cycle time, touch time a nd dwell time. Quality level, time and cost objectives were not within control given adequate importance. The integration of modular assembly in the gearbox (or direct machine drive) area 1 required additiona l attention. Single-piece flow, kanban, flow racks an d JIT-driven flow were not introduced. 2.4 Defining the Targets of Introduction of Lean in 2005. The management team agreed upon some overrearching principles that would guide the design process towa rds implementation of lean principles and modularity-in -production. Cross-functional teams were formed and worked on de sign concepts that covered layout and design of sho p floor area. The result was the production of the fo llowing working objectives: • elimination of waste, focus on over-production • standardised work • visualisation of quality issues and assembly sequen ce • one-piece-flow • Jit-driven flows and queues • minimal touch time • modularity as design and operations strategy: modul arity-in-design and modularity-in-production [13] • kanban delivery of small fixation parts: bolts, nut s, clips and others • flexible workforce and simple adaptable equipment • straight and short assembly line 2.5 Authors’ Contributions Solving the Problems. According to Starr [4], Feitzinger and Lee [5], Van Hoek and Weken [12], research on modularity-in-pro duction focuses mainly on the efficiency and flexibility ga ins stemming from the decoupling of the main and mo dule flows. A few studies have analysed the issues that firms must manage well in order to actually achieve these operational performance benefits. (Fredriksson) [11 ] 292 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT Contribution of this paper is the research on modul arity-in-production in combination with lean produc tion in a terry weaving machine assembly environment. This ma nufacturing approach is unique in the market of ter ry weaving machines. The authors and management of the company defined the concepts for the factory desig n launched in 2005 and 2007. Engineering and logistic al study of the projects 1 and 2 were performed by the authors and the operations management team. Modular ity design principles were discussed with machine designers and management team. 3.0 Project 1: Applying Lean Principles to Gearbox Pre-assembly Area 1 in 2005 Project 1 was launched in 2005 in order to start ap plying lean principles on the shop floor, train ope rations managers and operators in lean thinking and to intr oduce the principles of modular assembly. 3.1 Design Concepts • main focus of developing a lean system was the elim ination of waste, mainly over-production and waiting times, whereby waste was defined as ‘anythi ng that adds cost without adding value’ • a pull logistic system (kanban) supports eliminatio n of waste • an attempt to balance the flow of material to match the rate at which it is needed while making the mo st efficient use of labour • adopt the modular assembly approach with a flexible workforce and simple adaptable equipment. [7] 3.2 Performance Improvements The following improvements were recorded resulting from the re-design and implementation of the new operations system: • Waste was efficiently tackled by linking the outgoi ng flow of the gearbox pre-assembly to the incoming flow of the machine assembly. The operator who pre-assembled the gearbox was also responsible for mounting the gearbox on the machine . Over-production and waiting times disappeared quickly because the operator started to work at sam e tact as assembly line. Assembly time of the gearbox decreased from 17 hours to 14, 4 hours. • The modular assembly approach resulted in better qu ality, delivery and cost performance. The explanation is that people could better see the rel ation between their activities and the outcome, wer e better motivated and focused. • Other design concepts integrated in this small gear box line were made visible below in ‘actual configuration’ (fig. 3 ) such as one-piece-flow, st raight assembly line, same centre of gravity betwee n assembly and pre-assembly, improved logistics, dev elopment of an assembly fixture on wheels and optimized locations for incoming logistics (fig.4) • Reduced number of assembly activities made it more difficult to define a well-balanced work flow. 293 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT PREVIOUS CONFIGURATION ACTUAL CONFIGURATION gearbox assy left rack rack rack rackpress rack em pt y fix tu re em pt y fix tu re B A9 00 40 3 Li n ks D R EI EC KH EB EL BA 90 00 41 W EL LE N B A9 00 40 2 Re ch ts D R EI EC KH EB EL fin is he d ge ar bo x le ft fin is he d ge ar bo x rig ht gearbox assy right gearbox assy left gearbox assy right tools GE A RB OX M O UL D GE A RB OX M O UL D GE A RB OX M O UL D GE A RB OX M O UL D GE A RB OX M O UL D worktabl e worktabl e 14,34 m 5, 85 m Fig. 3. Evolution of a gearbox pre-assembly area us ing lean principles Fig. 4. Photo of storage racks location relative to production line 4.0 Project 2: Design Criteria of the New and Lean Terry Assembly Area in 2007. 4.1 Design Concepts The layout of the original, non-lean, machine assem bly area of 2004 was completely revisited in 2007. Lessons learned from project 1 were implemented in the new line layout. Additional design concepts were added to this produ ction line design (fig. 5) and these were as follow s: • modularity-in-production (components on modules in pre-assy areas) and final assembly (components and modules into end-products on the assembly line) [11] • reduction of total assembly time and work-in-proces s • optimised ergonomics 294 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT • dynamic line assembly instead of assembly cells • introduction of air-cushion technology to move the 3000kg heavy machine on the line. • reduction of walking distances by re-locating pre-a ssembly areas next to the assembly line • assembly fixtures to improve repeatability and to a ssure quality of assembly operations • optimised incoming logistics with barcode scanning linked to the MRP software. • use existing weaving machine platform and develop t he terry application on top it. Fig. 5. Optimized line layout with modular assembly areas and air-cushion technology 4.2 Performance Improvements with Justification. Important reduction in development cost and time-to -market was achieved by using an existing weaving machine platform and developing a niche application on top of this platform. Quantification of the imp rovement data was not made available by company. The project was delivered on time and within budget in January 2007. Start-up level of 1,5 machines per day was achieved immediately at re-start of the assembly line, comp ared with a maximum daily output of 0, 8 machines on the prev ious 2004 assembly line. In the assembly process, modules were build instead of parts being transported loosely around the line . The concept of modularity was explained to the assembly team as described by Baldwin/Clark and Asan et al. [15], [16] and [17]: ‘to divide a complex whole into deco upled and more manageable parts or modules.’ The ap proach defined by Ulrich [14] has been adopted during the development phase: ‘The modules can be designed and produced in parallel and still be combined with eac h other, as long as the interdependencies between t he modules are minimized and interfaces between them a re specified.’ The 2007 assembly facility was built on 800m². It had maximum output of 600 terry weaving m achines per year. Total assembly time of a terry ma chine was 52 hours. Quality and delivery levels increased with the lean line setup and the modularity-in-production concep t. As we noticed in the gearbox pre-assembly, the explanatio n is that people could better see the relation betw een their activities and the outcome, were better motivated a nd focused. Pre-assembly activities can also be per formed in ergonomically safe ways since they are not constrai ned by the physical shape of the base object. Kinut ani [18] and Blackenfelt [19] found that pre-assembly activi ties can be undertaken at a comfortable height thro ugh the use of supportive equipment, such as lifting device s. 295 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT To improve efficiency and to reduce complexity, the basic idea was to produce the new machine on a dyn amic assembly line linked to a modular pre-assembly area with dedicated assembly fixtures, torque tooling a nd lifting equipment. At the end of 2007, an average of 2 mach ines per day was reached. Overall delivery times we re reduced from 12 to 10 weeks. As a result of dispersion of activities and resourc es between main flow and module flows, operations h ad to handle disturbances quickly. Also transport modules (racks) were bigger and more complicated than befo re. The warehouse was integrated into the new assembly area in order to reduce internal transport, walking distances and to minimise stock level. Bar-coding w as introduced on the shop floor and linked with the MRP software in a period of 8 weeks. Its application wa s a key factor attributing to the successful integr ation. The team was able to reduce clutter and inefficienc y of the production environment using 5S. Five S is a philosophy and a way of organizing and managing the workspace and work flow with the intent to improve efficiency by eliminating waste, improving flow and reducing process unreasonableness ; 5.0 Conclusion An attempt was made to apply lean manufacturing tec hniques and modularity-in-production, both systems normally adapted in automotive manufacturing, to th e assembly of niche market textile machines. Defined problems were lack of knowledge of manufact uring techniques by management and operational supervisors. A reduced number of assembly activitie s made it more difficult to define a well-balanced work flow. Dispersion of activities and resources betwee n main flow and module flow calls for a need to han dle disturbances quickly. Transport modules were bigger than before and sometimes required special packagi ng. The results obtained showed significant improvement in the production output, surface utilisation, ove rall delivery times, quality and ergonomics. Flexibility in the design process and in the factor y design will help ensure that the production plant can maintain competitive and profitable in the fast changing mac hine building environment. Findings are useful because they form a complement to existing literature on modular process design. M ore research is required on operation of modular and le an assembly processes, in particular for production of niche market weaving machines. References [1] Herron C., Hicks C. (2007),The transfer of selected lean manufacturing techniques from Japanese automo tive manufacturing into general manufacturing (UK) throu gh change agents., Robot Computer Integrated Manufa cturing, doi:10.1016/j.rcim.2007.07.014. [2] Hines P, Holwe M, Rich N. (2004), Learning to e volve: a review of contemporary lean thinking., Int ernational Journal of Operations and Production Management, Vol. 24 No . 10, pp. 994-1011 [3] www.lean.org/whatslean [4] Starr M.K. (1965), Modular production a new con cept, Harvard Business Review, Vol.53 No.6, pp. 131 -142 [5] Feitzinger E, Lee H.L. (1997), Mass customizati on at Hewlett-Packard: the power of postponement, H arvard Business Review, Vol.75 No.1, pp.116-121. [6] Chakravorty S.S., Atwater J.B. (1996), A compa rative study of line design approaches for serial p roduction systems., International Journal of Operations and Production Management, Vol. 16 No. 6, pp. 91-108. [7] Hall, R.W. (1983), Zero Inventories, Dow Jones- Irwin, Homewood, IL, pp. 120-35. 296 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 APPLYING MODULARITY-IN-PRO D U CTION AND LEAN PRO DUCTI ON IN A TERR Y WEAVING MACHINE ASSEMBLY ENVIRONMENT [8] Womack J. (2005), Mr. Ford’s Wrong Turn, Lean M anagement Institute. [9] Spear S, Bowen K.H. (1999), Decoding the DNA of the Toyota production system, Harvard Business Rev iew, Vol. 77 No. 5, pp. 97-106. [10] Heilala J., Montonen J.(2007), Selecting the r ight system – assembly system comparison with total cost of ownership methodology, Assembly Automation, Vol. 21 No.1, pp. 44-54 [11] Fredriksson P. (2005), Operations and logistic s issues in modular assembly processes: cases from the automotive sector, Journal of Manufact.Techn.Mgt., Vol.17 No.2, 2006, pp.168-186. [12] Van Hoek R., Weken H. (1998), How modular prod uction can contribute to integration in inbound and outbound logistics, International Journal of Logistics: Rese arch and Applications, Vol.1 No. 1, pp. 39-56. [13] Sako M., Murray F. (2000), Modules in design, production and use: implications for the global aut omotive industry, paper presented at the International Motor Vehicle Program (IMVP) Annual Sponsors Meeting, October 5-7 , Cambridge, MA. [14] Ulrich K.T. (1995), The role of product archit ecture in the manufacturing firm, Research Policy, Vol. 24, pp. 419-440. [15] Baldwin C.Y., Clarck K.B. (1997), Managing in an age of modularity, Harvard Business Review, Vol. 75 No. 5, pp.84- 93. [16] Baldwin C.Y., Clarck K.B. (2000), Design Rules – The Power of Modularity, The MIT press, Cambridg e, MA. [17] Asan U., Polat S., Serdar S. (2004), An integr ated method for designing modular products, Journal of Manufacturing Technology Management, Vol. 15 No.1, pp.29-49. [18] Kinutani H. (1997), Modular assembly in mixed- model production at Mazda, Transforming Automobile Assembly, Springer, London, pp. 94-108. [19] Blackenfelt M. (2001), Managing complexity by product modularisation: balancing the aspects of te chnology and business during the design process, doctoral thesis , Royal Institute of Technology, Stockholm. 297 298 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES MATCHING HUMAN RESOURCES TO PROCESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES Fang Liu, Richard Weston Loughborough University Abstract The systematic development of computer models of two real asse mbly situations is described. The models are created to quantify business ben efits that can be generated when making manageable manufacturing engineering (ME) changes to process structures, operation procedures, work flow structures, and e specially human resource assignments. The models replicate process behaviour of different configurations of assembly systems based on the use of an actual case study scenarios and plant data. Static businesses models are described in t he form of CIMOSA Enterprise Models to explicitly represent the business environment of the assembly operations; this defines the business context (or ‘workplac e’) for these operations. Causal Loop Models (CLMs) enable reasoning about cause and effec ts impacting on the workplace dynamics of assembly systems. Collectively the EM and CLMs inform the design of Simulation Models (SMs) that can replicate exi sting assembly system behaviours and can generate possible future assembly system be haviours in a systematic and quantitative way. This paper explains how an integr ated use of EMs, CLMs and SMs can inform design and change in assembly systems . Also it explains how SM experiments can be carried out to support the matchi ng of human resources to work-loaded assembly processes. The concepts and modelling m ethod are illustrated in industrial case studies where coherent m odels of ‘role requirements’ and ‘role holders’ enable the design of process improvements. 1.0 Introduction Currently managers of manufacturing industry need t o plan and control an effective realization of prod ucts whilst thinking of better ways of deploying employe es (5). Typically, they aim to improve productivity by managing co-operative work loads within and between different departments. Therefore, improved ways of modelling human systems (including teams, workgroup s and individuals) are required that have potential to delver both business and social benefit in manufact uring workplaces that often are complex and changin g (7). 299 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES This paper reports on the application and testing o f a proposed integrated modelling approach in two distinctive furniture assembly processes. The resea rch created explicit, computer executable models of process- oriented roles and matches them to explici t models of alternative role holders; here role hol ders were explicitly modelled in terms of competencies and pe rformance levels that stereo-typical people can bri ng to the workplace. Case study models were built using Enter prise Modelling (EM), Simulation Modelling (SM) and Causal Loop Modelling (CLM) techniques. When modelling both similarities and distinctions w ere drawn between people and technical resources. A s significant variation occurs in the type or volume of products needed, or if there is significant vari ation in the number of workers available, the aim was to help pr oduction system supervisors to decide how best to r e- assign available resources to roles. In general, pe ople will need to play roles that define their diss imilar responsibilities and distinct positions when collec tively they are required to realise products of var ious types and in variable quantities. For example, the time t aken to realise any given set of products can infor m ‘Lean’ thinking. This in turn has to do with qualities of the people deployed, the roles they play, their ope ration times and the consecutive steps of the process they follo w. A dynamic model can help to understand and analy se unforeseen processing situations. With the help of models, managers can arrange employees so that indu stry can gain the biggest profits. In this paper, prime focus is on creating enterprise (static) and simula tion (dynamic) models of cabinet assembly processes. Als o considered in overview is the role of Simulation Modelling (SM) in support of consultative decision making by reflecting upon case study results. 2.0 Key Literature and the Approach Proposed An assumption researched is that ME enhancements ca n be explicitly specified and designed, and resulta nt production system behaviours can be predicted using simulation technologies. People are the prime resource of any ME, so it is v ital that the interoperation of ME personnel is sui tably systemised and coordinated. Suitably developed mode ls of human resource systems can help specify, implement and maintain timely & cost effective inte r working of human resources through specific ME lifetimes (1, 4). It was decided that it would prove impractical to c reate general purpose models of people. However it was presumed that specific purpose models of human syst ems could be developed and reused. This could inclu de models of individuals, work groups or teams created with respect to systematic work they are required to perform. To create those human systems models, CIMO SA Enterprise Modelling (EM) frameworks and tools were selected as a baseline modelling technique (5) . This choice was made in order to model people and manufacture processes in a coherent way; and by so doing to explicitly model the business and work con text of specific MEs. However it was observed that current EM frameworks and tools are deficient in regard to capturing competencies, behaviours and performances of human systems so that it was anticipated that extended EM concepts would be required. 300 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES As illustrated by Figure 1, previous experience of the authors had shown that CIMOSA EMs could usefull y capture key structural knowledge and data about the specific ME in which any production system and its associated human resources would need to operate. F igure 1 illustrates the modeling ideas where the EM would provide means of creating a graphical model o f the network of processes used by a subject ME, bu t that this graphical model would only encode relatively e nduring ME characteristics. Hence it was decided th at complementary causal loop modeling (CLM) and simula tion modeling (SM) techniques would be needed to analyze current and possible future dynamic behavio rs of subject MEs. Figure 1 Overview of systematic modelling approach- explicit & externalised organisational understandings knowledge &data multiple perspectives & life phases linked to: ‘what to do’,’ ‘how to do it’ & ‘when & how many to do’ different abstraction levels linked to: ‘scope of concern’, ‘time-frame of concern’, etc g en er a t es g en er a t es g en er a t es g en er a t es Step 1: Use of EM to create & validate visual process maps that provide a ‘big picture’ of the reality in the organisation & its segments Step 3: Use of causal loops to identify & design a coherent set of computer executable simulation models Step 4 : Use of SMs to replicate existing behaviours & predict potential future behaviours arising from decision making Step 2: Use of EM concepts & tables to populate process maps with ‘resource’ & ‘work’ data update models update models update models update models real-world knowledge holders various: ‘foci & time-fr ames of concern’; ‘responsi bilities’; ‘roles’; ‘competencies’; etc make change decisions & implement real organisational change & & & li ‘ ’ ’ ‘ i ’ & ‘ & ’ li ‘ ’ ‘ i ’ g n r a t s g n r a t s g n r a t s g n r a t s 1 & ‘ ’ & 3 & 4 & 2 & ‘ ’ & ‘ ’ 1 & ‘ ’ & 3 l l i i & i l i l i l 4 li i i i & i i l i i i i i i 2 & ‘ ’ & ‘ ’ l t l t l ari s: f ci & t i - fr a s f c nc rn ; r sp nsi bi i ti s ; r s ; c p t nci s ; t c ari s: f ci & t i - fr a s f c nc rn ; r sp nsi bi i ti s ; r s ; c p t nci s ; t c & Fig. 1. Overview of systematic modeling approach This paper explains how ideas illustrated by figure 1 were implemented and case tested when designing production systems and their associated resource sy stems. Here the Simul8 software tool (a discrete ev ent simulator) was used to create SMs capable of behavi ours of human systems in a virtual environment (3). 3.0 Modelling the Business Context in a Chosen Case Study Domain The present authors observed that three business mo dels are commonly used in the furniture industry to achieve the realisation of furniture products. Thes e business models are illustrated by Table 1. Key features of these business models were documented. The models a re termed: ‘fixed furniture’; ‘flat pack furniture’ and ’custom furniture’ business models. 301 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES The ‘fixed furniture’ concept (BM1) focuses on stan dard pre-constructed furniture items. Key to a comp etitive BM1 is some appropriate application of lean manufac turing principles (2); renowned for a focus on wast e reduction to improve overall customer value. In the BM2 concept specialist craftsmen realised customis ed furniture by meeting individual desires; BM2 is bas ed on the agility principle which complements lean manufacturing (2). The BM3 concept involves use of a ‘flat pack’ furniture catalogue, with a limited r ange of products with standard dimensions and finishes for users purchasing from ‘do it yourself’ stores. Table 1: Common business models used for home furni ture production BM1 Concept : ca talogued items allow economies of scale to be realised by sellin g enough of each item withi n a ra nge of fixed furn iture via a retail cha i n Outcomes; medium priced furniture of medium quality & medium to high availab ility Business Model (BM) Classes & Numbers of Actor involved in each Business Models BM2 Concept : customised furn iture exemplif ied by furniture portfolios is hand cra f ted to meet in dividual customer requirements by a h ighly skilled’ wood worker’ so as to meet specif i c construct ional, d imensional & f inis h cha rac terist ic s Outcomes; high pri ced furniture of good quality & low to medium availability BM3 Concept : a limited ran ge of pre-pac ked furniture components which conform to a range of furniture products (with st an d a rd dimensions & f in i s hes) are sold via catalogues & retail cha i n s; sellin g enough of each ‘ flat pack’ to realise economies of scale. In th i s model customers can save cap i tal outlay by assembling t he product s Hemselves. Outcomes; low priced furn iture of low to medium quality & low to medium availability Material Input Supply Chain Manufacturer (of fixed products) Retail Chain Custom Builder & Fitter User low multiples multiples large multiples (shops & stockists) not applicable to BM1-done by manufacturer large multiples & & Material Input Supply Chain Manufacturer Retail Chain Custom Builder & Fitter User low multiples low multiples low to medium multiples not applicable to BM3 done by user large multiples & Material Input Supply Chain Manufacturer Retail Chain Custom Builder & Fitter User low multiples not applicable for BM2 not applicable for BM2 medium multiples large multiples & 4.0 Simulation Modelling of Flat Pack and Fixed Furniture Static models describing relatively enduring charac teristics of the business context of flat pack asse mbly process were created by the authors using standard CIMOSA modelling constructs. One of the diagramming templates created (an example CIMOSA structure diag ram is shown in Figure 2. Following which a simulat ion model of flat pack furniture assembly was created, as shown in Figure 3. The reader should note that F igure 2 is one of six CIMOSA diagram templates used to char acterise the network of processes used by supply ch ain partners to realise flat pack furniture. Also it is important to point out that the SM shown in Figure 3 only focuses on one key segment of the flat pack process network; namely the assembly process where users o f flat pack products (in this specific case that of a flat pack cabinet product) are required to assemble the product after purchasing it and before being able to use it . 302 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES CIMOSA conformant Domain Processes (DPs) of the com plete flat pack business model modelled were successfully decomposed until selected process segm ents of reduced complexity that could be modelled effectively using a simulation tool. The process se gments were also further systematically decomposed; via the innovative use of low level function based modellin g concepts (called Functional Operations and Functi onal Entities) until viable process-oriented roles were explicitly defined. Following which competency and performance oriented models of people were built to draw comparisons between the potentials of differe nt human role holders to perform the defined roles whe n subjected to different work load conditions. Fig. 2. Flat pack structure diagram 5.0 Simulation Model 1: Flat Pack Assembly Process Fig. 3. Model 1 experiment 1: Flat Pack Assembly Pr ocess Model 1 (see Figure 3) was designed to analyse the use of human resources and time and cost factors associated with the ‘in the home’ flat pack cabinet assembly case. Here it was assumed after Do It You rself (DIY) shopping a family brings home a flat pack pac kage. The family might comprise one person or a cou ple, 303 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES or one parents with several kids who are intelligen t and/or skilled enough to contribute positively to the overall assembling activity. However as an initial learning problem, the first author decided to feed appropri ate parametric data into the first simulation model so that a set of simulation experiments could predict the time and cost involved when carrying out the cabinet ass embly with different numbers of family members. Res ults obtained from these three sets of experiments with Model 1 are shown in Table 2. Table 2: Predicted Total Cabinet Assembly Times: wi th different numbers of family members doing complementary assembly work One product Operation time (minute) Utilization (average) 1 people 117 5% 2 people 78 3% 3 people 56 2% Although at first sight the time taken by users to assemble flat pack furniture products is not a majo r concern to flat pack manufacturers and retailers it was obs erved that the kind of predictions made in by Table 1 could be useful in helping BM3 supply chain partners to s ell more products; if their product has competitive assembly times this could be a useful selling point . The seller should seek to improve customer satisf action, and this might be one approach to showing quantitat ively that they can do this relative to competitors . Therefore the information generated by this fairly simple SM1 could provide a competitive edge. 6.0 Simulation Model 2: Fixed Furniture Cabinet Assembly During fixed furniture cabinet assembly at a case f urniture manufacturing company, employees draw cabi net assembly work form two queues which are accessed si multaneously so that (1) frame (or carcass) assembl y and (2) drawer construction is carried out. Then th ese semi-finished products come into one store whic h functions to input to the last work centre, adding the plinths and castors to them. Following which, p roducts go to the paint shop where they are finished to meet c ustomer specifications. Fig. 4. Model 2 experiment 1: Fixed furniture Assem bly Process Model 2, experiment 1 (see Figure 4) was designed a nd run based on the assumption that all human resou rces would be trainees, with the same relative low level of payment. In theory there will be an expected co st saving 304 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES relative to the use of skilled and experienced peop le. But the cycle time is expected to be longer bec ause of a lack of key skills and possible lack of sufficient quality of work (and hence a need for rework). For Model2, experiment 2 (see Figure 5) all employees are high level trained. This means a higher salary bill but they can cut the lead time relative to trainees. With their better capability more entry points can be allowed, with greater relative work throughputs, leading to higher revenu es. Fig. 5. Model 2 experiment 2: Assembling process wi th high-skilled workers The authors observed some experimental conditions u nder which improved human resource utilization can be achieved by reducing the number of assemblers in th e process. Hence the same work could be loaded onto fewer workers. It was understood that this might re duce the work throughput but the new arrangement wa s expected to save processing costs; which were poste d on a final financial notice board. But it was obs erved that under favourable conditions the remaining peop le can make better use of their time (such as durin g times when previously they had been waiting) so that they can realise their own previous work plus the work of those made redundant. Fig. 6. Model 2 experiment 3: Modification of asse mbling process by deleting several workers During experimentation it was observed that when re duced numbers of experts were allocated roles it wa s possible for the same operation times to be achieve d as that found when experimenting with experiment 3 305 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES where a full complement of expert operators were as signed roles. Figure 6 explains this modification a nd suggests a possible future human resource arrangeme nt. 0 20 40 60 80 100 Tra ine e Ex pe rt Mo dif ica tio n B E H -40000 -20000 0 20000 40000 60000 80000 Profit Revenue Ope time Figure 7 Results of three models by analyzing the utilization of human resource Figure 8 Results of three models by analyzing the profit, revenue and operation time The line charts of Figures 7 and 8 illustrate the d evelopment of the new arrangement. A typical increa se in utilization of assemblers B, E and H was observed. These assemblers have been assigned more responsibi lity including that of the now redundant assemblers. Rev enue and profit improved with the higher capability of work centres and their assigned assemblers. For mo re entering components of the cabinet are allowed i n the operation and accordingly more products are suppose d to generate which bring higher revenue. Operation time was reduced with more effective use of skilled work ers. 7.0 Futures Predicting New role and competence modelling ideas were genera ted when modelling human resources for the two distinctive cabinet assembly processes. Consider th e ‘Design Specification’ stage of CIMOSA modelling. Because they possess different sets of Functional E ntities (FE), different people will play the same r oles differently. Hence it was considered feasible to di vide real people into several stereo-typical groups , such that they are modelled conceptually as shown in Table 2. Suppose that there are 24 functional entities neede d to realise all roles requirements in the complete cabinet assembling process. Then to each role a candidate r esource type can be assigned, with a known level of competence and performance (designated E=Expert, P= Practitioner, or T=Trainee). Having made these assignments a simulation model can be run to predic t performance (including financial) outcomes. Table 3: people with competencies possessed at spec ific performance levels C1 C2 C3 C4 C5 …… C24 Lily T E P T E T Bill E T E E T P David T E E T P E Mary E P E P E T Ben P T E P E P 306 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES In the real situation multiple product types may en ter the process simultaneously. If people assignmen ts to work are too rigidly maintained, then when a variet y of work comes in then more people would need to b e involved in realising the different work flows thro ugh the process, and their utilisation could fluctu ate significantly. Alternatively if people assignments can be modified in a flexible way then as different products types need to be realised it may be practical to re alise economies of scope; because under-utilised pe ople can be re-assigned for agreed time-frames. But this kin d of flexible working generally requires the use of people with a wider range of skills, so that they can be m atched to workloads which change in character. This is where the systematic approach to decomposing activities i nto elemental functional operations (FO’s) and func tional entities (FE’s) recommended in this paper can lead to well defined role requirements for different wor k types and quantities of work types. Here for each unique product relevant activity flows can be determined a nd the required activities can be decomposed into their el emental FO’s and FE’s. Where commonality of FO’s an d FE’s amongst product types is observed this might s implify the process of allocating people to the pro duct oriented roles so defined. This can be followed by a broader use of tables of people competencies poss essed at specific performance levels, of the type shown in T able 1. Theoretically at least this can provide a methodological and explicit way of enabling rapid a nd effective re-arrangement of the work-force in or der, so that they have the right role assignment at the rig ht time, to maximise value generation and minimise process costs. Normally there might be more than one assemb ler at each work centre, making a team, with the ai m of getting higher efficiency so as to shorten the tota l lead time. Another resource allocation policy mig ht be for one fitter to resource more than one work centre on ce someone is absent for some reason. Accordingly i t is necessary to make a table listing the competency an d performances of each assembler; and to use this t o formalise resource allocation policies, to facilita te computer execution of alternatively resourced pr oduction system models, and to guide training regimes. 8.0 Conclusion A method of improving the design of furniture assem bly systems has been presented in this project. A t ripod of modelling tools consisting of enterprise, causal lo ops, and simulation modelling has been used. This h as provided a methodological way of building simulatio n models of production processes resourced by alternative human system designs and by predicting key business performance characteristics. Also the approach encourages reuse of the models created so that they can be used when alternative and uncertai n product volumes and mixes need to be realised; or w hen process improvements need to be made on an ongoing basis. Importantly the models enable quanti tative decision making about feasible and profitabl e re- distributions of resources The project demonstrated case study examples of the combined use of CIMOSA Enterprise Models and Simul8 Simulation Models to support the design of t wo different real assembly processes, for flat pack and fixed cabinet furniture industry cases. Future project work is necessary to investigated fu rther the use of explicit definitions of process-or iented role requirements and of explicit descriptions of compet ency and performances characters of role holders. F or example managerial, supervisory and co-ordination c ompetency and performance characters could be used to specify capabilities that are possessed by candidat e role holders who could be made responsible for in tegrating 307 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MATCHING HUMAN RESO UR C ES TO PRO CESS- O RIENNTED ROLES IN MANUFACTURING COMPANIES process oriented roles. While ‘change’ competency c haracters may be used to differentiate role holders able to switch their responsibilities and actions performed between different value streams. References [1] Banks J., Carson, John S., Nelson, Barry L. (1996). Discrete-Event System Simulation. New Jersey: Pren tice-Hall, Inc. [2] Hirano, Hiroyuki and Makota, Furuya (2006), "JIT Is Flow: Practice and Principles of Lean Manufacturin g", PCS Press. [3] Kamath, Manjunath. ‘Process-modelling Techniques fo r Enterprise Analysis and Design’. A Comparative Evaluation. Oklahoma State University: Stillwater, OK 74078, USA. [4] Mujtada, M.S. (1994). Enterprise Modelling and Simu lation: Complex Dynamic Behaviour of a Simple Model of Manufacturing. Hewlett-Packard. [5] R.H.Weston, J.O.Ajaefobi and K.A.Chatha (2004). Pro cess thinking in support of system specification an d selection. Advanced Engineering Informatics 18(2004) 217-229. [6] Vernadat, Francois B (1966). Enterprise modelling a nd integration. Principles and Applications. Londo n: Chapman & Hall. [7] Whitman, L.E. and B.L.Hull (1997). ‘A Living Enterp rise Model’. In Proceedings of the 6th Industrial E ngineering Research Conference, Miami Beach, FL. 308 Extended Manufacturing Enterprises: Systems and Tools 309 310 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT Sara Saberi 1 , Charalampos Makatsoris 2* 1. Department of Mechanical and Manufacturing Enginee ring, University Putra Malaysia, 434 0 0 UPM, Serdang, Selangor, Malaysia sa60.saberi@gmail.com 2. Advanced Manufacturing and Enterprise Engineering g roup, School of Engineering and Design, Brunel University * Harris.Makatsoris@brunel.ac.uk Abstract Supply chain management (SCM) is an emerging field that has c ommanded attention and support from the industrial community. Supply chain (SC ) is defined as the chain linking each entity of the manufacturing and supply process f rom raw materials through to the end user. In order to increase supply chain e ffectiveness, minimize total cost, and reduce the bullwhip effect, integration and coordin ation of different systems and processes in the supply chain are required using infor mation technology and effective communication and negotiation mechanism. To solve this problem, Agent technology provides the distributed environment a great promise of effective communication. The agent technology facilitates the integration of the entire supply chain as a networked system of independent echelon. In this article, a multi agent system has been developed to simulate a multi echelon suppl y chain. Each entity is modeled as one agent and their coordination lead to control inventories and minimize the total cost of SC by sharing information and forecasting knowle dge and using negotiation mechanism. The result showed a reasonable reduction in total cost and bullwhip effect. Keywords: supply chain management, inventory manage ment, agent technology, negotiation mechanism. 1.0 Introduction In an era of globalisation, key account management and, products and services designed and delivered a round specific customer requirements, the way supply chai ns are managed distinguishes between success and fa ilure. A supply chain is a network comprising of raw mater ial suppliers, manufacturing plants and warehouses, distribution centres and subcontractors all engaged in the transformation of material into finished pr oducts and the transportation of those within the network and ultimately to end customers (Fig 1). Supply chain 311 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT management becomes however, increasingly complex as there is an increased reliance of businesses in outsourced services which include almost every acti vity from manufacturing to sales and distribution. This leads to increased uncertainty that must be address ed and complex and delicate business relationships that must be managed and maintained. To address this complexity, businesses are explorin g new forms of organization and collaboration mecha nisms to work together with their suppliers and customers . Hence, there is a move away from traditional type of organisation in linear supply chains towards engagi ng in federal partnerships between a number of larg ely independent businesses which aim at virtually pooli ng and time-sharing resources. The aim is to introd uce flexibility in managing and responding to uncertain ty in market place by pooling competencies and shar ing risks when responding to market needs. Such closely -knit federations can be in a dormant or a dynamic state at any time. These networks, however, regardless of st ate are always considered active and ready to respo nd. An arrangement such as this has been described as an a daptive value network (AVN) which is [9]: “An arrangement where companies form a web of close relationships and work together as a system that delivers the right customized product and expected service at the right quality in a coordinated manne r and are responsive and adaptable to changes in the environment.” Among various activities in supply chain managemen t, inventory management is the most important [12] and the effective management of supply chain inventorie s is perhaps the most fundamental objective of supp ly chain management [1]. The advent of software agent technology has trigger ed the development of new architectures and softwar e for modelling and managing the supply chain [7, 9]. Wi th this paradigm, activities in a supply chain such as procurement, planning, execution tracking etc. are represented by a software agent. Each agent acts ba sed on its internal model of that particular activity and interacts with other agents in the network. For exa mple an agent representing a particular warehouse can immed iately provide stock availability information to an other agent representing a particular customer when the l atter queries stock availability before placing an order. This process takes place in real time and potentially wi thout user interference. Fig. 1. Overview of a global manufacturing supply chain network 2.0 Multi Agent System for Supply Chain Management 312 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT Multi-agent system (MAS) is a fast developing infor mation technology, where a number of intelligent ag ents, representing the real world parties, co-operate or compete to reach the desired objectives designed by their owners. The increasing interest in MAS is because o f its ability to provide robustness and efficiency; to allow inter-operation of existing legacy systems; and to solve problems in which data, expertise, or control is distributed. The general goal of MAS is to create s ystems that interconnect separately developed agent s, thus enabling the ensemble to function beyond the capabi lities of any singular agent in the set-up [10]. The agent-based view offers a powerful repertoire o f tools, techniques, and metaphors that have the po tential to considerably improve the way in which people con ceptualize and implement many types of software. Agents are being used in an increasingly wide varie ty of applications — ranging from comparatively smal l systems such as personalized email filters to large , complex, mission critical systems such as air-tra ffic control. It is the naturalness and ease with which such a va riety of applications can be characterized in terms of agents that leads researchers and developers to be so exci ted about the potential of the approach [4]. Multi-agent systems try to solve the entire problem by collaboration with each other. In this way, MAS can help to solve complex problems and make decisions o r support humans to make decisions. Therefore, agen ts are especially suitable for coordination of supply chains due to the following characteristics: 1. Data, resources and control over data and resour ces are inherently distributed [3]. 2. A supply chain is adaptive and changes over time . Agents can serve as wrappers for the supply chain management components owned by a particular supply chain entity [6]. 3.0 Negotiation in Supply Chain Model Companies are required to comply with customer orde rs even if it may be hard to do so. Companies have to respond to the orders quickly and efficiently in th e limited time available to fulfill the customer’s requirements. Unexpected rush orders, however, in m ost circumstances causes delays in delivery and decreases efficiency in all of the supporting membe rs [2]. To coordinate different supply chain entiti es and solve these problems, negotiation decisions have be en identified as crucial for successful global manu facturing [5]. Negotiation techniques are used to overcome conflic ts and coalitions, and to come to an agreement amon g agents, instead of persuading them to accept a read y solution [8]. In fact, negotiation is the core of many agent interactions because it is often unavoidable betwee n different project participants with their particu lar tasks and domain knowledge whilst they interact to achiev e their individual objective as well as the group g oals. The importance of negotiation in MAS is likely to incre ase due to the growth of fast and inexpensive stand ardized communication infrastructures, which allow separate ly, designed agents to interact in an open and real -time environment and carry out transactions safely [11]. 313 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT 4.0 Multi-Agent Model Solution The model proposed in this paper is based on beer g ame four echelons, but assumes unlimited entities a t customer echelon and a single entity at others (Fig . 2). Three levels including supplier, distributor and retailer are allowed to have different inventory systems. Mu lti echelon agent based supply chain is simulated w ith multi agent systems. These agents are also capable of solving the problem of matching supply to demand and allocating resources dynamically in real time, by r ecognizing opportunities, trends and potentials, as well as carrying out negotiations and coordination. Operati ng in a multi agent environment, agent's plan expli citly represents interaction with other agents by exchang ing messages. The agents are implemented as JAVA-th read objects on top of the JADE toolkit that satisfies t he behavior as described above. This means that eve ry agent is a separate computational process with its own in ternal control and a mailbox. Agents can communicat e asynchronously using each other’s mailboxes. The ma ilboxes are implemented as databases with records representing incoming messages. Besides that, all v ariables that represent the state of an agent are a lso stored in a model state database. The simulation model, wh ich have been developed, is flexible enough to add new agents or to edit properties of existing agents to examine auction performance for different trading s cenarios. Fig. 2. abstract structure of simulated model In this model, two types of agents are employed to respond to various types of services for the entire supply chain. One of them is a control agent and the other is a demand forecast agent. These coordinated agen ts have the ability to specify both static and dynamic char acteristics of various supply chain entities. Custo mers generate random and stochastic demands at random ti me and only retailer give customers demand and send them goods. The same manner is employed between dis tributor and retailer and between supplier and distributor (Fig 2). Lead-time between echelons is deterministic and Supplier has infinite recourse. O ther variables like holding and ordering cost is fixed. These assumptions are important and necessary for forecasting agent who is responsible for computing the amount of order and time to order for retailer and distributor based on Economic order quantity (EOQ). Customer unsatisfied orders is changed to lost sale in retailer. For distributor facing a lack of inve ntory, retailer orders will be changed to backorder and sa tisfied, as goods are received form supplier. In th is system, in case of lack of inventory in retailer to satisfy customers needs, retailer agent open a negotiation mechanism which help to solve this problem. By starting the n egotiation mechanism retailer get new prices from customers and with the current orders and these new prices and by using a knapsack program, try to ide ntify 314 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT which customer needs have to be satisfied. In this manner, maximum benefit can be attained for retaile r (Fig 3). Generally, Fig. 4 shows the schematic model of agent-based model and demonstrates all of behaviors in each agent. We can see that there is only one datab ase which is shared for all of agents information. forecaster agent uses these information for prediction based o n new data from other agents and then send new upda te information for retailer and distributor. Fig. 3. Negotiation mechanism between retailer and customer Fig. 4. schematic model of agent-based supply chain CA CA CA Receive goods from supplier and update its inventory Retailer (RA) waits to receive customer order, compare with its inventory (S), and save orders into database Producing stochastic orders (q i) in stochastic times in customers (CA) Start negotiation mechanism between customers who give order and reta iler RACAi Sq >∑ Yes Start RA Send orders to customers and update its inventory Check its inventory to get to replenishment point RpS RA < No No Make a new order for distributor (DA) based on information from forecasting (FA) Yes RA DA Wait to receive request from retailer. Compare its inventory and the request Database DARA Sq > DA Send orders to retailer and update its inventory No Send its entire inventory and change the amount of order that is not satisfied to backlog Check its inventory. If get to replenishment point, request to supplier Yes Save their information in database for forecasting agent FA Based on database information, forecast and send it for all of agents except customers Information for agents Receive goods from distributor and update its inventory 315 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT 5.0 Result and Discussion Supply chain model as a discrete resource allocatio n problem under dynamic environment has been formulated, and demonstrate the applicability of th e virtual market concept to this framework. The pro posed algorithm facilitates sophisticated SCM under dynam ic conditions. This model shows that agent technolo gy can optimize supply chains by (a) reviewing intelli gent agent’s applications for supply chain optimiza tion and (b) illustrating how a multi-agent system can optim ize performance of this network. Model benefits and saving by making use of agent-ba sed systems for supply chain management include: - In retailer agent, fill rate ratio (customers ord er satisfaction rate) calculated. All simulation ru ns showed 95 to 96 percent fill rate, which shows a high costumer s atisfaction (figure 5). - despite of huge and sudden demand in customs orde r, by agent ability, bullwhip or whiplash effect de crease in upper demand, so, total cost dose not increase s o much (figure 6). 8 4 8 6 8 8 9 0 9 2 9 4 9 6 9 8 1 0 0 1 0 2 4 24 44 64 84 104 124 144 164 185 205 225 245 265 285 305 325 345 365 Time fil lra te pe rc e n t Fillrate Fig. 5. Fill rate in retailer 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 8 40 52 64 80 108 128 156 180 204 220 236 264 288 304 329 353 373 Time D em a n d Le v e l Distributor demand Retailer Demand Customer Demand Fig. 6 Bullwhip effect in SC 316 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MULTI AGENT SYSTEM FOR NEGOTIATION IN SUPP LY CHAIN MANAGEMENT 6.0 Conclusion Multi-agent system is a loosely coupled network of software agents that interact to solve problems tha t are beyond the individual capacities or knowledge of ea ch problem solver. The general goal of MAS is to cr eate systems that interconnect separately developed agen ts, thus enabling the ensemble to function beyond t he capabilities of any singular agent in the set-up in agent model. This research can demonstrate that ag ent technology is suitable to solve communication conce rns for a distributed environment. Multi-agent syst ems try to solve the entire problem by collaboration with e ach other and result in preferable answer for compl ex problems. For further works, we recommend developin g this model to have multi retailer and even multi distributor and apply the auction mechanism between them. References [1] Fu, Y., R., Piplani, R.D., Souza, and J., Wu (2000) Agent-based simulation applications: multi-agent e nabled modeling and simulation towards collaborative inventory mana gement in supply chains Winter Simulation Conferenc e, pp. 1763- 1771. [2] Ito, T., and M.R., Salleh (2000) A blackboard-based negotiation for collabora tive supply chain system Journal of Materials Processing Technology, Vol. 107, No. 1-3, pp. 398-403. [3] Janssen, M. (2004) Insights from the introduction o f a supply chain coordinator Business Process Manag ement Journal, Vol. 10, No. 3, pp. 300–310. [4] Jenning, N., K., Sycara, and M., Wooldridge (1998) A Roadmap of Agent Research and Development Autonom ous Agents and Multi-Agent Systems, Vol. 1, No. 1, pp. 7 – 38. [5] Jiao, J.R.., X., You, and A., Kumar (2006) An agent -based framework for collaborative negotiation in t he global manufacturing supply chain network Robotics and Com puter-Integrated Manufacturing, Vol. 22, No. 3, pp. 239-255. [6] Julka, N., I., Karimi, and R., Srinivasan (2002) Ag ent-based supply chain management 1: framework Comp uters and Chemical Engineering, Vol. 26, No. 12, pp. 1755–176 9. [7] Lianga, W.Y. and C.C., Huang (2006) Agent-based dem and forecast in multi-echelon supply chain Decision Support Systems, Vol. 42, No 1, pp. 390-407. [8] Lin, F., and Y., Lin (2006) Integrating multi-agent negotiation to resolve constraints in fulfilling s upply chain orders Electronic Commerce Research and Applications, Vol. 5, No. 4, pp. 313-322. [9] Makatsoris, H., Chang Y. & Richards, H. (2004). “C ollaborative sense-and-respond ICT for demand-drive n value network management.” In: Chang, Y., Makatsoris, H., & Richards, H., Evolution of supply chain manageme nt: Symbiosis of adaptive value networks and ICT (pp. 4 83-514), Boston : Kluwer Academic Publisher. [10] Ren, Z. and C.J., Anumba (2003) Multi-agent systems in construction–state of the art and prospects Aut omation in construction, Vol. 13, No. 3, pp. 421-434. [11] Wooldridge., M.J. (2002) "An Introduction to Multia gent Systems", John Wiley and Sons,. [12] Yung, S.K., C.C., Yang, A.S.M., Lau, and J., Yen (2 000) Applying Multi-Agent Technology To Supply Chai n Management Journal of Electronic Commerce Research, Vol. 1, No. 4. 317 318 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND THE STUDY OF INFLUENTIAL FACTORS FOR DEVELOPING INDUSTRIAL AGGLOMERATION IN THAILAND Somrote Komolavanij 1 , Masatugu Tsuji 2 , Yasushi Ueki 3 , Chawali Jeenanunta 1 , Veeris Ammarapala 1 1. Sirindhorn International Institute of Technology, Thammasat University, Pathumthani, Thailand 2. Graduate School of Applied Informatics, University of Hyogo, Japan 3. Bangkok Research Center, Japan External Trade Orga nization (JETRO), Bangkok, Thailand Abstract The focus of this study aims to identify influential factors to agglomeration and the innovation of the agglomeration by analyzing the data obtained from the survey. Twenty factors were investigated in the survey and the innovati on of agglomeration was identified by check if there were any upgrading in ter m of technology, product and supply in the past three years. It was concluded that the industrial agglomeration of Thailand could be divided into three periods (before 1985, 1986-1998 and after 1999). The earlier was the establishment of the large firms and the later was the establishment of the smaller firms to form themselves aroun d the large firms to become the agglomeration. Some of the factors were found signific antly affecting to the development of Thai industrial agglomeration such as the lar ge firms were attracted by investment incentives, legal systems and skilled labor while the small firms were also satisfied with the government policies in liberal trade and the system of intellectual property rights. However, the innovation of agglomerat ion cannot be concluded clearly since the result from the analysis showed that there were no significant common factors to explain the upgrading of industry among models. Keywords: Industrial agglomeration, Industrial Clus tering, Innovation 1.0 Introduction Strong economic background usually comes from stron g industrial section of the country that is why mos t of the countries try to strengthen their industries. T he strength of industry in each country may come fr om 319 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND different paths. For industrial countries, they hav e originally built their own technology and industr ial system. Until the present time as the world becomes smaller , many companies from industrial countries seek for the new opportunity to invest outside their countries. Non-industrial countries such as many countries in Asia are their targets of investment. Many non-industrial co untries have then turned up to be the new industria l countries; Thailand is one of them. Now, gross dome stic product (GDP) of Thailand is depended on indus trial section rather than on agriculture section as it us ed to be in the past. As the new comer in industry, Thailand has to find the right way to promote industry of th e country in the long term. Industrial agglomeratio n is one of the effective ways to strengthen the industrial sec tion of Thailand. Therefore, to understand the form ation of industrial agglomeration is quite essential for the country to allocate the limited resources to promo te industrial agglomeration. Not only helping in suitable resourc es allocating for the country but also by understan ding the formation of industrial agglomeration, it can help the country to understand the needs of them. By the concept Flowchart Approach [2], the formation of industrial agglomeration can be understood stage by stage. However, the detail of each stage is depended on ea ch country environment [1]. 2.0 Research Objectives and Methodology Foreign direct investment (FDI) has played the esse ntial role for the development of Thai industry. Wi th FDI at the earlier time, now some industrial agglomerat ions have been slowly formed. The agglomeration of industry can strengthen the industry of the country . The focus of this study aims to identify influent ial factors to industrial agglomeration and the innovation of t he agglomeration. The mail survey conducted in Nove mber 2007 to collect primary data for statistical analysis an d econometric analysis. Twenty influential factors were investigated in the survey and the innovation of ag glomeration was identified by checking if there wer e any upgrading in term of technology of production, prod uct, market and sources of supply in the past three years. 3.0 Descriptive Statistics from the Survey The mail survey was conducted in November 2007 by s ending q uestionnaires to 1,800 companies by mail, b y e-mail and some of the q uestionnaires were distribu ted in person by random. The response rate was 8.8% , with 160 valid responses returned and most of them came from management people. 3.1 Profile of Questionnaire Responders 42.5% of the responders are served as Top Managemen t such as CEO, President, Vice President, Business Owner, Managing Director, General Manager. 6.9% are served as Senior Management positions such as Financial Director, Regional (ASEAN) Manager, Manuf acturing Director, etc. 23.1% are in Middle Manager position such as HR Manager, Production Manager, Sa le Manager, etc. 10% are general employees such as Accountant and Engineer. 17.5% of the responders di d not declare their working positions. 3.2 Year of Establishing 320 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND Based on the year of establishing, the ranges of th e ages of the companies are between one year to fif ty years. Figure 1 shows how long the responding companies ha ve been established in Bangkok and/or surroundings area. The conclusion was that most companies have b een established for 11 to 15 years (22%). The secon d most companies have been established for 16 to 20 y ears (18%). 3.3 Capital Structure and Major Investment and Main Business Activities Most of the responding companies are local companie s (53%). 26% of them are joint venture and 21% of t hem are Foreign Direct Investment (FDI) as show in Figu re 2. For those who are join venture companies and foreign companies The most non-Thai investors were form other-Asia, 48%. The second one were EU investors, 21%. ASEAN and USA investors were 17% an d 9%, respectively. From Figure 3 most of the companies’ main business activities were involved i n manufacturing while a few of them were involved i n personal service 3.4 Factors Affecting Business To understanding the formation of industrial agglom eration, the influential factors for starting the b usiness in Bangkok, Thailand were investigated. There are seve ral factors that might affect the investment decisi on of investors. However, in this study, twenty factors a re selected based on from the pre-survey and interv iew on management people in the previous researches [3] as shown in Table I. The factors cover in three areas as social factors, governmental factors and demand/sup ply factors. In the questionnaire, the respondents were asked to identify the importance of factors in the viewpoint of business investors. The levels of impo rtance were classified by five levels as “very important,” “somewhat important,” “not sure,” “not very import ant” and “not important at all.” No. of Year Office First Established in Bangkok 5 % 4 % 4 %4 %2 % 11 % 22 % 18 % 9 % 4 %3 % 5 % 9 % 21 % 1 year 2 years 3 years 4 years 5 years 6 to 10 years 11 to 15 years 16 to 20 years 21 to 30 years 31 to 40 years 41 to 50 years More than 50 years Data Not Available Fig. 1. Ages of the responding firms 321 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND Company Capital Structure 100 % Loc a l 53% 10 0% F or e i gn 21% J oi n t Ventu r e 26% Fig. 2. Capital structures of the responding firms Table I: Factors affecting the business for present and future Number Influential Factors F1 investment incentives including tax incentives F2 liberal trade policy F3 customs procedures F4 local content requirements, rule of origin F5 physical infrastructure (roads, highways, ports, airports, etc.) F6 infrastructure (telecommunications, IT) F7 infrastructure (electricity, water supply, other utilities) F8 government institutional infrastructure F9 financial system F10 legal system F11 protection of intellectual property rights F12 size of local markets F13 access to export markets F14 proximity to suppliers/subcontractors F15 request by large/related company F16 availability of low-cost labor F17 availability of skilled labor and professionals F18 other companies from the same country are locat ed here (synergy) F19 access to cutting-edge technology and informati on F20 living conditions 322 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND 4.0 Econometric Analysis from the Survey Data Based on the econometric analysis, the years of est ablishment of firms in Thailand can be divided into three periods according to the trend in accumulation as f ollows: 1) before 1985; 2) 1986-1998; and 3) after 1999, as seen in Figure 3. This result agrees with the previ ous research done in 2005. [3] The model used to ex plain agglomeration in Thailand defined year of establish ment of the firm as the dependent variable. Size of firms, influential factors, and functions of an office in Bangkok were used as independent variables, as seen in Equation (1). 0 50 100 150 Th a ila n d(A cc u m u la te d n u m be r) 1 96 0 1 97 0 1 98 0 1 99 0 2 00 0 2 01 0 Year Fig. 3 Accumulated number of offices established in Thailand YoE = f (firm’s size, influential factors, function of an office) (1) where YoE = year of establishment In this study, the size of firm could be represente d in the function of the number of employees, firm’ s assets and paid-in capital. Therefore, Equation (1) can be expressed by three different ways in Equations (2) , (3) and (4). YoE = f (The number of employees, influential facto rs, function of an office) (2) 323 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND YoE = f (firm’s asset, influential factors, functio n of an office) (3) YoE = f (paid-in capital, influential factors, func tion of an office) (4) Table I: Results of Estimations: Agglomeration Mode l 1 10,000-24,999(US$)/ 10,000-24,999 (US$) [+] 2 100 - 199/25,000-49,999/25,000-49,999 * ** * ** 3 200 - 299/50,000-74,999/50,000-74,999 4 300 - 399/75,000-99,999/75,000-99,999 * + * 5 400 - 499/100,000-499,999/100,000-499,999 6 500 - 999/500,000-999,999/500,000-999,999 [*] [**] 7 1,000 - 1,499/1 M- 4.9M/ 1M- 4.9M 8 1,500 - 1,999/5M- 9.9 M/ 5M- 9.9M [*] [**] 9 2,000 & above/ 10M & above/ 10M & above 1 Investment incentives including tax incentives [**] [**] [**] [**] [**] [**] 2 Liberal trade policy ** ** ** * ** ** 3 Customs procedures + * + + * 4 Local content requirements, rule of origin [+] [+] 5 Physical infrastructure (roads, highways, ports, airp orts) + [+] [+] [+] [**] [**] 6 Infrastructure (telecommunications, IT) ** ** + * 7 Infrastructure (electricity, water supply, other util ities) + + ** 8 Government institutional infrastructure 9 Financial system 10 Legal system [**] [**] [**] [**] [**] [**] 11 Protection of intellectual property rights ** ** ** * ** ** 12 Size of local markets 13 Access to export markets 14 Proximity to suppliers/subcontractors 15 Request by large/related company 16 Availability of low-cost labor 17 Availability of skilled labor and professionals [*] [**] [+] [**] [**] [**] 18 Other companies from the same country are located here (synergy) 19 Access to cutting-edge technology and information 20 Living conditions 1 Retail/ Wholesale trade [**] [**] [**] [**] [**] [**] 2 Production (raw-material processing) * 3 Production (components and parts) [*] 4 Production (final products) [*] 5 Purchasing/ Procurement/ Logistics 6 R&D/ Consulting 7 Human resources development ** [**] ** ** ** ** 136 143 136 145 136 142 -110.674 -126.518 -112.496 -131.094 -109.073 -121.714 0.199 0.156 0.186 0.138 0.21 0.184 Assets Capita l Full model Selected model Full model Selected model Full model Selected model Employees Note 1: [ ] indicates that the coefficient is negat ive, and items without [] imply the coefficient is positive. Note 2: **, * and + indicates that coefficient is a t the 5, 10 and 20% significance level, respectivel y. The models based on Equations (2), (3) and (4) were analyzed for the significance of the models and th e summary of the result was shown in Table I, it can be concluded that for large companies who came earl ier, “investment incentive,” “legal system” and “availab ility of skilled labor and professionals” are the s ignificant factors that encouraged investors to establish thei r business in Thailand. The function of the office in Bangkok 324 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE STUDY OF INFLUENTIAL FACTORS FOR DELO PING INDUS TRIAL AGGLOMERATION IN THAILAND for large firms at the beginning was more related t o retail and wholesale trade. For smaller firms who usually came later after the large firms, the significant f actors for their concerns when establishing their b usiness are “liberal trade policy” and “protection of intellect ual rights.” In the Flowchart Approach, establishme nt of larger firms is very important since it is the star ting point of agglomeration. However, the formation of the smaller firms around the large ones is important as well since this formation can develop into an indu strial agglomeration later. From the models, it can be explained that the indus trial agglomeration of Thailand could be divided in to two stages. At the earlier stage, the large companies d ue to the attractive investment incentives, legal s ystems and skilled labor established firms related to sales ac tivities in Thailand. At the later stage, attracted by the country’s liberal trade policy and the system of in tellectual property rights, smaller firms followed suit. 5.0 Conclusion In conclusion, the industrial agglomeration of Thai land can be divided into three periods (before 1985 , 1986- 1998 and after 1999). The earlier establishment of the large firms who were attracted by investment in centives, legal systems and skilled labor, the smaller firms – who were also satisfied with th e government policies in liberal trade and the system of intellectual proper ty rights – to form themselves around the large fir ms. Although the result of descriptive statistics show that there are several upgrading of the firms in te rm of new goods, new production methods and new sources of ra w materials supply; the common factor that supports the upgrading is hard to find. Acknowledgement The authors wish to acknowledge the assistance and financial support of Japan External Trade Organizat ion (JETRO), Bangkok, Thailand. References [1] Kuchiki, A., and M. Tsuji (eds). (2005), Industrial clusters in Asia: analyses of their comp etition and cooperation. Place of publication: Pragrave Macmillan. [2] Kuchiki, A. and M. Tsuji (eds). (2008), The flowchart approach to industrial cluster policy, Place of publication: Pragrave Macmillan. [3] Tsuji, M., Ueki, Y., Miyahara, M. and Komolavanij, S., (2006) “An empirical examination of factors pro moting industrial clustering in Greater Bangkok, Thailand” , in Proceedings of 10 th International Convention of the East Asian Economic Association. City and country: Sponsor of convention 325 326 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 1 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES S N Khalil, R H Weston Manufacturing System Integration (MSI) Research Ins titute, Wolfson School, Loughborough University,UK Abstract Human systems are an important part of any organization . Recently at tention has been paid to modeling various aspects of people organized into pr oduction systems including production performance, efficient resource allocation and op timum resource management. In this research graphical and computer executable models of people have been conceived and used in support of human system engin eering. The approach taken has been systematically decompose processes and their e lemental activities into explicit descriptions of roles that potential human role hold ers can play. This approach facilitates quantitative analysis and comparison of diffe rent human system configurations that suit various manufacturing workplaces. The pap er describes how the approach leads to the design and runtime simulation of hu man system. In this paper, the researcher illustrates the application of this ap proach and observed advantages gained from the use of simulation technologies. This pap er described how the models enable prediction of the relative performance of alternative production system design comprises people and machines allocated to process oriente d roles. Keywords: Human system, manufacturing enterprise, s imulation modeling, process-oriented. 1.0 Introduction The literature review shows that ME’s (Manufacturin g Enterprise) are subjected to increasing dynamic impacts arising in the business environment in whic h they operate. To address these kinds of concern manufacturing philosophies like Agile Manufacturing [1], Group Technology[2], Reconfigurable Manufacturing Systems (RMS), Mass Customization & P ostponement and Holonic[3-5] manufacture have emerged to uniform ME’s how to achieve increased fl exibility and responsiveness. However, on general t hese philosophies are only supported by limited implemen tation tools to quantify relative benefits of choos ing alternative philosophies; and more particularly in the context of this research paper, to relative qua ntify benefits of alternative ways of resourcing process oriented roles in accordance with selected philosop hy. Also observed is that despite significant advance in bes t practice complex systems engineering , as yet in industry there is no model nor coherent means of modeling or ganizational structures and the related time based 327 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 2 behaviors of the human and technical (machine and I T (Information Technology)system) resources. On the other hand, various modeling techniques have been d eveloped to characterize machines (and their competencies and behaviors) and these techniques be comes commonly implemented using virtual engineerin g and simulation NC (Numerical Control) and robot sys tems of resources. But generally because ME modelin g to human system is so complex and generally the sof tware support tools are somewhat special purpose e. g. to model kinematics and ergonomic characteristics 2.0 The Scope and Focus of the Study This paper addresses the question: given a well def ined set of process-oriented roles how best should work roles be resourced? In this respect it is assumed t hat either (1) people or (2) some for of machine or IT system or (3) some combination of (1) and (2) will prove m ost effective; and that generally these kinds of ‘a ctive’ resource’; will be constrained in terms of their av ailability short and long term. Also assumed is tha t (a) the nature of roles and (b) the works loads placed on t he roles will determine the most effective match of ‘role holders’ to ‘the defined set of process oriented ro les’ Furthermore it is assumed that because the wor k loads in ME’s are typically determined by customers and rela ted factors in the ME’s environment then these work loads will frequently change. This latter points provides a baseline rationale for this study in that an imp roved systematic method and supporting modeling tools are needed to compare the match of different choice of candidate human and technical (active) resources to process oriented roles and their workloads; and al so that the request method and tools should conform short t erm planning of resource deployment as well as long er term strategic estimated (and risk taking) in achie ving people and technical resource systems. Fig. 1. Human system modelling in ME Figure 2 illustrates the systematic modelling appro ach under development by the authors. The underlyin g idea is to create multi-perspective models that can be c omputer executed in the form of simulation models ( SMs) such that they can provide a computer tool to infor m ‘ongoing planning’ and ‘longer term investment’ d ecision ME Processes & Workflow Perspectives • P1- Complex process network, organized into roles • P2- Dynamic workloads Resources perspectives • P3 Candidate holders/roles S tereotype Resources + Actual Resources + R 1 R 2 R 3 SM SM- Simulation Modeling R- Roles P-Perspectives Input work Output work 328 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 3 making leading to effective use of human and techni cal resources. Here specific ME models related to perspective P1 are created using an EM (Enterprise Modelling) technique which is geared toward specify ing sets of ordered activities (or process models) that can be decomposed into explicitly defined roles. T hese roles and role relationships specify a process oriented r elatively enduring structure for any ME being mode lled. A second dynamic workload perspective P2 is derived f rom (a) analysis of historical patterns of work tha t previously have passed through the defined roles an d/or (b) a forecast or prediction of likely future workloads. The third perspective P3 relates to candidate role holders in the form both of stereotypical and actua l human and technical resources. Here modelling can be with respect to (i) known competencies (of people) and capabilities (of machines), (ii) capacities and/or performance levels (of both resource types) and (ii i) psychological behaviours (of people). This multi-perspective modeling approach is designe d to enable: (I) independent change to the three perspectives P1, P2 and P3; (II) reuse of models of ME’s in the form of process and enterprise models; and (III) ongoing systematic reuse of models belonging to those three viewpoints, as required in support o f short, medium and longer term ME decision making. 3.0 Choice of Study Domain A bearing making company was chosen as the domain o f study. The products of this company fall into fou r categories which relate to the shape and dimensions of the raw material used to make them. The categor ies are: ‘Flat sheet’ , ‘Strip sheet’, ‘Round (narrow)’ and ‘Round(wide)’ products. The product type chosen for detailed study is the ‘Flat sheet’. Figure 3 explic itly portrays how the business process (BPs) and En terprise Activities (EAs) lead to flat sheet production. The human resource utilisation in each of the proce sses was determined by using the following theoreti cal equation: Number of operators needed = Total Cycle Time (TC T) / Takt time Hence predicted number of operators needed for each processing flat and round sheet product types was observed to be: Table I: Product TCT , Takt time and number of oper ators Product Types TCT Takt time No of operators Flat sheet 364 10 7 3 Strip Sheet 292 107 2 329 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 4 The TCT and Takt times were measured using time stu dy principles at the bearing company. These time st udy results refer to the ‘as is’ process in the activit y diagram as shown in Figure 3. 4.0 Enterprise Modeling and Simulation Modeling in Human System Models The previous researchers in MSI had developed CIMOS A[6] models for the bearing company. These models ‘context diagram’, ‘interaction diagrams’, ‘structu red diagrams’ and ‘activity diagrams’. Enterprise A ctivities were further decomposed into Functional Operations (FO) and Functional Entities (FE). The selected pro cess for this research is shown in Figure 3. This activi ty diagram concerns ‘Flat Sheet’ making, where at a chosen level of abstraction flat sheet making enterprise a ctivities are defined. These activity models can be semantically enriched with models of precedence rel ationships and information requirements for activit y execution. ACM Bearings Organisation Design Research Institute Title: Number: Design by: Version: Last update: 2 Events Physical Resource FinanceHuman ResourceInformation External Links Flow of Res./Mat. Flow of ProcessActivityCIMOSA Domain Non-CIMOSA Domain Context Diagram Site Map Alternative Flow K.A.Chatha Ver-1 19 – 09 -2005 Press Heat Laminate Sheets EA5211-14 Laminates Clean the Laminate Table EA5211-17 Empty Press and Clean it EA5211-15 Flat Sheets AD52 1.1 [Produce Raw Materials (Flat Sheets)] - Activity Diagram Produce Bearings Domain Take Job Card EA5211-1Production Supervisor Round Product Worker Collect Specified Cloth EA5211-2 Job Card Change Cloth EA5211-4 Specified Cloth Collect, Measure & Mix Chemicals if necessary EA5211-7 Job Card Wrap Wax Paper on Mandrel EA5211-8 Mixed Chemicals Wrap Cloth on Mandrel EA5211-9 Setup Bath for Flat Products EA5211-3 Set Mandrel EA5211-6 Mandrel Cut Cloth & Squeeze out Excessive Chem. EA5211-10 Setup Laminate Table EA5211-5 Take Wrapped Mandrel to Laminate Table EA5211-11WrappedMandrel Lay & Cut Laminate Sheets to Size EA5211-12 Wrapped Mandrel Required # of Sheets achieved no yes Wrap Laminates & Put in Press EA5211-13 Clean the Bath EA5211-16 Job Completion Next Job Needs Same Chem. But Not Same Clothyes no no yes More Chem. Needed yes no Fig. 2. Activity diagram- Flat sheet making These processes are then used as a reference for bu ilding simulation models (SMs). Two simulation mode ls SM1 and SM2 created in this way are described in th e following, but both share a common parent process described by the activity diagram of Figure 2. This process description defines processing routes foll owed by workflows in the SMs. At each work centre (WC) of S Ms resources (human, machine etc) are needed to realise EAs assigned to each WC. The simulation is laid out by referring to the activity diagram and t hese activities are represented in the simulation models with the icons such as ‘single proc’ for single pr ocesses, ‘workstation’ for work place or worker at processes , ‘workpool’ is for the place where all workers are gathered and assigned by a nominated supervisor known as a ‘ broker’. In the simulation models (SMs), enterprise activities numbered EA5211-1, EA5211-2 AND EA5211-1 are categorised into a Pre wrapping role for WC1. The process executed at WC2 are for a Wrapping role performed by EA5211-4 : then at WC3 the EAs numbered EA5211-5 to EA5211-11 are categorised toge ther as a PreLaminate role: then at WC4 a Laminate role is performed, EA5211-12; then at WC5 the post- laminate role is performed via EAs numbered EA5211- 14. Finally at WC6 the clean role is executed via E A5211-15. 330 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 5 The human/worker configurations for both simulation s are the same. In SM1, workers are assumed to be 7 5% - 85% efficient leading to ‘as is’ processing times at each process. The throughput rates and bottlene ck at each workstations were then viewed when the simulation m odels were run. SM2 input dynamics were changed so that the performance of the workstations and the wo rk pool in SM2 can be compared to the previous simulation model, SM1. The layout of the experiment is as portrayed in Fig ures 3 to 6: Fig. 3. Simulation Model 1 (SM1) Fig. 4. Simulation Model 2 (SM2) 331 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 6 Fig. 5. Results from SM1 Figure 6: Results from SM2 5.0 Observation and Conclusion The simulation experiments were found to replicate and predict overall effectiveness of enterprise act ivities when they deployed resources in different ways. Res ults obtained are shown in the graphs of process dy namics for both simulation models where bottlenecks change as the input of the processes varies. The time wai ting and blocked situations increase in SM2 relative to SM1 and resource utilization in the work pool decre ases by 2%. From this research, it is found that the simulation modeling approach can build upon information previ ously encoded by static models represented by CIMOSA acti vity diagram. The approach developed new understanding that enterprise activities can be mod eled and used to characterize the dynamic of the ov erall systems which includes human system dynamics. Thus these models can be used to predict the impact of t he overall performance of manufacturing enterprise due to product variance. Use of alternative SMs and their experimental input s enables analysis of a particular process segment of MEs. Several candidate behaviors can be modeled by chang ing the resources assigned to work centre roles. Al so, it 332 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 H UMAN SYSTEMS MODELING IN SUPP O RT OF MANUFACTURING ENTERPRISE ACTIVITIES 7 was observed that both EM and SM enable the execut ion of ‘to be’ production system behaviors with different human system configurations. The study is recommended for designing or re-designing manufacturing systems such as new product developme nt or the enhancement of existing production system s. Acknowledgment The principal researcher is currently a PhD researc her in Loughborough University and sponsored by Malaysian Ministry of Higher Education/Universiti T eknikal Malaysia Melaka References [1] Van Assen, M.F., Agile-based competence management: the relation bet ween agile manufacturing and time-based competence management. International Journal Agile Management Systems. 2 000. 142-155. [2] Hon, K.K.B. and H. Chi, A New Approach of Group Technology Part Families Op timization. CIRP Annals - Manufacturing Technology, 1994. 4 3 (1): p. 425-428. [3] Gou, L., P.B. Luh, and Y. Kyoya, Holonic manufacturing scheduling: architecture, coo peration mechanism, and implementation. Computers in Industry, 1998. 3 7 (3): p. 213-231. [4] Zhang, X., et al., Design and implementation of a real-time holonic control system for manufacturing. Information Sciences, 2000. 1 2 7 (1-2): p. 23-44. [5] Rodriguez, S., V. Hilaire, and A. Koukam, Towards a holonic multiple aspect analysis and modeling approach for complex systems: Application to the simulation of i ndustrial plants. Simulation Modelling Practice and Theory, 2007. 1 5 (5): p. 521-543. [6] Vernadat, F.B., Enterprise Modelling and Integration. 1996, London: Chapman and Hall. 333 334 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES THE IMPACT OF INDUSTRY GROUP CULTURE ON THE DEVELOPMENT OF OPPO RTUNISM IN CROSS-SECTOR ALLIANCES Susan B Grant School of Engineering and Design, Brunel University Abstract Over the last two decades the world economy has dramatically tran sformed, with strategic alliances and partnerships across industrial and glob al boundaries becoming an important means to maintaining and regaining competitive positi oning. In spite of an increase in partnership activity, alliances continue to face problems fuelled by factors such as partner opportunism, and cultural incompatibil ity. This paper highlights the emergence of opportunism in alliances arising from cross-sectoral partners’ differences in cultural values and norms. The li terature indicates that cultural differences are important factors for understandin g the behaviour of managers across sectors. Keywords: Industry group culture, cross sector sup ply chain alliances, opportunism, Product service systems. 1.0 Introduction More and more firms are realising the growth opport unities in developing alliances and partnerships bo th within and across industrial sectors. With this has come the recognition both amongst the academic com munity and supply chain practitioners, that a vital elemen t to effective alliances is not only the ability to pick the right channel members and then establish and develop stro ng integrated relationships with them [1] but also to consider the cultural mindset of partners, which is shown to be critically important [2]. Indeed, much research now recognises the value of understanding the cultu ral context of organizations, and the importance of bridging cultural differences in buyer supplier rel ationships. In particular, it has become important to consider the influences that come from beyond the specific c haracteristics and relationship of the parties to a n alliance, in the form of institutional arrangements that gove rn and constrain parties’ behaviours (such as infor mal rules, roles etc,)[3]-[5]. Different ‘cultural spheres’ o r distinct industry groupings, may share cultural v alues (such as trust and trust building) which may have a role in sustainability of alliances across such groupings [ 6] [7]. In a similar way, distinct industry groups may operate u nder a set of norms which lead to behaviours/attitu des that are not conducive to long term strategic partnering [5][8][9] within their own cultural spheres and wi th partners across sectors. Not only is a high degree of cultural consistency necessary for long term st rategies to 335 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES be successful [4], but also the right cultural norm s, values and expectations need to be generated and endorsed by the dominant players within an industry. In the light of the diversity of cultural norms, expectat ions, rules etc, in cultural groupings, building and managing a lliances across sectors and country boundaries can prove difficult. Even though a written agreement may be s igned by each partner, the real foundation underpin ning these alliances is based on trust, commitment and c o-operation between the parties. Given that such a lliances are based on trust, commitment and co-operation, th e partners’ perceptions of these attributes need to be congruent so that expectation on either side of the dyad will be reasonably similar. When alliances p artners share a similar cultural background, such consisten cy in expectation should be a matter of course beca use each would most likely share the same culturally defined norms of what defines trust, commitment and co- operation. But what happens when the cultural backg round of each channel partner differs as in the cas e of alliance members across industrial sectors (PSS’s), and across countries such as exporters with foreig n distributors [10]. 2.0 Role of Industry Group Culture Authors from a variety of backgrounds have develope d definitions and frameworks of organizational cult ure. Culture represents a shared set of meanings and und erstandings about the organization and its issues, objectives and practices [11]-[14]. Whereas culture may be visible via rituals, dress codes, stories, physical layout rules of conduct [15], these represent the o vert behaviours and other physical manifestations o f their organization [11]. At an even deeper cultural level , are the underlying assumptions, such as beliefs, habits of perceptions, thoughts and feelings that are the ult imate source of values and action [2]. Conceptualisation of culture remain problematic, an d continue to be greatly contested in different lit eratures. For the purposes of this paper, a definition of gro up culture is presented below. Group culture has be en defined as a group level phenomenon consisting of a set of shared, taken for granted implicit assumpti ons that a group holds, and influences how the members of th e group understand and respond to their environment . The content of ‘culture’- the specific assumptions, nor ms and values of the culture- shapes members’ patte rns of behaviour [13] and creates an environment in which attitudes/behaviours are generated, endorsed and to lerated [5]. This paper is concerned specifically with industry group cultures that incorporate a strong tolerance to behaviours such as opportunism. Given that strategi c alliances are sites in which tensions between co- operation and competition naturally occur [16], in those alliances where players operate under differe nt norms from one another, the result may be conflicting beh aviours and attitudes, leading to an intrinsic vuln erability and inherent instability of the alliance. Post con tractual opportunism in alliances can have severe c osts and consequences ranging from tentativeness in commitme nts, to alliance dissolution [17]. The costs of opportunism don’t end there as it can have a system s’ efficiency cost as total supply chain loses cred ibility in the eyes of the end user [18] Opportunism or cheating in the context of alliances refers to intentional self interest seeking at the expense of an other(s) assuming a prior in(formal) contract ha s been struck between parties. Typically in any tra ding arrangement, a minimum set of obligations will have been codified in a formal agreement. Under partner ship, 336 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES a fuller set of obligations is generally made infor mally. Intentionally failing to honour those obliga tions represents cheating, or opportunism. The assumptions surrounding the emergence of behavi ours such as opportunism seem to be generally based on the immediate or local characteristics of the allia nce dyad or partnership, such as ‘lock in investmen ts’ [19], [20] loopholes in contracts [21], difficulties of t erminating long term distribution contracts easily or cheaply [22], information asymmetry [23], and performance a mbiguity between partners [24] [25], decision makin g uncertainty and cultural inconsistency , psychic di stance [26] [27] monitoring, power [28]. The predominant proponents of this approach are typ ically rooted in economic theories of behaviour [29 ]of TCE [22]; [25], [30] [24][31]-[33. Today, many supp ly chain theorists investigating relational risk ar ising from opportunism in exchange have been and still ar e being influenced by literature that implicitly tr eats opportunism as a partnership level phenomenon. In contrast, very little consideration is given to the wider market group setting that players are par t of, with little or no account of the power of ‘group culture ’ that organizations knowingly or unknowingly are e mbedded in, which may be crucial in understanding why oppor tunism is more evident in some channel relations an d some sectors. This is surprising given that economi c behaviour, and in particular opportunism has been demonstrated to be largely constrained by social re lationships or institutions, with shared beliefs, n orms and mores (culture), [34][35]. Indeed, many studies dem onstrate the impact of cultural norms operating wit hin close-knit communities on the emergence of opportun ism, and imply the existence of some sort of broad or local community effect on behaviour or attitude [36 ] [9] [37] [5]. This is significant to cross indust ry alliances and partnerships that operate under different cultu ral norms. Despite this a limited amount of research has been conducting on understanding opportunism as a group based phenomenon, highlighting the need to examine the wi der market context of economic transactions, the informal social relations and obligations that econ omic behaviour are embedded in, that provide the ba sis for behaviour, or the community pressure towards confor mity and in-group norms [9]. As noted above, little theoretical or empirical att ention has been paid to understanding how opportuni sm can emerge within and across different group cultures. Indeed, little research has examined opportunism as a result of cultural norms. One potential research aim is to explore how different industry cultures come to us e opportunism as a strategy, and enact its developmen t, both internally, (within the industry sector cul ture) and externally (across industrial sectors and cultures) . How do different cultures’ ‘shared beliefs, ‘fabr ics of meaning’ dictate (in)appropriate beliefs attitudes and behaviours about opportunism among its members? How can apparently conflicting cultural norms, beliefs or attitudes be reconciled so that trust may develo p? 337 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES 3.0 The Role of Industry Group/Organizational Culture in Influencing the Level of Co-operation/Competition across Supply Chain Relationships There is an increasing amount of literature which h ighlights the notion that cultural differences (of a group) affect the dynamics of partnerships in different co untries [39] [39]. Johnson, Cullen, Sakano and Tak enouchi [40] concluded that developing cultural sensitivity and cross- cultural preparation was necessary when entering international alliances. Cultural sensitivity was t he extent to which managers understood and adapted to differences in their partner firm’s culture. Kim a nd Oh 2002 found that a supplier needed to understa nd the importance of societal and cultural influences on d istributor behaviour, especially with regard to com mitment and its impact on performance. Culture has been sh own to cause differences in the perception of ethic alness of negotation tactics [54], Boone & Witteloostuijn [53 ] found that culture has a substantial impact on th e co- operative or non-co-operative behaviour of American and Dutch participants in a Prisoner’s dilemma set ting. In the US construction sector, partnering is seen a s a way to achieve an optimum relationship between client and a contractor, it can be treated as a moral agre ement that facilitates effective resolution of prob lems and conflict without destroying the harmony between the clients and the contractors. For partnering to su cceed, Black et al [55] pinpointed that developing trust a mong partners is the most important factor. However they also indicated that “few industries suffer more fro m conflict than construction’. As such Hawke sugges ts that building mutual trust in construction is a myth, mi strust has been overwhelmingly deep seated and long standing, and seems to have become the acceptable y ardstick upon which to base transactions” [52]. Similarly in the UK grocery sector also, [5],[8],[1 8] found opportunism and mistrust to be endemic to the sector. Hardy and Magrath (1989) identify cheating as commonplace and indeed endemic to a whole indus try. In particular sectors of retailing, such as consume r grocery packaged goods, a great many sales promot ions negotiated with suppliers by retailers in order to encourage pass along savings to end customers, in f act never occur [18]. Cheating on payment terms is also commo n in channels. This is widespread in markets where huge retailers hold great power over suppliers. Grant’s(1989) exploratory research highlighted the appearance or relative absence of opportunism withi n and across sectors appears to be strongly linked to var iations in sector/market and sub market ‘culture’ characteristics, and the level of tolerance towards opportunism defined by market/group members and expectation of it. The findings reveal market membe rs define and enforce boundaries of ‘unacceptable’ behaviour using unofficial rules and sanctions, unw ritten penalties, standards of conduct, and concern s for reputation, which become embedded and transmitted v ia informal social and market networks. These differences create variations in-group member toler ance and vulnerability towards opportunism occurrin g not just between industrial sectors or markets, but als o within sub-markets. The Social Construction view also suggests that ‘ra tional maximizing’ (which manifests itself as oppor tunism) is not a fundamentally human drive or instinct, but rather a socially and culturally defined strategy. As such this perspective explicitly rejects the dominant ec onomic notion that levels of opportunism are nothin g more than the sum of individual actor's independent pref erences [9], or notions of psychological or moral inclinations of individuals. The idea here is that the differences in levels of opportunism that are observed 338 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES among cultural groups such as markets or sectors ar e explained by social conditions in each group/sect or. Opportunism in some groups like all other economic behaviours, is embedded in a specific social and cu ltural milieu. This theory further proposes that groups (i .e. within one market), can swing between periods o f opportunism and restraint. In financial markets, f or instance, market makers are pulled between the s hort-term attractions of extraordinary profits and the long t erm benefits of prudence. It is a dynamic process a ffected by pressures shaping behaviour. On the one hand, self interest as well as competitive pressures compel t raders to seek to maximize profits in the short term through the locally available strategies of aggressive trad ing. This aggression may sometimes slip over the line into op portunism. On the other hand, self- interest, as we ll as social pressures, compel traders to preserve their income stream in the longer term by maintaining interpersonal relationships and attractive markets. This elicits strategies of restraint. Similar explanations for group opportunistic behavi our can be found in the Business Ethics literature. While ethics are basically personal, and thus have to do with the behaviour of individuals, many authors bel ieve behaviour is influenced by the norms of the environ ment and the peer group to which a person belongs [ 41]. In business environments, Plank et al [42] suggest tha t purchasing professionals approach ethical decisio n making with a set of values related to the socializ ation process both inside and outside of their prof ession. Often in business an individuals ethics may be misa ligned with the ethics of the firm market or indust ry. Indeed, what may be disapproved of at a personal le vel, is often disregarded at a business level, and even accepted without comment. This is often regarded as ‘clever or sharp business’. All that matters is th at the company makes the maximum profit [43]. Studies have often shown that individuals with strong personal moral and ethical values do not maintain the same v alues in the business community [44]. Bowman & Carroll's [45] research similarly found that people feel under pressure to compromise their personal s tandards in order to achieve the goals of the organization. A survey by Pitney-Bowes Inc (1989), a manufacturer of business equipment, revealed that 95% of its manage rs feel pressure to compromise personal ethics to a chieve corporate goals. A similar study on Uniroyal manage rs, found 70% feel pressure to compromise ethics (w hich included types of opportunistic behaviour [46]. Mos t managers at the above companies believe most of t heir peers would not refuse orders to market off-standar d, and possible dangerous products [46]. These and other studies suggest unethical behaviour (including oppo rtunism) can stem from pressure imposed by superior s, absence of a corporate ethical code, the industry e thical climate and behaviour of peers [47]. Others however, within this discipline have focused on behavioural attributes as the root to unethical behaviour. For instance, Newstrom & Ruch [48] sug gest managers are motivated by self interest and wi ll therefore have a propensity to act unethically if i t is to their advantage, and if the barriers to une thical practices are reduced or removed. In addition to the influence of the climate (enviro nment) of the organization, the role of top managem ent, superiors and colleagues, the ethical level of beha viour is determined by the existence of limited (or a shortage of) productive resources [44]. These ‘economic’ con siderations are echoed in the findings by Ulrich an d Thielemann’s [49], who suggest the majority of mana gers take into account both ethics as well as econo mics in specific decision making problems. The implication is that ethical demands must not jeopardize a compa ny's continued economic success, as a company incapable of satisfying the requirements of the market will disappear from the scene before long. Managers reco ncile the requirements of achieving economic succes s with the ethical demands of which they are responsi ble, by legitimising their activities. As such ther e is no 339 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES average ethical profile for a manager but a whole r ange of thinking patterns that managers apply in th e field of tension between ethical and managerial aspects [49] . In this study the authors use the concept of ‘eco nomizing’ and opportunism synonymously, implying self-interes t seeking is not necessarily unethical, it can be j ust good business practice. Empirical findings on ethical business research rev eals unethical practices arise from a variety of so urces, including accepted practices within an industry. Th is may be the result of falling ethical standards s o that practices once considered unethical are no longer v iewed as such. Examples include the substitution o f materials without customer knowledge after the job has been awarded; mis-representing the contents of products; scheduled delivery dates that are known t o be inaccurate to get a contract [50]. Competitive pressures from outside the organization can also pu sh ethical considerations into the background. Soc ietal forces have also been blamed for causing lower ethi cal standards. These include an increase in social decay, materialism and hedonism, loss of church and home i nfluence, competition, current economic condition, political corruption; greed, desire for gain, ‘wors hip of the dollar’ as a measure of success, selfish ness of the individual, lack of personal integrity and moral fi bre; greater awareness of unethical acts, and TV/communications creating an atmosphere for crime [51]. This literature highlights the implications for the sustainability of ‘co-operative partnership’ acros s players operating under diverse societal and group cultural norms and values. Indeed can ‘partnerships’ across sectors ever be stable and mostly free from opportunistic a buse in those markets where high levels of toleranc e as defined by its members are considered acceptable? I n the light of these studies, and given the continu ing rise in ‘partnership structures’, and ‘outsourcing’, wit hin and across industry sectors, and the lack of re search comparing ‘market culture’ systematically across in dustrial sectors or geographical areas of a country , an extension to this work is timely and important. In the light of the literature presented, the resea rch proposes the following propositions: 1. Firms will tend to engage in activities and beha viours that reflect or at least are consistent with their values, culture, and predominant market group culture [5]. 2. Predominant industry group culture (values, norm s expectations, social obligations etc) influence a firm’s relational behaviours towards its partners in strat egic alliances. Market values play a role in influe ncing the types of relationships that will be established, an d determine the level of co-operation that can exis t between partners.[5]. 3 Cross-sectoral 'relational exchange' partners who do not share the same or similar market cultural v alues are less likely to be sustainable in the long term and vice versa. This may have implications for the sust ainability of cross sector/country partnership alliances. Given the continuing rise in inter-firm co-operatio n within and across industry divides, and the accel erated rise of global sourcing, this new corporate paradigm of increasing firm dependency on external organization s for 340 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES critical business processes, services, components a nd raw materials means that effective relationship management is crucial [29] [28]. Indeed, it is thro ugh understanding what causes behaviours such as opportunism to emerge, or what inhibits opportunism that practitioners are better able to predict and manage partners behaviours better. Failing to recognise th e impact of industry group cultures on behaviour by managers, means the knowledge of strategies for man aging potential opportunism remains incomplete. The implication for managers who disregard the wider cu ltural impact on behaviour (arising from group norm s, obligations etc) on individual players, may result in unnecessary costs of controlling behaviour[34], loss of reputation [28] and increasing costs. Additionally, the assumptions made about a partners’ behaviour w ill have an impact on relationship management strategies tha t managers select. I.e. maybe they are less open th an they would normally be, wary less willing to divulge con fidential information etc. The study suggests that greater understanding of the cultural and market group impl ications of collaborative working arrangements are needed if practitioners are to manage collaborative buyer- supplier relationships whether across the UK or glo bally. 4.0 Conclusions As co-operative activity amongst organizations, bot h within and across national and international bord ers acclerates, more needs to be known about whether pr edominant/strong social, or market based values, obligations and expectations within which transacti ons are conducted and embedded can influence the r isk of behaviours such as opportunism arising. Building tr ust, commitment and co-operation whilst refraining from opportunism among supply chain members operating in domestic channels, where the partners come from similar group cultures can be challenging. In those contexts where cultural backgrounds may be dissimi lar as in PSS’s for example, the challenge will be even gr eater. In short where cultures are substantially di fferent, trust, commitment and co-operation are more difficu lt to attain [10], and sustainability in alliances more difficult to achieve. Without a sound understandin g of the foundation of culture that influence behav iour, it will be difficult to successfully implement supply chain initiatives. Without the understanding of the importance and significance of cultural diversity a mongst partners involved in cross-sectoral partners hips, the sustainability of these alliances cannot be guarant ied. References [1] Fawcett, S.E., Magnan, G.M., Williams A.J. (2004) i n ‘Supply chain trust is within your grasp’, Supply Chain Management Review Vol8, No2 pp20-6. [2] Mello, J. E., Stank, T P (2005)’ Linking Firm Cultu re and Orientation to Supply chain Success’ International Journal of Physical Distribution and Logistics Mana gement Vol. 35 No.8 pp 542-554. [3] Costa A C (2003) Work team trust and effectiveness’ Personnel Review 32 issue 5 pp605-622 [4] Mc Afee RB, Glassman M, Honeycutt ED (2002)’ The ef fects of Culture and human resource management poli cies on supply chain management’ Journal of Business Log istics Vol23 no.1 pp1-18 [5] [Grant S B (1998 ) ‘Supplier’s Dilemma: Mixed motives in co-operativ e exchange’ PhD Thesis, Henley Management College UK. [6] Doney PM, Cannon JP and Mullen M (1998) Understandi ng the influence of national culture on the develop ment of trust’ Academy of Management Review Vol 23 no 6 pp 601-620 [7] Smagalla, D. (2004) ‘Supply Chain Culture Clash’ MIT Sloan Management Review Vol 46, No1 p6 [8] Whitehead, P. (1986) ‘Profitably Marketing Products through Grocery Retailers’ European Journal of Marketing Vol. 20 No.6, pp52-65 [9] Abolafia, M. (1996) ‘Making Markets: Opportunism and Strategies of Rest raint on Wall Street’ . Harvard University Press. [10] Mehta R, T Larsen, B Rosenbloom and J Ganitsky (200 6) ‘The Impact of Cultural differences in US B”B ex port Marketing strategic alliances’ Industrial Marketing Management Vol 35 Iss2 pp156-165 341 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES [11] Gordon GG (1991) ‘Industry Determinants of Organiza tional culture ‘ The Academy of Management Review V ol 16 No.2 pp396-415 [12] Reichers A E and B Schneider (1990) ‘Climate and cu lture: An Evolution of constructs in Schneider B (e ds) organizational Climate and culture Josssey-Bass San Fransisco [13] Rousseau D M (1990 ) ‘Assessing Organizational Cult ure: the case for multiple methods’ in SchniderB (E ds) Organizational Climate and culture Jossey-Bass Publ ishers San Fransisco CA pp153-192 [14] Schein E H (1991) ‘What is Culture? In Fronst PJ (E ds) Reframing Organizational Culture, Sage, Newbury Park CA pp243-253 [15] Marcoulides GA and RH Heck (1993)’ Organizational C ulture and performance: proposing and testing a mod ule ‘ Organization Science Vol 4 no.2 pp209-225 [16] [16] Das TK and B S Teng(1998) Between Trust and c ontrol: Developing confidence in Partner Co-operati on in alliances ‘ Academy of Management Vol 23 issue 3 [17] Kauser S and V Shaw (2004) ‘The influence of behavi oural and organizational characteristics on the suc cess of international strategic alliances’ International Ma rketing Review Vol 21 No1 [18] Hardy, K., Magrath, A. (1989) ‘ Dealing with Cheating in Distribution’ European Journal of Marketing 2 3 (2) pp123-129 [19] Barney, J.(1996) Gaming and Sustaining Competitive Advantage . Reading MA Addison Wesley. [20] Williamson, O. (1993 )‘ Opportunism and its Critics’ Management and Decision Economics 14(2) 97-108. [21] Rubin PA and Carter J R ‘Joint Optimality in Buyer -Supplier Negotiations’ Journal of Purchasing and M aterials Management Spring 1990 pp 20-26 . [22] Anderson, E., Weitz, B . ( 1992) ‘ The Use of Pledges to Build and Sustain Commitment in Distribution Channels’ Journal of Marketing Research 29 Feb pp8-34. [23] Wathne, K., Heide, J. (2000) ‘Opportunism in Inter- firm Relationships: Forms, Outcomes and Solutions’ Journal of Marketing Oct 2000 Vol. 64 issue 4. [24] Parkhe, A., Rosenthal, E C., Chandran R (1993) ‘Pri soners Dilemma Payoff Structure in Interfirm Strate gic Alliances: An Empirical Test ’ Omega International Journal of Management Science , 2 1 (5) 531-539 [25] Heide, J., Miner, A. S.( 1992) ‘The Shadow of the F uture: Effects of Anticipated Interaction and Frequ ency of Contact on Buyer- Seller Co-operation’ Academy of Management Journal 35 pp265-291. [26] Hallen, L., Johanson, J.(1991) ‘Interfirm Adaptatio n in Business Relations’ Journal of Marketing April, Vol.55 No.2 pp29. [27] Lusch R. F and Brown J R ‘A Modified Model of Power in the Marketing Channel’ Journal of Marketing Res earch Vol. XIX (August 1982) pp312-323 [28] John G ‘An Empirical Investigation of some Antecede nts of Opportunism in a Marketing Channel’, Journal of Marketing Research 21 August 1984 pp278-289. [29] [Williamson, O. (1985) ‘ The Economic Institutions of Capitalism’ NY: Free Press [30] Lee, Y., Cavusgil, T. (2006) ‘Enhancing Alliance Pe rformance: The effects of contractual-based versus relational- based governance’ Journal of Business Research Vol. 59, Iss. 8 pp896-905 [31] Axelrod R.(1984) ‘The Evolution of Co-operation’ , New York Basic Books [32] Deeds, D., Hill, C. (1998),’An Examination of Oppor tunistic Action within research alliances: evidence form the biotech Industry’ Journal of Business Venturing 14, 2 pp141-63 [33] Chiles, T., Mc Mackin, J..(1996) ‘ Integrating Variable Risk Preference, Trust and Tra nsaction Cost Economics’ Academy of Management Review (January) pp79-87 [34] Goshal S., Moran, P. (1996) ‘Bad for Practice: A Cr itique of the Transactions Cost Theory’ Academy of Management Review Vol. 21 No.1 pp13-47 [35] Granovetter, M. (1985) ‘ Economic Action and Social Structure: The Problem o f Embeddedness’ American Journal of Sociology 91(3) pp481-510. [36] Sable & Zeitlin 1985 In Lorenz E. H (p 196) ‘Neithe r Friends nor Strangers: Informal Networks of Subco ntracting in French Industry’ pp 194-209 1988. [37] Brickley J, Smith, C., Zimmerman, J. (1996) ‘Organizational Architecture: A Managerial Economic s Approach. Philadelphia: Richard D Irwin. [38] Aulakh M Kotabe and M Sahay (1996, ) ‘Trust and pe rformance in cross border marketing partnerships: A behavioural approach’ Journal of International Busi ness Studies 27 (5) pp1005-1-32. [39] Madhok A (1995)’Revisiting Multinationals firms’ to lerance of joint ventures: A trust based approach’ journal of International Business Studies 26 Vol 1 pp117-137. [40] Johnson, J.L., Cullen, J.B., Sakano, T, Takenouchi, H. (1996) ‘Setting the stage for trust and strateg ic integration in Japa nese-US cooperative alliances’, Journal of International Business Studies , Vol. 27 No5 pp981-1104. [41] Badenhorst J A ‘Unethical Behaviour: Causes and So lutions’ Business Ethics 1993. [42] Plank R E, Landeros R and Plank L F ‘Values driving Decisions in Questionable Purchasing Situations’ International Journal of Purchasing and Materials M anagement Vol. 30 No.4 Fall 1994 pp 45-53. [43] Hugo et al In Badenhorst J A ‘Unethical Behaviour: Causes and Solutions’ Business Ethics 1993. [44] Reilly & Kyj In Badenhorst J A ‘Unethical Behavi our: Causes and Solutions’ Business Ethics 1993. [45] Bowman & Carroll : In Ferrell O C, Gresham L G and Fraedrich J ‘A Synthesis of Ethical Decisions Mod els for Marketing’ Journal of Macromarketing Vol. 9 No.2 F all 1989 pp55-64. 342 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 THE IMPACT OF INDUSTRY GRO UP CULTURE ON THE DEVELOP MENT OF OPPO RTUNISM IN CRO SS- SECTOR ALLIANCES [46] Tsalikis J and Fritzche D J ‘Business Ethics: A Lit erature Review with a focus on Marketing Ethics’ Vo l. 8 No. 9 1989 pp695-743. [47] Arlow P & T. A Ulrich (1988) ‘A Longitudinal Survey of Business School Graduates’ Assessments of Busin ess Ethics’ Journal of Business Ethics Vol. 7 pp295-302 . [48] Newstrom J W & Ruch W A ‘The Ethics of Management a nd the Management of Ethics’ , MSU Business Topics Winter 1975 , 23 pp 29-37. [49] Ulrich P and U Thielemann(1993) ‘How do Managers Th ink about Market Economics and Morality? Empirical Enquiry’s into Business Ethical Thinking Patterns’ Journal of Business Ethics 12 pp879-898. [50] Brenner S N and Molander E A ‘Is the Ethics of Busi ness Changing?’ Harvard Business Review Jan-Feb 19 77 pp 57-75. [51] Brennan J J Jr, Valtz J J , Shallenberger J B and S tanton V P ‘How Ethical are Businessmen?’ Harvard Business Review 39 (4) 1961 pp6-19 156-176. [52] Hawke M ‘(1994) Mythology and reality: the perpetua tion of mistrust in the building industry ‘ Constru ction papers of the Chartered Institute of Building Vol 41 pp3- 6. [53] Boone C and van Witteloostuijn A (1999) ‘Competiti ve and Opportunistic behaviour in a prisoners dilem ma game: experimental evidence on the impact of culture and education ‘ Scandinavian Journal of management Vol 15 no..4 pp333-350 [54] Das TK (2005) Deceitful behaviours of alliance part ners: potential and prevention’ Management decision Vol 43 no.5 pp706-719. 343 344 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATION COMPETENCE AND INNOVATION CULTURE? Anu Suominen, Jari Jussila, Pasi Porkka, Hannu Vanharanta Tampere University of Technology Abstract The future competitiveness of organizations is dependent on their ability to innovate. On one hand, for organization’s personnel to develop themselves an d their innovation competence, the needs for development should be discovered. On the other hand, as the development of a positive innovation culture is regarded as a solid base for the innovations, the development requires that the management addit ionally discover the needs for development of the organization. The development need s, for individual innovation competence i.e. creative tension and for organizational in novation culture i.e. proactive vision, both can be portrayed with the difference between the current and the future state. The visualization can be carried out via personnel’s self- evaluation of linguistic statements and results shown both nu merically and non- numerically. Yet, an interesting question remains: do those r esults of creative tension and proactive vision have interrelations with each other? Therefore, this pap er aims to find interrelations between creative tension of individual’ s innovation competence and proactive vision of organizational innovation culture. The paper deals with the concepts of innovation, innovation competences, innovation culture. T hen the discussion goes on to describe the method of self-evaluation vi a software tool, creative tension, proactive vision and Friedman test for non-numeric dat a. And after that the results of an empirical study in organizations are presented. Keywords: Innovation, Innovation Competence, Innova tion Culture, Self-evaluation 1.0 Introduction Arthur VanGundy [23] asks a timely question: “Where has all the innovation gone?” The future competitiveness of organizations is dependent on th eir ability to innovate. On one hand, for organizat ion’s personnel to develop themselves, their individual i nnovation competences, should be discovered for bot h the current state and their future objectives. On the o ther hand, the development of a positive innovation culture is 345 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? regarded as a solid base for the innovations. For d eveloping such innovation culture, the management o f an organization should be aware in the same way of bot h the current state and the future vision of the in novation culture. Yet, evaluating innovation competences or culture from management top-down perspective does n ot give a holistic view. The needed bottom up view can be gained via self-evaluation carried out by membe rs of an organization. Here, self-evaluation is considere d as a method for a single person to evaluate [e.g. 15] subjectively one’s individual competences portrayin g creative tension [20] and objectively the environ ment – this case the organizational environment thus illus trating proactive vision [e.g. 4, 18]. Those indivi dual’s innovation competences [e.g. 8, 10, 21, 25] and als o organizational innovation enablers and barriers [ e.g. 1, 2, 3, 7, 16, 22] have been discovered from scientific literature. In practice, this self-evaluation can b e carried out with two web-based software tools using fuzzy logic [13] – one representing individual’s innovation competence and the other organization’s ability to support one’s innovativeness, i.e. innovation cultu re. For holistic view the evaluations of the various indivi duals can be combined, thereby visualizing the need s for improvement for both individuals and their organiza tion. The results of these two tools evoke question s; therefore, the goal of the proposed paper is to ans wer the research question: “Is there interrelations between the results of self-evaluation of creative tension of innovation competences and proactive vision of organizational innovation culture? Related to this question, the attempt of this paper is to find proof that there either is or is not interrelations between organiza tions individuals’ innovation competence and innova tion culture results. In answering this question, first the paper deals with the concepts of innovation, in novation culture and innovation competences. Then the discus sion goes on to describe the method of self-evaluat ion via software tool. And after that the results of an emp irical study in organizations are presented. 2.0 Innovation Innovation has been defined various ways by many au thors. However, the OECD [17] definition of innovat ion is “...the implementation of a new or significantly improved product (goods or services), or process, a new marketing method, or a new organisational method in business practices, workplace organisation or exte rnal relations.” Therefore, the minimum requirements for innovation are its novelty, whether it has been de veloped by the organization itself, or adopted from others; and its implementation, either to the market or in use within the organization [17]. This broad definition allows studying the phenomenon of innovation from larger point of view, not only from the narrow product angle. The link between individual creativity and organiza tional innovation has been found in the “Componenti al Theory of Organizational Creativity and Innovation” (CTOCI) [1, 2, 3]. Thus, outcome of innovation re quires both individual creativity and the innovation respo nsive organizational environment. According to Amab ile [1, 2, 3], the individual creativity is formed of three components: creativity skills, expertise and task motivation, whereas organizational innovation is a combination of resources, organizational motivation and managem ent practises. Co-dependently, the individual or team c reativity feeds the organizational innovation, wher eas the work environment impacts the individual creativity. 346 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? 2.0.1. Innovation Culture and Competence Innovation has been defined both as a specific comp etence [e.g. 9, 24] and a collection of competences [e.g. 25]. A competence, or competences in the plural, wi ll be used to refer to those specific characteristi cs of an individual that are causally related to superior pe rformance in innovating. Several competences relate d to innovation have been presented [1, 5, 9, 21, 25], o f which 27 have been discovered as essential to inn ovation. (Table 1) All organizations have organizational culture [e.g. 19, 23]. Additionally, in the long run all organi zations face changes and introduce new techniques and technologi es that can be regarded as innovations. Therefore i t can be claimed that all organizations have innovation c ulture as well. Here we see innovation culture in a n organization as the part of the organizational cult ure that either produces innovations by enabling th em – or doesn’t. Many authors have presented such enablers for organizational innovation [1,2, 3, 7, 22] of wh ich we have discovered 22 various enablers for organizatio nal innovation (Table 1). Table 1: Individual innovation competences and Orga nizational innovation enablers Individual innovation competences Organizational in novation enablers Connection to innovation enablers 1. Absorptive capacity 2. Accurate self-assessment 3. Achievement orientation 4. Change orientation 5. Communication 6. Flexibility 7. Independence 8. Initiative 9. Stress tolerance 10. Leveraging diversity 11. Professional and technical expertise 12. Relationship building 13. Risk orientation 14. Seeking information 15. Self-development 16. Teamwork and cooperation 17. Trustworthiness 18. Analytical thinking 19. Conceptual thinking 20. Divergent thinking 21. Imagination 22. Intuitive thinking Connection to innovation competences 1. Absorptive capacity 2. Constructive feedback 3. Challenge 4. Change-able 5. Communication 6. Flexibility 7. Freedom 8. Empowerment 9. Stress management 10. Requisite variety 11. Organization support learning 12. Networking 13. Risk tolerance 14. Organization support development 15. Seeking information 16. Team work and collaboration 17. Trust and openness 18. Idea generation No connection to innovation enablers Conflict management Responsibility Self-control Self-confidence Understanding others No connection to innovation competences Idea documentation Idea screening and evaluation Understanding Strategy Situational constraints 347 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? 3.0 Methods 3.0.1. Self-evaluation In this research the self-evaluation is performed u sing a computer application based on fuzzy logic. I n practise, the subject evaluates linguistic statements describ ing the innovation competences and the innovation c ulture using a web-based graphical user interface [cf. 14] . For each statement the subjects evaluates the cur rent and the future target level. The scale is fuzzy, for ex ample, from continuum of strongly agree to strongly disagree. 3.0.2. Creative Tension and Proactive Vision Zwell [24] proposes three competence cornerstones t hat support the success of an organization: the competence of its leadership, the competence of its employees and the degree to which the corporate cu lture fosters and maximizes competence. If these cornerst ones are strengthened, Zwell [24] believes that an organization can improve virtually every side of it s operations and come close to achieving its establ ished vision; yet all these cornerstones do interrelate, i.e. co-evolve. Senge [20] points out that envisioning the future i n the desired destination requires realizing the pr esent state. The present state can be clarified by involving all the members of the organization in a participatory manner. Focusing attention on oneself, an individual can br ing out the current state of one’s innovation compe tences. Similarly by focusing the attention to the outside world and its processes enables people to form thei r perception of how things should be related to the c urrent situation and additionally the vision of the desired future. More specifically, at the individual level it is possible to evaluate each person’s own percep tion of the current and desired level of organization, i.e. whe ther the organization’s characteristics enable or h inder personal innovativeness and creativity. Self-evaluation generates a difference between the current level (reality perceived) and the vision of the desired future (what is wanted). This difference in each individual’s perception of one’s current and desired future state is the individual’s creative tension, as termed by Senge [20] [15]. Similar energy, or th e will to develop one’s organization i.e. the gap between the current and target state is referred to as the pro active vision [e.g. 4, 18]. Proactive vision activates peo ple to mould the environment to their liking. Colle ctively, the individual creative tensions and proactive visions form a group level results that reveals the most cr itical areas in need of development in terms of organization’s i nnovation competences and organizational culture. W ith the above described participatory manner, it is pos sible to direct the organizational inputs toward cr eating an innovation enabling culture in the organization. 3.0.3. Friedman Test of Non-numeric Data The data collected with linguistic variables is by nature weakest in the statistical sense. Let’s assu me, that there are three different spaces in different cities. The se spaces are inside a house, outside a house and i n the freezer. 348 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? Next we ask from a person how cold it is in each sp ace. The answers are given in linguistic means, wit h no thermometer involved. The answers may vary from “re ally hot” into “extremely cold”. It is impossible t o define from two different person’s answers which on e is warmer in June, Brunel or Pori. However, it is possible to define within one single person’s answe rs the rank of different spaces. In February in Por i warmest is inside and coldest probably outside. Statistical ly these measurements are of nominal scale, where v alues are used merely as means of separating the properties o f elements into different classes [6]. With nominal scale, traditional statistical methods (sums, correlations, etc.) are not applicable. Our self- evaluation creates numerical values for each compet ence within one answerer. These values can be ranke d within one answerer and there are valid statistical methods for calculating group results based on the rankings. We have used the Friedman test [6] to rank creative tensions and proactive visions. In Friedman test w e have hypotheses that there are significant differences b etween rankings. Friedman test sums the rankings an d gives us a value within which the sums should be consider ed equal. We have used significance level of 0.05 a nd divided the sums into three groups: the most signif icant, middle group and the least significant. With creative tensions we came into groups where the first 7 rank ings were the most significant and the last 6 the l east significant. With proactive vision the rankings wer e divided into two groups, where 11 first ones were the most significant. In empirical results we study how the corresponding innovation competencies and innovatio n enablers were divided into these groups and what th at division means. 4.0 Empirical Results The empirical case study was conducted with test gr oups in two organizations in different branches: information technology and university. The test gro up in the information technology organization was 1 0 persons and in the university was 10 persons. Both groups carried out a self-evaluation of linguistic statements with two web-based software tools using fuzzy logic [10] – one representing individual’s innovation competence and the other organization’s ability to support one’s innovativeness, i.e. innovation cultu re. The test persons self-evaluated 108 statements regardin g their 27 individual innovation competence and 94 statements regarding 22 enablers and barriers in th eir organization’s innovation culture that have bee n formulated according to the conceptual research on the literature. For collective view, the rankings o f the various individual’s evaluations were accumulated , thereby visualizing the needs for improvement for both all the individuals in the organization and their organ ization. The test results of these two case organi zations were analyzed by seeking interrelations. In the first round of analysis, the average values of the collective results were compared. The result s, viewed by the ranking of each competence or organizational feature, showed no specific interrelations between any competence or feature, but gave a slight indication , that with other statistical methods there might b e found some interrelations. 349 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? 4.0.1. Results of Friedman Test of Non-numeric Data Thereby, in the second round of analysis, the Fried man test, a test method for non-numeric data was ap plied for finding statistically relevant interrelations. For this test, the entire data of the two test grou ps were handled together as one group (Table 2). From this data we have come up three kinds of conclusions. Table 2: Results of the interrrelations between Ind ividual innovation competences and Organizational innovation enablers with Friedman test Intuitive thinking7 Professional and techical expertise6 Conceptual thinking5 Analytical thinking4 Understanding others3 Self-control2 Absorptive capacity1 NAMERANK Absorptive capacity9 Idea generation8 Organization support learning7 Constructive feedback6 Organizational flexibility5 Idea documentation4 Teamwork and collaboration3 Networking2 Stress management1 NAMERANK Risk orientation27 Independence26 Divergent thinking25 Responsibility24 Leveraging diversity23 Achievement orientation22 Requisite variety20 Freedom19 Changeability18 Risk tolerance17 Challenge16 1 2 3 INNOVATION COMPETENCE INNOVATION CULTURE 7 6 5 4 3 2 1 9 8 7 6 5 4 3 2 1 27 26 25 24 23 22 20 19 18 17 16 1 2 3 1. High score of creative tension of individual’s inno vation competence and additionally high score on organizational innovation culture enablers. In othe r words, people find a great need for development a t their innovation personal competences, and as great need for improvement also for organizational cultu re enablers. This result can be interpreted so that an individua l finds great need for improvement in one’s individ ual capabilities and currently organization does not su pport one’s innovation competences at such a level that it enables his or her individual needs for improvem ent to be realized at adequate level. 2. Individual’s competences that the people did not fi nd need for improvement, i.e. the creative tension was rather low. With those competences, for their count erparts in organizational level, people did not reg ard needing development either. 350 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? This result can be interpreted so that in those ind ividual competences a person does not find great ne ed for improvement, the same people do not require any org anizational support either. 3. As a third group could be identify such individual innovation competences that does not have a clear counter part in organizational innovation culture e nablers. Those interrelations to organizational innovation culture enablers have to be studied furt her. 5.0 Conclusion The interrelations between individual innovation co mpetence and organizational innovation culture stil l requires additional research. In this paper we have sought to offer a brief illustration on the proble m area, though the concepts of creative tension and proacti ve vision. On the basis of the discussion above, we can conclu de that the creative tension of individual’s innova tion competence and proactive vision of innovation cultu re in organizations do not have clear interrelation between when compared with simplistic ranking of average va lues of each competences and organizational feature s. Yet, according to our finding with Friedman test, a statistical method for non-numeric data, there are at least two types of interrelations between individual comp etences and organizational innovation culture featu res, when compared with creative tension and proactive v ision: 1. High scoring in both creative tension and proactive vision portraying both individual and org anizational needs for improvement. 2. Low scoring i n both creative tension and proactive vision portraying in novation competences that do not require developmen t, thus not requiring support from counterparts in organiza tional innovation culture. Additionally there was found innovation competences that do not have direct coun terparts in organizational innovation culture featu res, yet having need for development. To conclude our findings, competences that people w ish to enhance in the future have, to some extent, interrelations to those organizational features tha t people wish to be enhanced in the future. However , those interrelations have to be studied further with more data and also with other statistical methods. References [1] Amabile, T.M. 1997. Motivating Creativity in organi zations: on doing what you love and loving what you do. California Management Review. Vol. 40, No. 1. pp. 3 9-58. [2] Amabile, T.M. 1998. How to kill creativity. Harvard Business Review. September-October 1998. pp. 77-87 . [3] Amabile, T.M, Conti, R., Coon, H., Lazenby, J, Herr on, M. 1996. Assessing the work environment for cre ativity. Academy of Management Journal. Vol. 39, No 5. pp. 1 154-1184. [4] Aramo-Immonen, H., Kantola, J., Vanharanta, H. and Karwowski, W. 2005. The web based trident applicati on for analyzing qualitative factors in mega project manag ement. Proceedings of the 16th Annual International Conference of IRMA2005, information resources management assoc iation international conference. San Diego, Califor nia, May 15-18, 2005. [5] Boyatzis, R. 1982. The competent manager. New York, John Wiley & Sons. [6] Conover, W.J. 1999. Practical nonparametric statis tics. 3 rd ed. John Wiley & Sons, Inc. New York et al. [7] Ekvall, G. 1996. Organizational Climate for Creativ ity and Innovation. European Journal of Work and Organizational Psychology. 5(1). pp. 105-123. 351 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INTERRELATIONS OF DEVELOPMENT NEEDS BETWEEN INNOVATI ON COMPETENCE AND INNOVATION CULTURE? [8] Farr, J.L., Sin, H-P., Tesluk, P.E. 2003. Knowledge Management Processes and Work Group Innovation. Sh avinina, L.V. (Ed.). The International Handbook of Innovatio n. Oxford, Elsevier. pp. 574-586. [9] Goleman, D. 1998. Working with emotional intelligen ce. London, Bloomsbury. [10] Goleman, D., Boyatzis, R., McKee, A. 2003. The New Leaders. Transforming the art of leadership into th e science of results. London, Time Warner. 349p. [11] Hamel, G. 2007. The future of management. Boston, H arvard Business School Press. [12] Jackson, C.M. 2004. Systems Thinking: Creative Holi sm for Managers. West Sussex, John Wiley & Sons Ltd . [13] Kantola, J. 2005. Ingenious Management. [14] Kantola, J. 1998. A Fuzzy Logic Based Tool for the Evaluation of Computer Integrated Manufacturing, Organization and People System Desig n. Doctoral Thesis. University of Louisville. [15] Kantola, J. Vanharanta. H., Karwowski, W. 2005. The evolute system: a co-evolutionary human resource development methodology. International encyclopedia of human factors and ergonomics. [16] Martins, E.C. & Terblanche, F. 2003. Building organ isational culture that stimulates creativity and innovation. European Journal of Innovation Manageme nt. Vol. 6, No. 1. pp. 64-74. [17] OECD. 2005. Oslo Manual: Guidelines for Collecting and Interpreting Innovation Data, 3rd Edition. 166 p. [18] Paajanen, P., Kantola, J., Vanharanta, H. 2004. Eva luating the organization’s environment for learning and knowledge creation. 9 th international Haamaha conference: human & organiza tional issues in the digital enterprise. Galway, Ireland, August 25–27, 2004. [19] Schein, E.H. 2004. Organizational culture and leade rship. 3 rd ed. San Francisco, CA, Jossey-Bass. 438 p. [20] Senge, P.M. 1990. The fifth discipline: the art and practice of the learning organization. New York, C urrency Doubleday. [21] Spencer, L.M., Spencer, S.M. 1993. Competence at wo rk: models for superior perform-ance. New York, Joh n Wiley. [22] Trott, P. 2005. Innovation management and new produ ct development. 3 rd ed. Essex, Pearson Education Limited. [23] VanGundy, A.B. 2007. Getting to innovation: how ask ing the right questions generates the great ideas c ompany needs. New York, Amacom. [24] Zwell, M. 2000. Creating a culture of competence. N ew York, John Wiley & Sons, inc. [25] Van Assen, M.F. 2000. Agile-based competence manage ment: the relation between agile manufacturing and time- based competence management. International Journal of Agile Management Systems. Vol. 2, No. 2. pp. 142 -155 352 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT Wei Fu 1 , Yoon Seok Chang * 1 , Myo Min Aung 1 , Charalampos Makatsoris 2 , Chang Heun Oh 1 1. School of Air Transport, Transportation & Logistics , Korea Aerospace University, Republic of Korea 2. School of Engineering and Design, Brunel University , UK Abstract This paper presents a cold chain monitoring system which is implemented by using ubiquitous computing technologies, Radio Frequency Identificat ion (RFID) & Wireless Sensor Network (WSN). In this paper, we discuss h ow cold supply chain works and how we can monitor and control cold supply chain by us ing wireless tracking and sensing technologies. We propose a prototype design wh ich will provide a well controlled and transparent cold chain system, which could he lp the users to manage their products’ environmental data in real time during the life cycle. Moreover, we highlight how the availability of product trace data in combination with historical condition-monitoring data can facilitate decision-maki ng processes enhancing supply chain’s performance. Finally we discuss the integration works of these two technologies together in the cold supply chain manage ment system. Hardware and software platform of WSN used in this system are also descri bed in this paper. Keywords: Cold Chain, RFID, WSN. 1.0 Introduction A cold chain is a temperature-controlled supply cha in. An unbroken cold chain is an uninterrupted seri es of storage and distribution activities which maintain a given temperature range. Cold chains are common i n the food and pharmaceutical industries and also some ch emical shipments. One common temperature range for a cold chain in pharmaceutical industries is 2 to 8 ° C. But the specific temperature (and time at temper ature) tolerances depend on the actual product being shipp ed [1]. Cold chain is very important for the supply chain management of the food and pharmaceutical industrie s because a good cold chain can reduce the risk and cost. The loss of a trailer of temperature sensitive prod ucts due to improper transportation or inventory wi ll cost 353 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT thousands of dollars; a pharmaceutical shipment, in millions [2]. Perishable food products must be continuously monitored for safety concerns througho ut the whole supply chain. A breakdown in temperatu re control at any stage will impact on the final quali ty of the product [3]. The key to managing the cold chain is to measure product temperature at each stage. Currentl y, food, dairy and pharmaceutical companies are alr eady tracking their environmentally sensitive products u sing temperature data loggers placed in their trans portation vehicle, containers or even pallets. However, this technology only provides information for a certain part of the cold chain [4]. But for the whole life cycle tr acking is still under research. To achieve the capa bility of monitoring and controlling of every link in a cold chain, real time data should be communicated via da ta retrieval devices. Radio Frequency Identification (RFID) technology ha s been available for decades but it has only recent ly risen to prominence. Increasing competitiveness between t ag manufacturers has lead to significant reductions in the cost of the technology, which in turn drives indust ry sectors’ attention on RFID for adoption. Current RFID systems, however, lack intelligence on their operat ional environment such as temperature, humidity, et c. In order to tackle this problem, we have been investig ating the possibility of integrating RFID with a re latively new technology, WSN. WSN are made up of many small devices with processing and sensing abilities [5]. Integrating RFID systems with condition-monitoring systems will enhance existing track and trace applications, not only the location, but also the c ondition of perishable and valuable products. Moreo ver, the availability of product trace history data in combi nation with historical condition-monitoring data ca n facilitate numerous decision-making processes [6]. In this pap er, we aim to discuss the benefits that such integr ation could provide, whilst also paying close attention t o the implementation view. In our research, we used a sensor network system based on Nano-Qplus platform which i s composed of NANO HAL for abstracting the hardware part sensing and actuating, task managemen t, power management, and message handling module. I n addition, the Nano-Qplus platform includes ATmega12 8 MCU and CC2420 Zigbee, IEEE802.15.4 RF communication module. Based on Nano-Qplus platform, we developed the integrated system which have the functionalities of both RFID and wireless sensor ne twork system [7]. Overall, this paper focuses on th e design and implementation of the WSN based cold chain mana gement system which monitors the condition of temperature sensitive objects during their supply c hain processes. 2.0 Wireless Sensor Network Platform 2.1 Hardware of Nano-Qplus Platform The sensor hardware which is called Smart Sensor No de focuses on low-cost, low-power, and high- modularity. The sensor node is composed of four blo cks: Main, Base, Sensor, and Actuator. The main blo ck has ATmega128 microcontroller and CC2420 IEEE802.15 .4 compliant RF transceiver. The base block is used for Anchor node with RS-232 serial I/F, parallel I/ O and external power source. For sensing of physica l environment, the sensor block has several sensor en tities, such as light, humidity, temperature, and u ltra sound. The actuator block is made up several electrical sw itches and can be combined with electric appliances in order to turn Off/On power. In case of normal appli cation, the sensor node is power-supplied by two AA 3.3 batteries [7]. 354 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT 2.2 Software of Nano-Qplus Platform The architecture of the Nano-Qplus resembles as cla ssical modular and layered design and consists of dynamically-loaded modules included in hardware, Na no-OS, and application parts respectively. The hardware part is composed of MCU using ATmega128, R F module, which is CC2420 for wireless communication, and Sensors/Actuators. The Nano-OS p art has a role as kernel scheduler and network prot ocol stack for handling RF messages, and it have a devic e driver modules, called as nHAL, for abstracting t he hardware part. Furthermore, the Nano-OS part also o ffers the system APIs for convenient developments o f WSN applications to sensor networking programmers. In the end, the application part consists of one or more modules interacting via system APIs with Nano-OS pa rt [7]. 2.3 RFID Reader Embedded Sensor Technology As the name indicates RFID is used to identify obje cts using radio communication. Different applicatio ns have different demands on range, power consumption and s o on. The most commonly used carrier frequencies ar e 125 kHz, 13.56 MHz, 868 MHz and 2.4 GHz. The tags can be either passive or active, this system uses 1 3.56 MHz carrier with inductive coupling, has ability to read from and write to the tag and read multiple t ags simultaneously [8]. The normal reading range for th e RFID system used in this system is about 5-10 cm, which means that reader and tag cannot be physically sepa rated more than this distance in order to function. The nodes in the network include an RFID reader and RF transceiver as we have mentioned before. When the RFID reader read the tag, it will transmit the tag’ s information from the sensor node to the sink node , and the distance between the sensor node and sink node coul d be 10-30 meters [9]. 3.0 Analysis of the Cold Chain 3.1 Typical Cold Chain Process From the factory to the consumer, products follow c omplex logistic circuits that are subjected to intr insic constraints. First, the required chilling time betw een harvest, or the end of cooking, and loading, is a constraint encountered by the producer and the carrier. The ca rrier’s liability comes into play from the moment t he products are taken over, and it is up to him to che ck the temperature manually during loading products , then reach the temperature-controlled distribution hub a nd send to the customer via supermarket or store. Temperature control needs to be improved throughout the cold chain, to ensure food safety and hygiene and to maintain the product quality [10]. 355 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT Fig.1. The sequence of events within a typical cold chain [11] 3.2 Hazard Analysis and Critical Control Point (HACCP) A cold chain can be managed by a quality management system: it can be analyzed, measured, controlled, documented, and validated. The food industry uses t he process of Hazard Analysis and Critical Control Point (HACCP) as a useful tool. Its usage continues into other fields [1]. HACCP is an important element in the control of safety and quality in food production. W hen properly applied, it provides a management tool aimed at complete commitment to product quality and safet y. HACCP is useful in identifying problems in food production and works well for simple products and p rocesses. There are 7 principles of HACCP as follow s: 1)Identify hazards, assess risk, and list controls; 2) Determine critical control points (CCPs); 3) Sp ecify criteria to ensure control; 4) Establish monitoring system for control points; 5) Take corrective acti on whenever monitoring indicates criteria are not met; 6) Verify that the system is working as planned; 7 )Keep suitable records [12]. Therefore according to these principles, it is clear that we need to define the control points to monitor and keep the time-temperature his tory throughout the chain for products’ safety and quality. 3.3 Lot Traceability and Expiration Many process industries, notably those involving fo od and beverages, drugs, cosmetics, and medical apparatus, are subject to government regulation, an d must maintain records that detail the lot identif ication of materials used in the manufacture of these products . Many businesses follow this practice to protect themselves against liability. Shelf life or lot exp iration tracking systems also require supporting in ventory record subsystems. Typically, they track lot creati on dates and expiration dates and provide for first -in, first- out (FIFO) use of material as well as periodic agin g reports used to predict material that is potentia lly expiring [13]. 3.4 Advanced Planning Systems Adopting Cold Chain Management Advanced planning system incorporates long-term, mi d-term and short-term planning levels. With the sup port of effective information flows among these levels m ake it a coherent software suite. APS do not substi tute but supplement existing Enterprise Resource Planning (E RP) systems. APS now take over the planning tasks, while an ERP system is still required as a transact ion and execution system. APS are intended to remed y for the inefficiency of ERP system through a closer int egration of modules, adequate modeling of bottlenec k capacities, a hierarchical planning concept and the use of the latest algorithmic developments. APS is seen in 356 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT its ability to check whether a (new) customer order s with a given due date can be accepted. By adoptin g cold chain management functions into the APS could help us to track not only where the product is but also the temperature information around it. To adapt the sen sing based APS, in our cold chain management system , we can define the environmental parameters to record a nd transmit those in every process of the cold chai n system and send back to APS and ERP system which could hel p us to better manage the forecast, inventories, resources, orders and distribution information [14] . 4.0 Proposed System Description 4.1 RFID&WS N in Cold Chain The monitoring of the cold chain should be from the birth of the product to its’ destination with its whole life cycle. When the product is ready to be palletized, an RFID and temperature (or humidity, etc) sensor n ode is activated and placed on the product stack, each pro duct item will be packaged with the RFID tag contai ning information such as description, destination, and d ate of departure etc. This information is stored in a secure database [4]. So during the whole life cycle of the cold chain from the plant to the hand of customer will be traceable by using the RFID and Sensor Networks. As RFID record the individual information of an item and sensor network could detect the environmental statu s, by combining the two domains we can check the st atus of every product in real time. The system will auto matically integrate new nodes which are placed in t he range of the network. RFID sensor nodes in the network wi ll have the ability to read RFID tags of a certain type and pass that information to the sink node. The sink no de will connect to a PC or Internet from which the user can collect and analyse the data. In every critical poi nts of cold chain we deploy wireless sensor network s in which sensor node should communicate with each other (Fig .2). Fig.2. Integrated RFID&USN Cold Chain Management S ystem Design One or more RFID reader sensor is being clustered w ith some of the temperature sensor node. When the product have been read from the RFID reader node, t he temperature data being detected from the same cl uster by the temperature sensor node will be sent to the database and build up the product environment histo ry information. By combining the RFID data and tempera ture sensor data in the data logger, we get the rea l-time 357 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT temperature information of products. Also by using the web server we can make the web application in w hich remote users could monitor and access the cold chai n data. Whenever the temperature is out of the cont rol range in the part of the cold chain, the system cou ld sent the SMS message to the administrators’ mobi le phone to give the alert by using the CDMA server and SMS service. 4.2 Supply Chain Planning and Execution Supply chain planning refers to a set of supply cha in activities that focus on: evaluating demand for material and capacity; formulating plans; and scheduling to meeting the demand and company goals [15]. Supply c hain execution means set of supply chain activities that focus on fulfillment rather than planning-raw mate rial delivery, manufacturing operations and shipments to customers. Execution functions receive requirement s from the planning cycle and provide the actual data for actual measurements [16]. Obviously, supply chain planning is an important f actor for the success of a company. Poor planning w ill result in loss of profits and revenue while accurat e planning allows a company to operate smoothly and to minimize expenses. The question then is how to more effectively create business forecasts for supply c hain activities and control supply chain execution [17]. In our research, the prototype system will not onl y act as a monitoring tool but also could be used as a plannin g and execution tools of the supply chain control p oints. As an easy example, it is easily provide the gap betwe en planning and execution during the cold chain pro cesses. 5.0 System Implementation 5.1 Programming Tool Description Our prototype system is implemented by LabVIEW (Lab oratory Virtual Instrumentation Engineering Workbench) which is a platform and development envi ronment for a visual programming language by National Instruments. LabVIEW is commonly used for data acquisition, instrument control, and industria l automation on a variety of platforms including Micr osoft Windows, various flavors of UNIX, Linux, and Mac OS [18]. VISA is a standard I/O language for instru mentation programming. VISA by itself does not prov ide instrumentation programming capability. VISA is a h igh-level API that calls into lower level drivers [ 19]. In our system, the sink node is connected to the compu ter by serial connecter, in order to get the data e asily by using NI-VISA module. For our prototype system, we use Microsoft Access database to store the records. 5.2 Cold Chain Management System As a part of cold chain execution tool, RFID tag sh ould be assigned to a product and it will represent the product during the whole life cycle. For example, w e assigned an RFID tag for product ‘ice-cream’ and its attributes (e.g. Product ID, Product Quantity, Cust omer, Preferred High and Low Temperature, Due Date etc.). 358 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT Fig. 3. Data foundation for cold chain management Fig 3. shows example data foundation for cold chain management. After assigning the tag with product information, the next step is to identify the locat ion of a product in the cold chain. When we scan th e RFID tag using the wireless RFID reader sensor node, the sys tem can show us the location information where the tag has been read. In monitoring point of view, we include four steps in the cold chain such as Manufacture, Inventory, Transportation and Retailing which simul ate the whole life cycle of a product. In each step , there is a monitoring window that clearly shows the function of each step. We placed LED indicators to indicate whether the sensor node is working well or not, and to alarm whether the temperature is out of the ran ge from preset interval. By using waveform chart we could c learly see the changes of temperature during monito ring process. We also designed a user interface which co uld help us to monitor the process visually. If the temperature is out of the range, it will also give the alarm for enabling proactive measures for our c old chain (Fig 4). Fig.4. Cold Chain Monitoring System User GUI 359 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 WSN BASED INTELLIGENT COLD CHAIN MANAGEMENT 6.0 Conclusion Highly integrated and inexpensive smart nodes netwo rk is much cheaper and more flexible compared with heterogeneous network. Smart nodes read fewer tags and can be deployed densely as self-organizing WSN . They run autonomously and translate data informatio n to the sink node. The gathered information is transmitted through multi-hops. Energy constraint i s an extremely crucial problem when smart nodes are applied in industry because of battery consumption. Currently ZigBee protocol is considered as the bes t candidate as it satisfies reliable, low cost, low p ower consumption requirements [20]. For a large sca le application, collecting a huge amount of data from the sensor nodes is also an issue to consider. Alth ough we use one temperature sensor in each location, it is also possible to deploy more than one temperature s ensor for multi-temperature modeling. We have presented a des ign model for the cold chain traceability and the c oncept of the cold chain. After that, the hardware and sof tware architecture for WSN and RFID integration pla tform is described. Then, we have mentioned a prototype of a traceability system using Nano-Qplus wireless sens or networks and user friendly LabVIEW working environm ent. RFID/WSN integration works are also introduced as a challenging area of research. References [1] Wikipedia, http://en.wikipedia.org/wiki/Cold_chain . [2] T. Kevan, “Control of the Cold Chain”, “Frontline S olutions, Managing Supply Chain Strategies with Tec hnology, 2004. [3] South Australian Research and Development Institute (SARDI), “Maintaining the Cold Chain – Air Freight of Perishables”. [4] Syntax Commerce, “Cold Chain Traceability”. White P aper [5] A. Mason et al., “RFID and Wireless Sensor Network Integration for Intelligent Asset Tracking Systems” . [6] J. Mitsugi et al., “Architecture Development for Se nsor Integration in the EPCglobal Network”, AutoID Labs White Paper WP-SWNET-018 , July 2007 [7] H. S. Choi, I. G. Park, Y. S. Shin, S. M. Park, “A Design and Implementation of Wireless Sensor Networ k Routing on Nano-Qplus Platform”, Advanced Communication Tec hnology, ICACT 2006. The 8th International Confere nce. [8] C. Englund, H. Wallin, “RFID in Wireless Sensor Net work”. Chalmers University of Technology, April 200 4. [9] H.I System Co., Ltd, “Nano-Qplus for Ubiquitous Sen sor Network User Manual”. [10] B. Commere, “Controlling the Cold Chain to ensure F ood Hygiene and Quality”, Bulletin of the IIR – No 2003-2. [11] M. George, “Managing the Cold Chain For Quality and Safety” Flair-Flow Europe Technical Manual, May 2 000. [12] New Zealand Food Safety Authority (NZFSA), “An Intr oduction to HACCP”, May 2003. [13] J. Clement, A. Coldrick, J. Sari “Manufacturing Dat a Structure: Building foundations For Excellence wi th Bills of Materials and Process Information” 1992. [14] G. Knolmayer et al., “Supply Chain Management Based on SAP System: Order management In Manufacturing Companies”., 2002 [15] IBM Business Consulting Services “A Retailer’s Guid e to Supply Chain Management”. Available online via www- 304.ibm.com/jct03004c/tools/cpeportal/fileserve/dow nload5/29826/RCC_WhitePaper.pdf?contentid=29826 [16] Bridgefield Group ERP/ Supply Chain Glossary, Avail able online via http://bridgefieldgroup.com/bridgefieldgroup/glos8. htm [17] Supply Chain Planning System Available online via h ttp://www.epiqtech.com/supply_chain-Planning-System s.htm [18] Wikipedia, “LabVIEW”.Available online via http://en.wikipedia.org/wiki/LabVIEW [19] LabVIEW VISA Tutorial Available online via http:// www.ni.com/support/visa/vintro.pdf [20] L. Zhang, Z. Wang, “Integration of RFID into Wirele ss Sensor Networks: Architectures, Opportunities an d Challenging Problems”, Grid and Cooperative Computi ng Workshops, 2006, Fifth International Conference. . 360 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES LINKING SUPP LY CHAIN MANAGEMENT CAPABILITY AND MANUFACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERFO RMANCE: A CASE STUDY OF THAI INDUSTRIES N. Chiadamrong 1 , N. Suppakitjarak 2 1. School of Manufacturing Systems and Mechanical Engi neering, Sirindhorn International Institute of Technology, Thammasat University, Pathumthani, T hailand, 121 2 1 2. Faculty of Commerce and Accountancy, Chulalongkorn University, Bangkok, Thailand, 103 3 0 Abstract This paper conceptualizes, develops, and validates two dimension s of SCM practices (strategically managed buyer-supplier relationships and intern al manufacturing operations competence). The study is designed to identify im portant factors that influence a firm’s internal operations and involvement in su pplier development, develop reliable and valid measures of these factors, and test hypotheses on their interrelationships. Data for the study were collected from 245 companies in Thailand and the measurement scales were tested and validated using structural equation modeling. The results of this study shed light on the impor tance of managing the cooperative relationship between buyer and seller and its ef fect on the financial situation of the firm in Thai industries. Keywords: Supply chain management, Strategic purcha sing, Internal manufacturing operations, Organizational performance, Structural equation mod eling. 1.0 Introduction As competition becomes more intensified and markets become global, the challenges associated with gett ing a product and service to the right place at the right time at the lowest cost have been more critical to organizational performance. Organizations begin to realize that it is not enough to improve efficienci es within an organization, but their whole supply chain has t o be made competitive. It has been pointed out that understanding and practicing supply chain managemen t (SCM) has become an essential prerequisite to sta ying in the competitive global race and to growing profi tably. Many firms have responded to these condition s by 361 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES focusing on their core competencies, and outsourcin g non-core activities that were previously performe d in- house. As a result, there is a need to achieve the right b alance between internal and external outsourcing ac tivities. The successful implementation of SCM requires integ rating internal functions of a firm and effectively linking them with the external operations of its partner fi rms in the supply chain. Largely missing from this body of knowledge has been a full and complete understandin g of how cooperative supply chains are effectively managed overtime and whether internal manufacturing operation efficiencies within an organization can contribute to such cooperative relationships as wel l as to the organizational competence. Our study is an attempt to fill this gap in the res earch by developing and testing a framework that lo oks into how these relationships function and contribute to the firm’s success. This paper has three objectives : (i) to establish linkages between some aspects of firm’s i nternal manufacturing operations with some aspects of supply chain management: (ii) to study the impact o f these linkages on an organization’s performance: (iii) to study the effects of these linkages in the context of Thai industries. It is hoped that the results of this study will help senior management to better understand these i nherent relationships. For researchers, this case s tudy of Thai industries may provide a stepping stone toward s examining the impact of the relationships between manufacturing and supply functions on organizationa l performance. 2.0 Theoretical Framework and Hypotheses This section focuses on organizational measures in terms of financial performance of the firm in an ef fort to place supply chain management’s capability and inte rnal manufacturing operations’ competence in a theoretical context. The factors under study are co nceptually categorized as follows: Supply chain management capability factor: This includes strategic purchasing (SP) and buyer- supplier relationship (BR). These components influence the f irm or the purchasing function within the firm to a dopt a strategic perspective on long-term cooperative rela tionships. Internal manufacturing operations competence factor: This includes quality expectation (QC) and produc tion and inventory management (PIM). These components ar e an integral part of the firm’s internal manufactu ring operations. Organizational performance factor : This includes the firm’s financial performance (i .e., return on investment, profits as a percentage of sales, net income before taxes, and present value of the firm), which is co nsidered to be the main criterion for judging the organizationa l performance of the firm. 362 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES The following hypotheses and the model depicted in Figure 1 specify the antecedent factors and their interrelationships. 2.1 Supply Chain Management Capability in Relation to Organizational Performance Hypotheses 1 to 3 propose that strategic purchasing (SP) has a positive impact on buyer-supplier relat ionships as well as the firm’s financial performance. These hypotheses are based on the idea that firms that pr actice long-term planning and considered strategic purchas ing are likely to build long-term cooperative relat ionships with their key suppliers which eventually benefit t he firm’s financial performance. Two dimensions of the effective supply chain management were measured: st rategic purchasing and buyer-supplier relationships . Buyer-supplier relationship architecture is concept ualized as being information sharing and trust betw een buyer and supplier, and aspects relates to centrali zed planning which is referred in terms of JIT as “ supply chain proximity”. The following hypotheses are purp osed: H1: Strategic purchasing (SP) has a positive impact on buyer-supplier relationships (BR) H2: Strategic purchasing (SP) has a positive impact on the buying firm’s financial performance (FP) H3: Buyer-supplier relationships (BR) have a positi ve impact on the buying firm’s financial performanc e (FP) 2.2 Relationships of Internal Manufacturing Operations Competenc e and Organizational Performance SP BR FP QC PIM H1 H2 H3 H6 H5 H7 H4 Figure 1: Research Hypotheses 363 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES Hypotheses 4 to 6 postulate that positive relations hips between the independent factors of the interna l manufacturing operations can impact on the organiza tional performance. According to Fogarty et al.[1], in spite of the media attention devoted to the develop ment of supply chain, two mistakes are far too comm on. First, developing effective internal manufacturing operations may seem mundane next to strategy formul ation, but these methods are critical to long-term surviva l and competitive advantage. Second, many analysts assume that implementing sophisticated manufacturing opera tions will solve all problems. These analysts may a chieve a high level of efficiency by optimizing given the lead time and demand variability the firm observes. However, they do not understand that this efficienc y is insignificant compared to the benefits availab le by changing the givens. Therefore, the following hypot heses are purposed: H4: Quality expectation (QC) has a positive impact on production and inventory management (PIM) H5: Quality expectation (QC) has a positive impact on the firm’s financial performance (FP) H6: Production & inventory management (PIM) has a p ositive impact on the firm’s financial performance (FP) 2.3 Relationship of Production and Inventory Management Capability and Buyer- Supplier Relationships Hypothesis 7 suggests a positive relationship betwe en internal manufacturing operations competence (production and inventory management) and buyer-sup plier relationships. To attain the quality in the s upply chain, it is essential to develop a stable buyer-su pplier relationship, which requires that the firms involved work beyond organizational boundaries to improve pe rformance throughout the supply chain. The stabilit y of relationships goes beyond a simple, positive evalua tion of the other party based on considerations of the current benefits and costs associated with the rela tionship. It implies the adoption of a long-term or ientation towards the relationship-a willingness to make shor t-term sacrifices to realize the long-term benefits of the relationship [2]. Therefore, it is hypothesized tha t production and inventory management has a positiv e impact on buyer-supplier relationships H7: Production and inventory management (PIM) has a positive impact on buyer-supplier relationships (B R) 3.0 The Survey A survey was undertaken to gather data for testing the research hypotheses. The survey included multip le scale items for each of the factors. The sample was drawn across industries. A relatively large pool of resp ondents was necessary due to the number of variables in the model. From the 1,000 subjects in the target sampl e, 245 responses were received. Twelve surveys were exclud ed from the analysis because of incomplete informat ion. Thus, the response rate was about 24%. To achieve a s high a response rate as possible, telephone calls and company visits were used to remind potential respon dents of the questionnaires. 364 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES The respondents were drawn from a variety of indust ries and industrial areas drawn from the membership list of the Federation of Thai Industries (FIT). In orde r of frequency, industries most frequently represen ted were automobile, plastics and packaging, electronics, an d consumer products. Registered capital investment was used as indicator of the firm’s size. The response sample was comprised of high ranking plant mangers and purchasing executives. The responses were recorded using a five-point Likert-type scale (1 = strongly disagree to 5 = strongly agree) on 5 groups of the questions representing each node of the hypothesis. This paper uses Structural Equation Modeling method ology with Statistical Analysis Software (SAS) to analyze the research hypotheses. SAS provides a wid e range of statistical analysis ranging from tradit ional analysis of variance to exact methods and dynamic d ata visualization techniques. Then, Structural Equa tion Modeling is performed by using a two-step procedure . The measurement model is developed; this is follo wed by development of the structural model [3]. The mea surement model examines the relationship between th e underlying variables and the factors they are suppo sed to measure. The structural model differs from t he measurement model because it includes causal paths based on hypothesized relationships between specifi c factors in the model. 4.0 Results Confirmatory Factor Analysis (CFA) was used to vali date measures of constructs for developing the measurement model. CFA is a more effective method f or assessing unidimensionality than exploratory fac tor analysis, coefficient alpha, and item-to-total corr elation. The purpose is to ensure unidimensionality of the multiple-item constructs or low item-to-constructs and to eliminate unreliable items [4]. Items that l oaded on multiple constructs were deleted from the model pri or to testing. The measurement model was analyzed u sing the SAS program and CALIS procedure. An adequate fi t was achieved for the measurement model. The chi- square to df freedom ratio = 1.756, the Bentler’s Comparative F it Index (CFI) = 0.9027, Bentler and Bonett’s Non-normed Fit Index (NNFI) = 0.9, all of the t-statistics for the indicator variables were greate r than 2.576, significant at p<0.01, and no standard errors were near zero. The confirmatory factor analysis resulte d in the elimination of a few individual items (V5 and V18) because of low factor loadings or high residuals. T his result was not surprising because many of the surve y items had been developed specifically for the stu dy, and other items had been adapted from other literature streams. Table 1: Measurement model: unstandardized coeffici ent, standard error, t-value and standardized coefficient for each item Individual items, their respective factors and coef ficient alpha for each factor. (All scales were 5-point Likert sc ales where 1 = Strongly disagree – 5 = Strongly agree) Unstandardi zed coefficient Standar d error t-value Standardized coefficient Production and Inventory Management (PIM) (α = 0.6658) V1: The firm has never experienced supply shortage in the past 6 months V2: The firm has never experienced late order deliv ery to customers in the past 6 months. V3: Demand forecasting of the firm is quite accurat e and has a high potential to be used for its production planning. V4: The firm has managed its inventory control syst em effectively so that its level of inventory is suita bly set and managed. Buyer-supplier Relationships (BR) (α = 0.7861) 0.4388 0.6056 0.6603 0.7143 0.0787 0.0812 0.0639 0.0671 5.5741 7.4605 10.3358 10.6486 0.3926 0.5118 0.6823 0.7010 365 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES V6: The firm has firm policy to build a long-term r elationship with key suppliers. V7: The firm has frequent face-to-face planning/communicate with key suppliers. V8: The firm has frequent face-to-face planning/communication with its customers. V9: There are direct computer to computer links wit h key suppliers. V10: The firm has exchanged important information w ith key suppliers such as supply/production/delivery pl ans, current inventory level, etc. Strategic Purchasing (SP) (α = 0.8507) V11: The firm has an effective supplier certificati on program V12: The firm has a formal program for evaluating and recognizing suppliers. V13: Purchasing plan is effectively established by considering various types of relationships established with sup pliers V14: Purchasing plan is reviewed and adjusted to ma tch changes in the firm’s strategic plans on a regular basis. V15: The firm is currently satisfied with key suppl iers. Quality Expectation (QC) ( α = 0.7100) V16: The firm gives the highest regard to the quali ty of its products. V17: There are effective quality control tools to m onitor and control the quality of its products. V19: There is no complains from our customers about the quality of our products. V20: The firm has an effective equipment and machin e maintenance program so that all machines and equipm ent are always in good conditions. Financial Performance (FP) (α = 0.8960) V21: Return on investment V22: Profits as a percentage of sales V23: The firm’s net income before taxes V24: The present value of the firm 0.4909 0.8036 0.6955 0.8075 0.7685 0.6812 0.7350 0.5837 0.6740 0.5313 0.3108 0.7075 0.4761 0.7025 0.7986 0.9094 0.8863 0.5776 0.0594 0.0618 0.0635 0.0852 0.0710 0.0483 0.0502 0.0502 0.0504 0.0541 0.0394 0.0566 0.0737 0.0566 0.0474 0.0487 0.0517 0.0567 8.2658 13.0011 10.9547 9.4806 10.8245 14.1144 14.6463 11.6198 13.3597 9.8179 7.8912 12.4931 6.4573 12.4109 16.8591 18.6789 17.1504 10.1916 0.5347 0.7688 0.6738 0.5999 0.6674 0.7880 0.8082 0.6845 0.7582 0.5997 0.5145 0.7494 0.4307 0.7455 0.8693 0.9259 0.8788 0.6055 Note that V5 and V18 were eliminated during measure ment purification Table 1 provides unstandardized coefficients, stand ard errors, and t-values for each individual item. These numbers provide information about the local fit, th at is, how well each individual item related to its respective factor. Each of the coefficients is large and signi ficant at the p<0.01 level. Table 2 also provides c oefficient alphas for each factor after the measure purificati on process. The coefficient alphas ranged from 0.66 58 to 0.8960. De Vellis [5] noted that alpha levels below 0.6 are unacceptable. Goodness of fit is determined by comparing the stru ctural model (full maintained model) to alternative models. One tests alternative models by sequentially deleti ng or adding paths [6]. The measures of goodness-of -fit are shown in Table 2. After deleting 2 paths representi ng Hypothesis 7 and Hypothesis 2, the results of th e test of the overall fit of the model in Figure 2 are provid ed below. The chi-square statistic is significant. Other goodness-of-fit indices indicate an acceptable fit of the structural model to the data, especially giv en the exploratory nature of the study. Bentler’s Comparat ive Fit indices (CFI) and Bentler and Bonett’s Non- normed Index (NNFI) are above the desired 0.90 level [7,8] and thus indicate good fit. The ratio of chi-square to 366 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES degrees of freedom is 2.138, which is below the rec ommended 3.0 threshold [8,9], which indicates a goo d fit. The adjusted goodness of fit index is above the des ired 0.80 threshold [9-11], although below the cons ervative 0.90 threshold recommended by Bagozzi and Yi [12]. Table 2: Measures of goodness of fit for the struct ural model Model χ2 df χ2 / df ratio* CFI** NNFI*** Null model Uncorrelated factors Full maintained model H7 path deleted H2 path deleted 2355.1991 1360.2646 431.8911 431.9934 433.5983 210 209 201 202 203 - - 2.15 2.14 2.14 - 0.4997 0.8997 0.9000 0.9004 - 0.4470 0.8847 0.8857 0.8858 * χ2 / df ratio < 3 (Hartwick and Barki [9]; Hair et al. [8]) ** CFI > 0.9 indicates a good fit of the data to th e model (Bentler and Bonett [13]) ***NNFI > 0.8 indicates an acceptable fit (Wheaton et al. [10]; Segars and Grover [11]; Chau [12]) These results are indicative of an acceptance fit o f the model to the data, especially given that many of the measures used in this study were either developed f or the study, or adapted from other scales. The R 2 values for the three structural equations, which represent the variance explained by the endogenous factors o f FP, PIM, and BR, are 0.23, 0.52 and 0.4 respectively. F or example, 0.4 is the variance in buyer-supplier relationship (BR) explained by strategic purchasing (SP) and production and inventory management (PIM) . The results of the hypothesis tests, represented by individual paths between factors within the model, are included in Table 3, shown in Figure 2 and addresse d in the following paragraphs. SP BR FP QC PIM H1 H2 H3 H6 H5 H7 H4 Denotes non-significant paths Denotes significant paths Figure 2: Structural M odel .446* .3077* .2410* .26* .723* * Standardized path coefficients (p < 0.01) 367 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES Table 3: Test results of the structural model Hypothesis Path from To Regression weight Standard error t-value 1 3 4 5 6 SP BR QC QC PIM BR FP PIM FP BR 0.4180 0.4031 0.7424 0.3185 0.2554 0.0907 0.1176 0.1012 0.1162 0.0985 4.6074* 2.5933* 7.3338* 2.7417* 2.5933* * Significant at p < 0.01 H1, representing the path from the strategic purcha sing factor (SP) to the buyer-supplier relationship factor (BR), was positive and significant (p<0.01). This r esult supports the hypothesis that as strategic pur chasing increases, it is expected that the firm will increa se communication, cooperation, and coordination wit h key suppliers. Firms that do long-term planning and con sider purchasing to be strategic are also likely to build long-term cooperative relationships with their key suppliers. This trend has also shown in Thai indust ries where the purchasing firm seeks to identify potenti al suppliers and determine their qualifications as a supplier of the firm either through formal or informal suppl ier evaluation systems. H2, testing the effect of the strategic purchasing (SP) on the factor of the firm’s financial performa nce (FP), was not significant and recommended to be dropped d uring the Walt test, which analyzed the fitness of the structural model to the data, there was no signific ant increase in the model chi-square. This may refl ect the fact that SP does not appear to have a direct impact on FP. However, it appears to have an indirect impact on the firm’s financial performance via the factor of BR ( as path H1 is positive and significant). This sugg ests that improving strategic purchasing without building coo perative buyer-supplier relationship may not improv e the firm’s financial performance. H3, testing the effect of the buyer-supplier relati onships (BR) on the factor of the firm’s financial performance (FP), was significant and positive (p<0.01). This r esult lends support to the notion that when buyer-s upplier relationships become more cooperative with key supp liers, these firms will also have higher levels of financial performance with respect to return on investment, p rofits as a percent of sales, net income and presen t value of the firm. H4, representing the path from the quality expectat ion (QC) to the production and inventory management factor (PIM), was positive and significant (p<0.01) . Defective units from a production line need to be returned for rework or rejected as scrap. Appropriate contro l mechanisms are then required to integrate the ret urned defective units into the producers’ production plan ning. From a production management point of view, t hese activities are no different from other production p rocesses. H5, representing the path from the quality expectat ion (QC) to the firm’s financial performance factor (FP), was significant and positive (p<0.01). This result lends support to the notion that quality practice c an generate a sustainable competitive advantage. In the traditi onal paradigm, firms are concerned with company-cen tered issues such as price, product quality, and delivery time. In the new supply chain quality paradigm, su pplier- customer relationships and co-making quality produc ts have evolved as the major issues. 368 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES H6, testing the effect of the production and invent ory management (PIM) on the factor of the firm’s fi nancial performance (FP), was not significant and should be discarded as during the Walt test, which analyzed the fitness of the structural model to the data, there was no significant increase in the model chi-square . This result is surprising in that it suggests that the producti on and inventory management has little direct impac t on the firm’s financial performance. Moreover, this result contrasts with the result of H5, which suggests th at the QC has a direct effect on FP. The results of the analy sis indicate that the linkage between specific inte rnal capability factors and overall financial performanc e is not always clear. As this study reflects an ex tension of the manufacturing enterprise to encompass the entir e supply chain as the competitive unit, other facto rs may have a stronger effect on it. As a result, improvin g this individual performance does not necessarily imply immediate improvement in organizational financial p erformance, since there are many other intervening factors, which may have an impact on organizational results. H7, testing the effect of the production and invent ory management (PIM) on the factor of buyer-supplie r relationships (BR), was significant and positive (p <0.01). It reflects an extension of the manufacturi ng enterprise to encompass the entire supply chain as the competitive unit. High levels of process integr ation across firms are characterized by greater coordinat ion of the firm’s logistics activities with those o f its suppliers. This blurs organizational distinctions b etween the production and inventory management acti vities of the firm and those of its suppliers. Thus, a sin gle construct termed “logistics integration” is use d to describe the study of the steps taken by firms towards proce ss integration along the supply chain. 5.0 Conclusions The study provides empirical evidence for ever-grow ing strategic nature of supply chain management functions in Thailand. It reemphasizes the critical role of strategic purchasing functions in building strategic and collaborative buyer-supplier relationships. It also shows that internal operational aspects includ ing production and inventory management can impact on s trategic alliance formation. This strategic allianc e formation, in turn, exerts a significant impact on the financial performance of the firm. The signific ant positive link of strategic alliance to performance construct s reaffirms the importance of collaboration in a su pply chain context. Cooperation between buyers and suppliers is a key f actor for successful integration. To successfully m anage a buyer-supplier relationship, a firm may need to dev elop a strategic purchasing function. In addition, to achieve business success, it is imperative for firms to exc el in their internal operations, on which the effic ient and effective flows of goods and information in the sup ply chain depend. The internal manufacturing operat ions function is related to the goals of a business, the types of resources it utilizes, and the tasks of m anagement. It organizes, plans, controls, and improves the use of process, inventory, work force, and facilities and equipment in order to appropriately determine the ranking of the competitive priorities – price, quality, depend ability, flexibility, and time – thereby providing short-ter m profit, long-term profit, and improved market sha re. References [1] Fogarty, W.D, Blackstone, H.J. and Hoffmann, R.T. ( 1991) Production & Inventory Management, Second edition, South-Western Publishing Co., USA. 369 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LINKING SUP P LY CHAIN MANAGEMENT CAPABILITY AND MANU FACTURING OPERATIONS COMPETENCE WITH ORGANIZATIONAL PERF O RMANC E: A CASE STUDY OF THAI INDUSTRIES [2] Dwyer, R., Schurr, P, and Oh, S. (1987) Developing buyer-seller relationships, Journal of Marketing, V ol. 51, pp. 11-27. [3] Anderson, J.C. and Gerbiing, D.W. (1988) Structure equation modeling in practice: a review and recomme nded two-step approach, Psychological Bulletin, Vol. 103 , No. 3, pp. 411-423. [4] Bollen, K.A. (1989) Structural Equations with Laten t Variables, Wiley, New York, U.S. [5] DeVellis, R.F. (1991) Scale Development: Theory and Applications, Sage Publications, Newbury Park, CA. [6] Handfield, R.B. (1993) A resource dependence perspe ctive of just-in-time purchasing, Journal of Operat ions Management, Vol. 11, No. 3, pp. 289-311. [7] Byrne, B.M. (1994) Structural Equation Modeling wit h EQS and EQS/Windows: Basic Concepts, Applications , and Programming, Sage Publications, Thousand Oaks, CA. [8] Hair Jr., J.F., Anderson, R.E., Tatham, R.L., Black , W.C. (1995) Multivariate data analysis with readi ngs, 4 th edition, Prentice-Hall, Eaglewood Cliffs, NJ. [9] Hartwick, J., Barki, H. (1994) Explaining the role of user participation in information system use, Ma nagement Science, Vol. 40, No. 4, pp. 440-465. [10] Wheaton, B., Muthen, B., Alvin, D.F., and Summers, G.F. (1977) Assessing reliability and stability in panel models, In: Herse, D.R. (Ed), Sociological Methodol ogy. Jossey-Bass, San Francisco, pp. 84-136. [11] Segars, A.H. and Grover, V. (1993) Re-examining per ceived ease of use and usefulness: a confirmatory f actor analysis, MIS Quarterly, Vol. 17, No. 4, pp. 517-52 5. [12] Bagozzi, R.P., Yi, Y. (1988) On the evaluation of s tructural models, Academy of Marketing Science, Vol . 6, No. 1, pp. 74-93. [13] Bentler, P.M. and Bonett, D.G. (1980) Significance tests and goodness-of-fit in the analysis of covari ance structures, Psychological Bulletin, Vol. 88, pp. 58 8-606. 370 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGRATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS Colm Burns *, Nola Hewitt-Dundas Queen's University Management School, Queen's Unive rsity Belfast, Belfast, BT7 1NN * c.burns@qub.ac.uk Abstract This paper critically reviews the literature on the deter minants of IT adoption, specifically differentiating between discrete IT and int egrated IT systems. Integrated IT is exemplified by ERP; discrete IT is represented by the wide variety of discrete IT/IS examined in the adoption and implementation literature , including advanced manufacturing technology (AMT), CAD/CAM and expert systems. The central argument of the paper is that integrated IT systems differ markedly from discrete systems, and that differences exist in the organisational factor s that determine the successful implementation of each. A taxonomy of proposed differ ences between the determinants of integrated and discrete IT implementation is prese nted. Keywords: Information systems, Integration, Adoptio n, Implementation. 1.0 Introduction Advances in business software are supporting greate r integration within organisations of previously di screte functional IS. Since the mid-1990s, the interfacing of functions that were previously programmed and t ested separately has led to the growth of systems such as enterprise resource planning (ERP), customer relat ionship management (CRM), supply chain management (SCM) and digital manufacturing (DM). Prior to these developments, research on technology adoption and i mplementation focused primarily on single, discrete new technologies, producing perhaps the most mature str eam in IS research [1]. In general, this research h as led to broad consensus on the factors important in the imp lementation of discrete IT/IS. However, with the gr owth of integrated IS, research on the factors influencing implementation and adoption [2]-[3] has been less c onsistent than that for discrete IT/IS [4]. Research on the adoption and implementation of inte grated IS/IT, and on ERP in particular, has tended to study its implementation in a similar manner to tha t of discrete IT systems [5]-[6]. This approach how ever 371 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS ignores the integrative dimension of such a system, which defines and elevates it above a legacy syste m servicing a single department [7]. It is arguable t hat ERP in fact belongs to a different category of IT than the standard legacy system (that of ‘integrated IT’), a nd that a distinction should be drawn between ERP implementation research and IT/IS implementation re search [8]. In this paper we review the literature on the deter minants of IT adoption and implementation, specific ally differentiating between discrete and integrated IT systems. Discrete IT is represented by a range of I T/IS in the literature including advanced manufacturing technol ogy (AMT), CAD/CAM and expert systems, while integrated IT/IS is exemplified by ERP. The central argument of the paper is that integrated IT system s differ markedly from discrete systems, as reflected in the organisational factors determining the successful implementation of each. The remainder of this paper is structured as follows. In Section 2, we highlig ht the range of factors identified in the research that in fluence the adoption of discrete and integrated IT systems. In Section 3 these factors are discussed, considering in particular, if these factors differ for discrete as compared to integrated IT systems. Finally, a taxonomy of di fferences is presented in the concluding section wi th some consideration of the organisational implications of this. 2.0 Determinants of IT/IS Implementation & ERP Implementation A review of the literature reveals that the determi nants of adoption and implementation of IT/IS and E RP fall into three categories (Fig. 1). First, decisions ma de prior to adoption lay the foundations for the pr oject and, as such, strongly influence its ultimate success or fa ilure. Second, the inherent characteristics of the organisation determine its receptivity toward change [9], and th erefore its capacity for successful uptake of IT. F inally, the appropriateness of the managerial approach to the p roject is believed to strongly influence successful implementation. In this process (Fig. 1), no distinction is made be tween discrete and integrated IT. Rather, factors a re classified according to the nature of their influence, which c losely parallels with the stage of the process duri ng which they are most important. Initial selection is deter mined by technology characteristics and by the cons traints of the project. After the adoption decision, the organ isation’s characteristics become critical in determ ining the initial interface between technology and organisati on. In the implementation phase, success becomes a function of how effectively the project is organise d and managed. In Section 3 we consider how specifi c determinants differ for discrete and integrated – a s typified by ERP – adoption and implementation. 3.0 Comparison of Determinants of IT/IS Implementation & ERP Implement ation 3.1 Pre-Adoption: Preliminary Decisions After deciding to introduce IT, managers select a t echnology with the specific features addressing the perceived need. Rogers [10] identifies five techno logy constructs as determinants of ‘innovation diff usion’: relative advantage, compatibility, complexity, tria lability and observability (Table I). With the exce ption of ‘complexity’, the greater the extent to which an IT possesses these traits, the smoother will be its a doption and 372 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS implementation. Goodhue and Thompson’s [11] concept of ‘task-technology-fit’ is analogous to Rogers’ ‘compatibility’, advocating selection of IT which i s equal or suited to the task for which it is being adopted. Rogers’ construct, however, also incorporates “comp atibility with what people feel or think” [12, p. 3 3]. Davis et al. ’s [13] perceived usefulness and perceived ease of use concepts, meanwhile, correspond to Rogers’ ‘relative advantage’ and ‘complexity’ respectively. An IT’s position along Rogers’ constructs has impl ications for adoption and implementation success; as such, m anagers should evaluate the complexity, trialabilit y, etc. of the various options before selecting an IT solut ion. Process of Adoption & Implementation Pre- Adoption Adoption Decision Adoption Implementation Preliminary Organisational Implementation Decisions Influences Management Contributing Variables Determinant Fig. 1. Consolidated determinants of IT/IS & ERP a doption & implementation by category Table I Technology Characteristics and IT Selecti on Decisions Technology Characteristic Issue for Consideration in Selecting IT Relative Advantage (Perceived Usefulness) I s it sup e ri o r to the pre c e d i n g tec h n o l o gy? Compatibility (Task-Technology Fit) Do e s it fit with tas k s for whi c h it is nee d e d & sta f f atti t u d e s ? Complexity (Perceived Ease of Use) Ho w easy/di ff i c u l t is it to ado p t and us e ? Trialability I s it pos s i b l e to ex p e ri me n t & try it out? Observability I s the tec h n o l o gy vis i b l e & are its ben e f i t s vis i b l e ? Source: [10]-[11]-[13] With regards ERP, Somers and Nelson [4, p. 260] sug gest that applications should be selected based on projected “budgets, timeframes, goals and deliverab les”. In addition, it is also important to consider potential vendors [14], and to take into account the number a nd experience of previous adopters [15] [4]. As is also System Selection Decision of Scope Workforce Traits Firm Size Cultural Traits Formal i ty Central i sati on Compl exi ty Age Tenure Diversi ty Attitude Structural Traits TTF R. A dvantage C ompati bi l i ty Compl exi ty (Ease of use, Useful ness) Trial abi l i ty Observabi l i ty Budget Timeframe Vendor Performance Innovati veness TR Data Accuracy Cost Risks Benefi ts Time Resources Technology/ Organisation Balance C ross-functi onal Internal & external Implementation Team Communication D i ssemi nati on Adverti sement 2-way channel Top Management Support P rovi si on Consi stency Mediati on Presence of Champions Customisation/ BPR V endor support Focus on use & understandi ng Training Performance Measurement Planning U sage rates Benefi ts Impact on firm 373 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS found with discrete IT, compatibility is important in terms of acquiring high-performance modules for the critical areas of the firm [16], and matching the s oftware to the business’ needs [17]. Yet, due to th e larger scale of ERP systems and the IT vendor’s role in im plementation success, selecting a suitable integrat ed system is more complex than in cases of discrete IT . As well as selecting a system, in ERP projects th e scope of implementation must be determined, often in term s of the number of modules needed [3]. In contrast, the scope of discrete IT implementation is more fixed, typically comprising one application within one fun ction. ERP system selection therefore requires greater con sideration, particularly in terms of the scope of implementation and the potential support of vendors . 3.2 Post Adoption Decision/Pre-Adoption: Organisational Influences 3.2.1 Workforce Characteristics Research suggests that a workforce with a high prop ortion of younger employees [18] from diverse backgrounds [19], who have worked in the firm for l onger [20] is most suited to IT adoption and implementation. The workforce ‘complexity’ of the o rganisation (i.e. the extent to which the workforce is specialised, highly educated and professional) is a lso positively linked to IT implementation success [21]. Successful IT adoption and implementation also depe nds on the support of a “critical mass of stakehold ers” [22, p. 98], so monitoring and understanding workfo rce attitudes is important. This helps ensure that the IT fits with stakeholder needs and that the necessary ‘ener gy’ for implementation is present [23]. These workf orce characteristics, and fostering a positive attitude towards implementation, are important for both disc rete IT and ERP adoption and implementation. 3.2.2 Firm Size Larger firms are consistently found to be more capa ble of adopting and implementing new technology [24 ]- [25]. These firms perceive less risk in the decisio n to adopt new technology and as such, tend to be m ore willing, and better equipped, to adopt [24]. At th e same time, Robinson [26] suggests that in smaller organisations with greater physical and cognitive p roximity of functional areas, more effective inter- function coordination and communication can be achieved with out integrating their systems. The pressure to adop t an ERP system therefore increases as firm size increas es. The feasibility of both discrete and integrated IT adoption and implementation therefore increases wit h firm size. 3.2.3 Structural Characteristics In general, structural complexity (e.g. the number of different divisions within the organisation) is positively correlated with IT adoption and implementation succ ess [27]. An organic, informal organisational struc ture is found to facilitate IT/IS adoption [28]-[29]. In o ther words, less centralised authority and decision -making within an organisation is conducive to workers adop ting the unfamiliar [30]. However, it is suggested that a centralised organisational structure is more conduc ive in the implementation stage, when tight control leads to more optimal and faster implementation of the IT [3 1]-[32]. 374 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS Essentially, formalisation and centralisation deter mine how suited an organisation’s structure is to a dopting IT. Optimum ERP implementation, however, involves changing the organisation to fit with the new system [4]- [33], so pre-existing organisational characteristic s should have a lower effect on implementation succ ess. Therefore, organisational structure would be expect ed to be less important in ERP adoption and implementation, where the choice of IT ought to dic tate the company’s level of centralisation and formalisation, not vice versa. 3.2.4 Cultural Characteristics An organisation’s culture is the “commonly held bel iefs, attitudes and values which...provide...rules for behaviour” [34, p. 70]. Looking at certain imbedded cultural features of an organisation can help info rm IT adoption decisions. Morrison’s [35] concept of ‘org anization dispositional innovativeness’ (ODI), for example, gauges how inherently innovative and recep tive to new technology the firm is. ‘Personal innovativeness’ (PI) [9] and the Technology Readine ss Index (TRI) [36] are similar constructs, but the y cumulatively measure receptivity to technological c hange at the individual, as opposed to organisation al, level. Higher levels of these constructs mean less inheren t resistance to new IT, and a smoother implementati on [23]. While culture influences IT/IS adoption and impleme ntation in general [37], it is regarded as “the uni que factor affecting ERP systems implementation success ” [17, p. 57]. An underlying predisposition to inte gration, high TRI [36], positive attitudes to change, and an existing capability for computing are found to red uce the upheaval of integrated IT implementation [3]. ERP u sers are required to input data in a timely and pre cise fashion [17]: a culture of data accuracy is therefo re more important for ERP implementation success th an for discrete IT [26]. 3.3 Post-Adoption: Management of Implementation 3.3.1 Organisation/Technology Balance IT implementation is determined not only by the ben efits of the IT, but also by how it links to the sp ecific objectives of the organisation, and by how effectiv ely the organisation is reconfigured in order to accommodate the IT [23]. As such, understanding and appropriately modifying the organisational context in which the technology is implemented is of great imp ortance [2]; this may mean training and reskilling of workers, or new managerial approaches [23]. Benjami n and Levinson [23] advocate an approach to plannin g IT implementation in which the organisation’s curre nt state, desired final state and a number of trans itional states are set out. The target final state is one o f equilibrium, so the change process can be thought of as a quest to rebalance the organisation, and its processes, in line with the new IT. IT/IS implementation success therefore requires that changes made in one of thes e three areas be offset by change to one or both of the other areas [23]. Successful management of ERP implementation also de pends on maintaining equilibrium between technical and organisational issues [38]-[39], with neither d ominating implementation decisions. Success depends on 375 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS optimising the fit between the ERP application and the organisation; this requires effective monitoring and management on both fronts. The greater scope of an ERP project and the obligation to change the organi sation in line with the new system [4] make it even more important and more difficult to a chieve this balance. 3.3.2 Planning Planning is deemed crucial in discrete IT/IS implem entation [40] with research tending to focus on pre - implementation planning, i.e. measures to ensure th at adoption is justified and that the best technolo gy is selected. For example, assessing the organisation’s current level of IT sophistication [41]-[42] and s earching for appropriate technologies [37] is advised, but p lanning how the technology is then to be implemente d is not a prominent consideration. An explanation for this may be that, if an appropriate IT is selected and a timeframe set, the actual adoption and implementati on will, to some extent, ‘take care of itself’. A thorough implementation plan is also considered c ritical to ERP success, preferably detailing the “p roposed strategic and tangible benefits…costs, risks and tim eline” of the project [3, p. 291]. The resources re quired for implementation should be carefully calculated, and the anticipated benefits set out for different area s of the organisation [2]-[16]. The needs and best interests of the organisation should be kept at the heart of the project plan [16], which should, in turn, be kept at the he art of the ERP implementation [2]. Studies have sho wn greater rates of ERP success in firms who formulate and adhere to an implementation plan and adhere to it [16], and poor performance in cases of inadequately planned ERP implementation [43]. In contrast with the IT/IS implementation literature, ERP findings empha sise the need for detailed planning of post-adoption, implementation activities, as these are much more c omplex and unpredictable [33]. Pre-adoption plannin g is necessary for both discrete IT and ERP implementati on, but detailed planning of the actual implementat ion process is much more important in cases of integrat ed IT. 3.3.3 Implementation Team Cross-functional implementation teams are found to facilitate discrete IT/IS implementations [40]-[44] . These almost invariably involve staff from diverse discip lines, and allowing these staff to co-operate and c ollaborate means all interests are represented and fully discu ssed [23]-[28], leading to optimal implementation. The use of dedicated cross-functional teams is also consistently linked to ERP implementation success [23]- [45]-[46]. ERP is, by nature, a cross-functional co ncept, so it stands to reason that its implementati on should be coordinated among the members of different funct ions. Implementation team membership ought to refle ct both business and technical interests within the fi rm, in order to achieve balance between organisatio nal factors and issues arising from the IT itself [2]. All divisions involved in or affected by the implem entation should be represented [46], and managers placed in teams alongside lower-level personnel [45]. Also, effective implementations involve careful use of ex ternal consultants [4] alongside the firm’s own bes t staff [47]. This juxtaposes expertise on ERP systems and their implementation with expertise on the organisa tion and its processes [17], optimising the interface be tween the ERP system and the organisation and leadi ng to successful implementation [46]. As with discrete IT implementation, concurrent input from all stakehol ders 376 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS will ensure all interests are represented, and is s hown to shorten the ERP implementation period [16]. However, while cross-functional teams are desirable for standalone IT implementation, they are indispensable for successful ERP implementation. 3.3.4 Customisation / Business Process Reengineering (BPR) The best discrete IT/IS system for an organisation can generally be chosen from among a wide range of standard alternatives, which are essentially sold c ustomised. As such, there is good fit between the t echnology and the organisation from the outset, and customisa tion and BPR are therefore of little concern. ERP, on the other hand, is a “packaged solution” [3 , p. 286]. Optimum performance occurs when it is ap plied exactly as it was developed and programmed, and the organisation’s processes are moulded to fit with i t, while altering the software will preclude the future adop tion of newer versions and updates, eroding value f or money and risking obsolescence [3]. As such, keeping cust omisation of the ERP software to a minimum should b e a priority during introduction and implementation [4] . Successful adoption of new IT requires balance be tween designing the IT to fit with the organisation and r estructuring the organisation to accommodate the IT [48]. The implication of there being no changes to the ER P software is that all the changes necessary to ali gn the firm and the new system must be made to the organis ation’s processes [17]. Optimal ERP implementation is therefore characterised by low levels of software c ustomisation and high levels of BPR [4]; failure is linked to underestimation of the need for organisational chan ge [17]. The greater range of tailored discrete IT solutions, meanwhile, makes alterations to the technology or t he organisation less of a necessity. 3.3.5 Communication Effective communication is shown to support discret e IT/IS implementation [37]. Pre-implementation, st rong communication among those installing the IT will op timise its deployment; in the implementation phase, organisation-wide communications become more import ant, as the effective dissemination of information about the technology can help train, persuade and r eassure staff members [49]. Communication is also emphasised as a determinant o f ERP implementation [3]. Initially, the implementa tion must be justified to stakeholders [16]. There shoul d be clear communication from management of the functionality and benefits of ERP [2], of what the workforce should expect from the system [4], and of what is, in turn, expected of the workforce [3]. Informing w orkers helps to minimise resistance to the new appl ication [4] and build enthusiasm. As well as communicating general information about the ERP system, it is imp ortant to specifically ‘advertise’ the progress of impleme ntation and associated improvements within the firm [33]. Provision for two-way communication should be made [3]-[23], ideally extending to user involvement in key decisions about ERP implementation [16]-[33]. This creates the impression among users that they are in control of the implementation [17], and maximises f it between system functionality and user needs. Ign oring users’ “needs and expectations” is linked to IT imp lementation failure [2, p. 509]. All stakeholders m ust be consulted; otherwise, the project will be biased to wards the interests and agendas of certain groups, and long- term implementation success will be jeopardised [2] . 377 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS These ERP communication factors are generalisable t o cases of discrete IT/IS implementation [37]-[49]. However, there is likely to be no precedent for, an d therefore little knowledge of, ERP within the org anisation, so a greater volume of information may need to be c onveyed. Also, the scope of communication will need to be broader than for discrete IT/IS implementation, as potentially every division of the organisation w ill be involved in or affected by the implementation. Comm unicating justification, features, benefits and pro gress is important for success in any IT implementation proj ect, as is provision for user feedback, but more in tense communication with a greater proportion of the work force is necessary during ERP implementation. 3.3.6 Champions A champion is a manager “who is intensely intereste d and involved with the overall objectives and goal s of the project” [50, p. 15]. Their role is to promote the IT to their superiors and subordinates [51], and to provide, or encourage the provision of, necessary funding and r esources for the implementation project [23]. Champ ions may be appointed or adopt the role voluntarily, but it is argued that the appointment of a champion is symptomatic of a lack of spontaneous support, and o f problems with the implementation as a whole [23]. The visible support of champions is found to facili tate implementation across a range of IT [52], incl uding ERP systems: Sumner [46], for example, advocates th e appointment of an individual to promote the integration initiative to the whole organisation. T he role of champions is particularly important in e arlier stages of ERP implementation [4]: the positive mess ages conveyed will then disseminate and take hold d ue to the reputability of the source. The presence of a c hampion is strongly correlated with implementation success for both discrete IT and ERP systems. 3.3.7 Top Management Support The support of top management for any newly adopted IT is consistently cited as crucial to its success ful implementation [53]-[54]. It is ultimately managers who decide whether to adopt a technology or not [4 2], so their backing is instrumental from the outset. Howe ver, managerial support should not be limited to in itial enthusiasm: it is important that managers continue to show their commitment, and that they periodicall y ‘energise’ the implementation [53]. Their advocacy of the IT can serve as an example to the rest of th e organisation, radiating through the workforce and b oosting the chances of successful implementation [2 1]. Top management support has also been correlated wit h successful ERP implementation; Somers and Nelson [4], in fact, rate it as the greatest predictor of ERP success. Senior managers should emphasise the s ignificance of the project, while demonstrating their own commi tment and involvement. This will involve allocating sufficient resources and time to the implementation effort [33], and generally providing strong leader ship [16]. Several factors make management support and all-rou nd effective management particularly important for ERP implementation. Implementing ERP requires cooperati on and coordination among members of many divisions : strong management will help settle disputes and giv e cohesive direction to all those involved [17]. Al so, major organisational change is often a part of ERP implem entation, and effective management is crucial in su ch 378 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS times of change [17]. Top management support is cri tical for any IT implementation, but attentive and effective management is more important in the conte xt of the disruption and broad scope of an ERP proj ect. 3.3.8 Training Training in new IT is among the most direct and eff ective ways to encourage adoption [37]. In particul ar, expert training and support from onsite staff of th e IT vendor is linked to successful implementation [52]. IT/IS training supports implementation by optimisin g use of the new system, but also by increasing understanding thereof [2]. ERP system training is similarly important, but mus t focus even more on increasing understanding [17]. It must encourage proper use of the software, but this first requires that users develop personal underst anding of, and enthusiasm for, the integrative dimension of ER P [3]. As such, effectively training integrated IT users is even more challenging, and arguably more important, than discrete IT training. Negative outcomes are recorded in cases where efforts are not made to pro mote optimal use and clear understanding of new ERP systems [17]. Somers and Nelson [4] recommend initi ating training and education as early as possible i n ERP implementation (even prior to adoption), and contin uing it throughout the process. However, while ongo ing training is important for any IT/IS implementation, it is the need to focus on r einforcing understanding which sets ERP training apart from the more practical foc us for discrete IT. 3.3.9 Performance Measurement As discussed in Section 3.3.5, any discrete IT/IS i mplementation should be preceded by a meticulous cost/benefit analysis, in order to justify and get backing for the project. This is, regrettably, as f ar as the evaluative efforts of many IT adopters go [2]. Howe ver, by then comparing predicted versus actual post - implementation costs and benefits, managers can mon itor how effectively the IT is operating [16], how successful it is [17] and whether it should be pers evered with. What is more, failure to adequately sp ecify and evaluate the anticipated benefits from an IT is bla med for heavy financial losses in IT/IS implementat ion [55]. Clear procedures for performance measurement are al so linked to successful ERP implementation [56]. Us age rates provide the most basic and reliable measure o f the system’s success: whether it is used by its i ntended users [2]. It is equally important to gauge and coo rdinate “how the organization should operate behind the implementation effort” [3, p. 291]. Performance sho uld therefore also be gauged in terms of the manage ment of the project and of its contribution to firm perf ormance [3]. It is crucial to set out desired benef its, performance levels and usage rates for any discrete or integrated IT prior to adoption, and to subsequ ently measure how the implementation project itself is go ing, and how the IT is affecting other areas of the firm. 4.0 Conclusion 379 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS Somers and Nelson [4, p. 270] conclude that “some o f the players and activities that are critical duri ng any IT implementation play an equally crucial role in ERP implementation”. This review of the IT/IS and ERP implementation literatures supports this position. However, the key point to focus on is not the parallels between the determinants of discrete and integrated IT adoption and implementation, but the significan t differences that appear to exist in the influence of certain f actors on one or the other. These differences fall into one of two categories. Firstly, there are those factors which influence al l IT implementations, but which are of greater importance during integrated IT adoption and implementation. F or example, organisational culture determines an organisation’s predisposition to change, and is the refore determinative of any IT adoption and impleme ntation. However, because the level of change required durin g ERP implementation is much greater, organisationa l culture is more influential on the process. Similar ly, communication is critical to the success of any IT implementation, but more so in cases of ERP, when m ore information must be conveyed to more people. System selection, implementation planning and top m anagement support also fall into this category. Secondly, certain differences arise due to fundamen tal disparities between discrete IT and ERP. The fa ct that the scope of an ERP implementation (i.e. the number of modules to be adopted) is variable, adds extra decision-making responsibility, for example. Also, ERP systems are inherently cross-functional, so a c ross- functional implementation team is a necessity, wher e it is merely desirable in cases of discrete IT implementation. The uncustomised nature of ERP soft ware gives rise to further differences: changes to either the organisation or the software are inevitable; th e influence of existing organisational structure is less due to the obligation to change the organisation; and main taining balance between organisational and technica l considerations is more difficult than in cases of d iscrete IT/IS implementation. The implication here is that ERP adoption and imple mentation is at least partly subject to different considerations than discrete IT adoption and implem entation, and that managers therefore need to appro ach such projects with a different configuration of fac tors in mind. It is arguable that the typically hig h ERP failure rates during the last decade [4] have, to some exte nt, been due to managers’ persistence in managing E RP implementations as they would manage any discrete IT implementation, where they ought to have been managed as integrated IT. References [1] Venkatesh, V. 2006, "Where to go from here? Tho ughts on future directions for research on individu al-level technology adoption with a focus on decision making ", Decision Sciences, vol. 37, no. 4, pp. 497-518. [2] Dhillon, G. 2005, "Gaining benefits from IS/IT implementation: Interpretations from case studies", International Journal of Information Management, vol. 25, no. 6, pp. 502-515. [3] Nah, F.F.H., Lau, J.L.S. & Kuang, J. 2001, "Cri tical factors for successful implementation of ente rprise systems", Business Process Management, vol. 7, no. 3, pp. 285-296. [4] Somers, T.M. & Nelson, K.G. 2004, "A taxonomy of players and activities across the ERP project life cycle", Information & Management, vol. 41, no. 3, pp. 257-278. [5] Ifinedo, P. 2007, "An empirical study of ERP su ccess evaluations by business and IT managers", Information Management & Computer Security, vol. 15, no. 4, pp. 270-282. [6] Hong, K.K. & Kim, Y.G. 2002, "The critical succ ess factors for ERP implementation: An organization al fit perspective", Information & Management, vol. 40, no. 1, pp. 25-40. [7] Al-Mudimigh, A., Zairi, M. & Al-Mashari, M. 200 1, "ERP software implementation: an integrative fra mework", European Journal of Information Systems, vol. 10, no. 4, pp. 216-226. 380 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS [8] Milford, M. & Stewart, G. 2000, "Are ERP implem entations qualitatively different from other large systems implementations?", Americas Conference on Information Systems, Association for Information Systems, Atlanta, GA. [9] Zmud, R.W. 1984, "An examination of 'push-pull' theory applied to process innovation in knowledge work", Management Science, vol. 30, no. 6, pp. 727-738. [10] Rogers, E.M. 1983, Diffusion of Innovations, 3rd edn, The Free Press, New York. [11] Goodhue, D.L. & Thompson, R.L. 1995, "Task-tec hnology fit and individual performance", MIS Quarterly, vol. 19, no. 2, pp. 213-236. [12] Tornatzky, L.G. & Klein, K.J. 1982, "Innovatio n characteristics and innovation adoption-implement ation: A meta- analysis of findings", IEEE Transactions on Engineering Management, vol. 29, no. 1, pp. 28-45. [13] Davis, F.D., Bagozzi, R.P. & Warshaw, P.R. 198 9, "User acceptance of computer technology: A compa rison of two theoretical models", Management Science, vol. 35, no. 8, pp. 982-1003. [14] Geishecker, L. 1999, "ERP vs. best-of-Breed", Strategic Finance, vol. 80, no. 9, pp. 62-67. [15] Roberts, H.J. & Barrar, P.R.N. 1992, "MRPII im plementation: key factors for success", Computer Integrated Manufacturing Systems, vol. 5, no. 1, pp. 31-38. [16] Mandal, P. & Gunasekaran, A. 2003, "Issues in implementing ERP: A case study", European Journal of Operational Research, vol. 146, no. 2, pp. 274-283. [17] Zhang, Z., Lee, M.K.O., Huang, P., Zhang, L. & Huang, X. 2005, "A framework of ERP systems implem entation success in China: An empirical study", International Journal of Production Economics, vol. 98, no. 1, pp. 56-80. [18] Beatty, C.A. & Gordon, J.R.M. 1988, "Barriers to the implementation of CAD/CAM systems", Sloan Management Review, vol. 29, no. 4, pp. 25-33. [19] Aspelund, A., Berg-Utby, T. & Skjevdal, R. 200 5, "Initial resources' influence on new venture sur vival: a longitudinal study of new technology-based firms", Technovation, vol. 25, no. 11, pp. 1337-1347. [20] Stuart, W.D. 2000, Influence of sources of communication, user charact eristics and innovation characteristics on adoption of a communication technology, PhD, University of Kansas. [21] Hadjimanolis, A. 2000, "A resource-based view of innovativeness in small firms", Technology Analysis & Strategic Management, vol. 12, no. 2, pp. 263-292. [22] Beckhard, R. & Harris, R. 1987, Organizational Transitions, 2nd edn, Addison Wesley, Reading, MA. [23] Benjamin, R.I. & Levinson, E. 1993, "A framewo rk for managing IT-enabled change", Sloan Management Review, vol. 34, no. 4, pp. 23-33. [24] Laosirihongthong, T., Paul, H. & Speece, M.W. 2003, "Evaluation of new manufacturing technology implementation: An empirical study in the Thai auto motive industry", Technovation, vol. 23, no. 4, pp. 321-331. [25] Cohen, W.M. & Levinthal, D.A. 1990, "Absorptiv e capacity: A new perspective on learning and innov ation", Administrative Science Quarterly, vol. 35, no. 1, pp. 128-152. [26] Robinson, P. 2003, Best of Breed (BOB) v Fully Integrated (FIS) [Homepage of Business Performance Improvement Consultancy], [Online]. Available: http://www.bpic.co.uk/articles/bobvfis.htm [2007, October 18]. [27] Damanpour, F. 1996, "Organizational complexity and innovation: Developing and testing multiple co ntingency models", Management Science, vol. 42, no. 5, pp. 693-701. [28] Ahmed, P.K. 1998, "Culture and climate for inn ovation", European Journal of Innovation Management, vol. 1, no. 1, pp. 30-43. [29] Argote, L. & Goodman, P.S. 1986, "The organiza tional implications of robotics" in Managing Technological Innovation: Organizational Strategies for Implement ing Advanced Technologies, ed. D.D. Davis, Jossey-Bass, San Francisco, pp. 127-153. [30] Cormican, K. & O'Sullivan, D. 2004, "Auditing best practice for effective product innovation mana gement", Technovation, vol. 24, no. 10, pp. 819-829. [31] Damanpour, F. 1991, "Organizational innovation : A meta-analysis of effects of determinants and mo derators", Academy of Management Journal, vol. 34, no. 3, pp. 555-590. [32] Zaltman, G., Duncan, R. & Holbek, J.M.A. 1973, Innovations and Organizations, Wiley, New York. [33] Holland, C.P. & Light, B. 1999, "A critical su ccess factors model for ERP implementation", IEEE Software, vol. 16, no. 3, pp. 30-36. [34] MacKenzie, S. 1995, "Surveying the organizatio nal culture in an NHS trust", Journal of Management in Medicine, vol. 9, no. 6, pp. 69-77. [35] Morrison, P. 1996, Testing a framework for the adoption of technological innovations by organizations and the role of leading edge users, Institute for the Study of Business Markets, Penn sylvania. [36] Parasuraman, A. 2000, "Technology Readiness In dex (TRI): A multiple-item scale to measure readine ss to embrace new technologies", Journal of Service Research, vol. 2, no. 4, pp. 307-320. [37] Mirvis, P.H., Sales, A.L. & Hackett, E.J. 1991 , "The implementation and adoption of new technolog y in organizations: The impact on work, people, and cult ure", Human Resource Management, vol. 30, no. 1, pp. 113- 139. [38] Laroia, A. 2002, "EAI facts and fiction", EAI Journal, vol. 4, no. 11, pp. 8-10. [39] Themistocleous, M. & Irani, Z. 2001, "Evaluati ng application integration: An exploratory case stu dy", Americas Conference on Information Systems , Association for Information Systems, Atlanta, GA, p p. 1376-1380. [40] Meyers, P.W., Sivakumar, K. & Nakata, C. 1999, "Implementation of industrial process innovations: Factors, effects, and marketing implications", The Journal of Product Innovation Management, vol. 16, no. 3, pp. 295-311. 381 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EXPLORING THE ADOPTION AND IMPLEMENTATION OF INTEGR ATED AND DISCRETE IT SYSTEMS -A FRAMEWORK FOR ANALYSIS [41] Bendoly, E. 2007, "Resource enablement modelin g: Implications for studying the diffusion of techn ology", European Journal of Operational Research, vol. 179, no. 2, pp. 537-553. [42] Chen, A.N.K., Sen, S. & Shao, B.B.M. 2006, "St rategies for effective Web services adoption for dy namic e- businesses", Decision Support Systems, vol. 42, no. 2, pp. 789-809. [43] Cooke, D. & Peterson, W. 1998, SAP Implementation: Strategies and Results , The Conference Board, New York. [44] Sohal, A.S. 1996, "Assessing AMT implementatio ns: An empirical field study", Technovation, vol. 16, no. 8, pp. 377-384. [45] Bahli, B. & Ji, F. 2007, "Ana assessment of fa cilitators and inhibitors for the adoption of enter prise application integration technology: An empirical study", Business Process Management, vol. 13, no. 1, pp. 108-120. [46] Sumner, M. 1999, "Critical success factors in enterprise wide information management systems proj ects", Americas Conference on Information Systems , pp. 297-303. [47] Light, B., Holland, C.P. & Wills, K. 2001, "ER P and best of breed: A comparative analysis", Business Process Management, vol. 7, no. 3, pp. 216-224. [48] Yusuf, Y., Gunasekaran, A. & Abthorpe, M.S. 20 04, "Enterprise information systems project impleme ntation: A case study of ERP in Rolls-Royce", International Journal of Production Economics, vol. 87, no. 3, pp. 251-266. [49] Burkhardt, M.E. & Brass, D.J. 1990, "Changing patterns or patterns of change: The effects of a ch ange in technology on social network structure and power", Administrative Science Quarterly, vol. 35, no. 1, pp. 104-127. [50] Materials Advisory Board 1966, Report on the Ad-hoc Committee on Principles of Res earch-Engineering Interactions, US National Materials Advisory Board, Washington D.C. [51] Leonard-Barton, D. & Deschamps, I. 1988, "Mana gerial influence in the implementation of new techn ology", Management Science, vol. 34, no. 10, pp. 1252-1265. [52] Efstathiades, A., Tassou, S.A., Oxinos, G. & A ntoniou, A. 2000, "Advanced manufacturing technolog y transfer and implementation in developing countries: The case of the Cypriot manufacturing industry", Technovation, vol. 20, no. 2, pp. 93-102. [53] Beatty, C.A. 1992, "Implementing Advanced Manu facturing Technologies: Rules of the Road", Sloan Management Review, vol. 33, no. 4, pp. 49-60. [54] Ettlie, J.E. 1984, "Implementation Strategy fo r Discrete Parts Manufacturing Innovation" in Microprocessors, Manpower and Society, ed. M. Warner, Gower Press, Brookfield, VT, pp. 3 1. [55] Keil, M. & Robey, D. 1999, "Turning around tro ubled software projects: an exploratory study of th e de-escalation of commitment to failing courses of action", Journal of Management Information Systems, vol. 15, no. 4, pp. 63-87. [56] Mabert, V.A., Soni, A. & Venkataramanan, M.A. 2001, "Enterprise resource planning: Common myths v ersus evolving reality", Business Horizons, vol. 44, no. 3, pp. 69-76. 382 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES CHARACTERISATION OF SME MIGRATION TOWARD ADVANCED TECHNO LO GIES Barton R.A. 1 ,Thomas A.J. 2 1. Manufacturing Engineering Centre, Cardiff Universi ty, UK 2. Cardiff Business School, Cardiff University, UK Abstract British SMEs are seen to be migrating toward higher value mark ets, by utilising advanced technologies. In the face of competition from low labou r cost economies, manufacturing best-practices are being promoted which reduc e cost and/or increase profit. However, it is not clearly understood by SMEs how increased product val ue can be offered without compromising control of material, labour and investment c osts. One possible solution is the introduction of Advanced Manufac turing Technologies (AMT) in order to improve product quality, reduce lead time s, and offer additional functionality or customisability. Cost must be controlled thr ough careful selection, detailed preparation and planning, and swift implementation and development stages. As part of a 3 year investigation, 300 companies were surveyed in orde r to determine their attitude toward AMT. Subsequently the companies were categorised into three levels of technological capability, and an ideal AMT implementat ion model created. The companies were revisited with a view to identify what c hanges had been adopted toward the end of the project, and how they had influenced bottom line perf ormance. Whether it was due to implementation of the model, or other s ources, it was seen that companies were migrating toward the higher categories of technol ogy utilisation. There were also significant numbers of companies falling bet ween the previously described categories; either side of the level two category, due to discrepancy between attitude and associated achievement, hence the categorisation it self needed to be re- developed. By considering the characteristics of these transitional comp anies, a revised strategy for AMT implementation should be developed for the f uture to tackle the weaknesses being revealed. Keywords: SMEs, Advanced Manufacturing Technology, Management, Survey. 383 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES 1.0 Introduction SMEs in manufacturing must recognize the threat of half-heated approach to change, and particularly th e potential for failure incurred by ineffective techn ology development. This will not only result in fin ancial loss, but more importantly a loss of ground in the race t o lead the market. As competition from low labour c ost economies continues, EC manufacturing companies mus t continue to move toward advanced technologies to remain competitive (Ettlie [1], Beaumont et al [12] ) by offering higher levels of customisation and te chnical service while maintaining Quality Cost and Delivery (QCD) performance against the global manufacturing market. At a base level, the requirement to keep abreast of technological developments in order to improve productivity, quality, range of products and other performance measures is now paramount (Quarashi et al [2], Jhang et al [13]) Despite the clear evidence for a need to acquire technical skills and implement new and effective technology into SMEs to ensure survival a nd sustainable growth, many companies are reluctant to move towards major investment to enable competitive advantage through technological capability. Furthermore, many companies are especially reluctan t to invest in technology pertaining to Automated, Intelligent and Computer Aided Engineering systems. The reasons for this are many. Primarily, companie s are initially deterred by the extensive capital investm ent required to develop such technologies and secon dly, the capabilities of the technology and the advantages i t brings to the average SME are not fully appreciat ed by the companies concerned. This, along with the fact that many SMEs do not have the technical and manufactur ing infrastructure to support AMT severely limits their success in various technology transfer initiatives (Thomas [3]). 2.0 Methodology 300 manufacturing based SMEs were targeted for asse ssment over a three-year period. A questionnaire wa s devised that identified accurately the current tech nological platform of the SMEs as well as defining their aspirations towards developing their operations, co mpany infrastructure, financial strength, skills ba se etc. It was decided that each SME would be visited by a res earcher rather than rely purely upon questionnaire feedback since this allowed for a more realistic an alysis of the company’s operations. The companies w ere selected from a range of industrial sectors and all were registered as Small to Medium Manufacturing Enterprises. 2.1 SME Categorisation The companies were profiled at an initial survey vi sit, considering the above aspects, and the detaile d results were published by Thomas et al [4]. This describe d a three level categorisation of the surveyed SMEs , which is outlined below. Category 1. Companies 384 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES These were companies that were happy with their cur rent customer base and had no significant aspiratio ns to develop their companies through the use of AMT. The se companies tended to be sceptical about the benef its of AMT and the benefits that it could bring by way of increased customer base, improved technology qua lity, reduced product cost and improved delivery performa nce etc. It can be argued that the development and implementation of AMT into such companies may not s imply be relevant to the markets they operate in. (22% of companies were considered to fall into this Cate gory) Category.2. Companies These were companies who constantly try to improve and develop their company operations. These compani es in general could see the long term benefits of usin g AMT in developing their product or process but, d id not have the in company skills base, financial resource s and knowledge on how to introduce AMT into their companies. (65% of companies were considered to fal l into this Category) Category.3. Companies These were companies who had considerable experienc e of introducing AMT into their company and as such had a financial model which returned a large amount of their yearly profits into the purchasing of tec hnology to continually develop their manufacturing operatio ns. Without exception it was these companies that a chieved a greater market share and mostly had contracts wit h the larger manufacturing industries producing pro ducts in high value markets. (13% of companies were consider ed to fall into this Category) 2.2 Technology Implementation Model Subsequently, a model for AMT implementation was pr oposed to the companies as a composite model of aspects seen in best practice throughout the lifecy cle of a technology implementation project, these w ere seen to be largely common to improving, and especially c at.3, companies. The model is described in full by Thomas et al [4] but essentially comprises the foll owing 5 stages: Stage 1 - Company Analysis and Planning Stage Stage 2 - Technology Planning Stage Stage 3 - Technology Selection Stage Stage 4 - Technology Process Engineering Stage Stage 5 - Technology Development Stage • Technology and Resource Capacity Building • Knowledge Development and Management • Technology Performance Enhancement • Technology Improvement Once this model was proposed, some companies natura lly selected elements of the model as particularly pertinent to their business and adopted those pract ices, while others chose to adopt the carefully str uctured hierarchy to use the model as a chronological ‘road -map’ for technology development, and as expected a number of companies chose to ignore the model compl etely. This begs the question; could this mixed res ponse be monitored to try and evaluate the overall effect iveness of the model? 385 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES The initial answer was that the spread of applicati on was seen to be too varied, and the nature of the mixed industrial sectors proved to be too broad, such tha t there was no like-for-like comparison. However, a number of test companies were examined and the results pro ved favourable for those adopting the model, which correlated directly with the character of the compa ny as profiled by the initial categorisation. Gains were seen in on-cost delivery, on-time delivery and right-fir st-time delivery of technology development projects . Each of which yielded tangible improvements to the company in QCD performance, hence bottom line profitability . However this was a sample, hence a fairly superfici al view and continuity of project management is not assured, regardless of success experienced historic ally. In order to review what progress was being ma de by the SMEs surveyed, a second Phase review survey was undertaken by means of a second round of visits. In the interests of research integrity, the survey rep orts were similar in format so that the categorisat ion of the SMEs could be compared directly with the initial re sults. 3.0 Results The results of the second phase survey are shown in the summary table, table I, and indicate that ther e was some variation against initial findings; cat.1 and cat.3 companies reduced, as a percentage of the tot al, while cat.2 companies increased, as a percentage of the t otal. It is important to note that the percentages given in the results, see table I, do not demonstrate tha t the overall number of companies surveyed in the second phase wa s lower than original. This was different due to tw o additional conditions, which could be demonstrated as ‘virtual’ categories zero and four: a small num ber of category 1 companies had ceased trading when the la ck of ambition had led to the demise of that compan y (there were 5 instances), and a similarly small num ber of companies had been merged/acquired by larger partners when the level of development had made the m a prime candidate for acquisition before becoming a threat to the larger corporations (there were 3 ins tances). In Addition to the three fundamental categories ide ntified, the review survey highlighted a distinct s plit in the cat. 2 companies into three factions: • Category 2.1 Companies; is a mixture of previously cat.1 and cat.2 companies. The cat.1 companies are those which have realised that survival of the busi ness may rely on adoption of new technologies, to prevent competition from low cost economies and hig h product specification from cat.3 competitors continuing to erode the customer base beyond that w hich will support the company. The old cat.2 companies are those which have previously performed well on the basis of a strong product or process portfolio, but have failed to maintain an effective succession plan and obsolescence has led to a redu ction of company performance. The key for these companies is to recognise this and that action is required t o update the business capabilities. • Category 2.2 companies; are the same in characteris tic to those seen before, but less in number, and m ore significantly consists of the same companies as bef ore, i.e. many companies judged to be underachievin g in 386 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES terms of failing to maximise the effect of the deve lopment they are implementing are still not learnin g from mistakes and continue to invest in new short term d evelopments to cover gaps in capability which could already be tackled with fine-tuning of existing tec hnology, as opposed to the 'leapfrog' developments that will promote them to become market leaders. • Category 2.3 Companies; are completely made up of c at.2 companies who have learned from their mistakes and have made the transition toward forwar d-planning of AMT development with a view to leading the market and ensuring sustainability of t he business. The factors affecting the success of these companies, essentially the 'to-do' list for the cat .1.1 companies are detailed in the following discus sion section. In each of these sub-categories there is a profile of company that fits comfortably with each of the t hree possible trends in category change; improved attitu de to AMT and associated improvement in performance (promotion from cat.1 to cat 2.1 and from cat. 2.2 to cat. 2.3), maintenance of a stable operation (ca t. 1, cat. 2.2 and cat. 3) and decline in business profile due to a lack of activity in implementation (cat. 2.2 to cat 2.1). It was seen that there were no clear ‘relegation’ comp anies moving from cat.3 to cat.2 or cat.2 to cat.1) , which indicates that once the practices included in the i mplementation model are truly incorporated into the operations management of a company then they are ro bust enough to be self-sustaining, which is the key to continuous improvement, or at least continuous acti vity. 4.0 Discussion The survey highlighted that in general, SMEs still did not fully appreciate that the effective impleme ntation of AMT could assist the company in improving business performance and customer satisfaction. Furthermore, of the companies that had implemented AMT, all stated that the implementation phase was the most problema tic area of the process. This can have two effects; of deterring companies from further investment, as the y do not perceive that value for money has been realized, or alternatively of necessitating additional resource to meet the original expectation, preventing optimization o r specialization which would provide greater perfor mance. These difficulties must be overcome to prevent comp anies falling into the 2.1 and 2.3 categories. Howe ver, it is now seen that to move between the basic three ca tegories, a significant change process has to be un dertaken beyond that of using applied tools and techniques. While a difficult implementation can lead to a lack of top- end performance, hence progression to a higher cate gory, it is a symptom of the more cultural issues i n the business. For example, precision engineering companies that h ave been founded on traditional machine-shop (often family owned) business values are happy to move tow ard higher technology as followers, minimising risk , but less likely to develop the unique capabilities that would make them a cat.3 company. While a precision engineering company that has been more recently est ablished on the strength of a particular skill set or capability will already appreciate the value of thi s, and will be both more likely to dedicate resourc e to developing a faster, more flexible and cost effecti ve technology and also more able to do so through application of a simpler management framework to ma intain this advantage 387 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES Table I: Summary of Results; SME Categorisation Business Type Technology Strategy Business Strategy Cat.1 Low tech’ SMEs with static progress (22% of original survey result, 19% of review survey) Single-process based business using conventional production techniques which are unchanging. Has the single advantage of being flexible in small batch sizes. Very little interest in technology development and equipment is initially replaced with improved conventional equipment. Lack of management vision. SMEs in craft activities with slow technical progress, often with highly skilled labour, reducing profits. AMT is viewed as a financial burden, unable to retain specialist staff or customers. Cat. 2.1 Cat.2 companies falling behind / cat.1 companies catching up (12%) Sales are being lost to competing economies. These companies have usually relied on particular product or process strength with little succession planning. Understands that a core capability needs to be quickly re-established but inexperience in technology management prevents meeting the required pace of change. Lack of clear understanding of what is required to maintain market share. There is often a discrepancy between business strategy and customer requirements which must be addressed. Cat.2.2 Growing SMEs profiting from basic technology development (65% of original survey result, 47% of review survey) Higher-performance business which purchases technology without really managing or understanding it. Technology can be low cost but is underutilised due to limited knowledge and lack of workforce development. Believes technology is essential to development of company but do not tend towards forward planning and an ability to match the technology to company needs. Tend to be taken by 'technology fashion'. Results in technology being under- utilised. Business with a relatively traditional mode of production. See that technology implementation as the key to success but no real idea of how to implement. Financial risk is seen as an issue but not one that will prevent AMT implementation. Cat 2.3 Cat. 2 companies who have experienced success and are developing unique AMT. (10%) Cat. 2 companies that are achieving increased performance, but consistency is required before competitive advantage is sustained. These companies have developed an AMT implementation strategy in light of success, or just increased experience. Increased performance of the business is capitalised in terms of continued development of product or process capability. Management attitudes are changing toward ‘front-end’ activities, appreciating the full potential benefits of AMT in terms of sustainability. Business performance is improving as customer base is extended, achieved by change from a reactionary attitude (to particular customer demands) towards a proactive development plan designed to extend scope of the company. Cat.3. Innovative SMEs with high tech’ profile and culture (13% of original survey result, 12% of review survey) Business marked by its strong capacity for technological innovation and enterprise, offering new products or innovative processes, creating new markets and customers. On the fringe of losing SME status. Continuous technological innovation strategy. Forward thinking with MD's who will devote limited resources to implementing technology in house hence maintaining knowledge base. Independent innovative SMEs focusing on process in order to manufacture the product with high quality. Tend to be able to keep workforce who develop skills that keep company ahead of competitors. Culture of continuous performance. 388 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES 4.1 Factors Affecting Technology Implementation The ancillary benefit to interviewing the companies after the initial categorisation, and proposing th e subsequent technology implementation model, is that companies can more accurately assess themselves in terms of performance and attitude. This makes the p rocess quicker the second time round, but also prov ides a deeper insight into the areas for improvement and w hat factors are preventing promotion to a higher ca tegory by increasing the effectiveness of technology devel opment. As mentioned, the implementation phase was stated t o be the most difficult, but almost all companies a greed that this could be overcome in part by a tighter fr ont-end specification and a more detailed return on investment (ROI) calculation which should drill-dow n into expected QCD improvements to identify potent ial problem areas. Burcher [5 ] also proposes that this will prevent missed opportunity / not realising ca pabilities of technologies, as closer interrogation of supplie rs and the existing knowledge base will expose aspe cts which may not be commonly utilized by the competition, or better still areas of weakness that could be overc ome to provide the critical competitive edge. Commercially this is also important to align market requirement s with new technology capabilities. Small [6 ] advocates a n approach to identify selective factors in order t o create a selective technology portfolio, In addition, the need for a disciplined approach to technology planning through close adherence to a d etailed project plan prevents deviation from achieving the core objective, a view supported by Salimi [7], and also precludes unrealistic expectation about the impleme ntation time-scale and the associated cost of TI. I n combination, the specification and planning assesse s all possible risks of failure and deals with issu es before they happen (Smith and Reinertsen [8]). To achieve this, on an operational front, there are several common features which have led to sub-opti misation of new technology. Failure to integrate existing co nventional and advanced technologies can lead to unbalanced process chains and potentially little im provement in overall lead-time, which is supported by Beatty [9] and Singh et al [14]. Support for the d evelopment of a new technology can leave it isolate d through inadequate organizational planning, again compromis ing this integration of this key technology to the core of the business where it should quickly become a uniqu e selling point. This also calls for capacity build ing of the human resource so that the technology can be fully developed and expanded upon into the future (Castri llón & Cantorna [10]). Therefore, it is suggested that as a development of the Technology Implementation Model (TIM) proposed , there could be a more flexible Technology Implement ation Strategy (TIS) which diverges across a range of approaches depending on existing constraints/parame ters/characteristics. The TIS will continue to prov ide a framework for the tools and techniques at each stag e, but in an optimised progression, weighting activ ity by requirement. Cooper [11] and Hutchison & Das [15] p ropose that this must also incorporate detailed pee r review, including top level management, at suitable decision points to provide commitment to expected results. 5.0 Conclusions It is important that implementation of AMT is not v iewed to be a panacea for all production related manufacturing problems, but it is believed that a c ompanies specific attitude toward AMT reflects its strategy 389 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C HARACTERISATION OF SME MIGRATION TOWARD ADVANCED TE CHNO LO GIES for achieving long term economic sustainability. To this end, it is seen that there are three main cat egories of SME: low tech’ companies with low ambition, growing companies which are undertaking AMT development but not optimizing potential, and also high tech’ c ompanies that are high achieving and largely stand- alone. In addition to this, there are two additional sub-c ategories of category 2 companies: those failing to keep pace of growth and development, and those who are changi ng attitude to development of new technology and developing unique skills, but still not realizing c ompany performance increases which will take them t o the next category. Future work identified includes creation of an anal ysis tool for companies to be able to self-assess, against the characteristics identified through the surveys, in order to place themselves into this improved perfor mance matrix. Subsequently, a decision tree structure can be created to identify a critical route for progre ssion to the next category, incorporating which TI tools and tec hniques are required to achieve this. Which will ul timate result in fine-tuning and refinement of the TI mode l proposed by Thomas et al [4]. References [1] Ettlie, J.E, [1990], "Intrafirm mobility and manufa cturing modernization", Journal of Engineering and Technology Management, 6, pp 281-302. [2] Quarashi Z.N, Polkinghorne M.N, Bennett J.P, [1998] , “The Economic Impact of the Teaching Company Sche me on the Local Manufacturing Base” Internal Report, Plymouth Teaching Company Centre . [3] Thomas A.J [2002] “The College Business Partnershi p Scheme and its Impact on the Technical Developmen t of SMEs in South Wales’, Proceedings 18th NCMR , Leeds Metropolitan University [4] A.J. Thomas, R. Barton, E.G. John, [2008], “Advanc ed manufacturing technology implementation”, International Journal of Productivity and Performance Management, Vol 57, Issue 2, Pp 156 – 176 [5] Burcher, R.G., Lee, G.L., [2000], "Competitiveness strategies and AMT investment decisions", Integrated Manufacturing Systems, Vol. 11 No.5, pp.340-7. [6] Small, M. H., [2007], “Planning, justifying and ins talling advanced manufacturing technology: a manage rial framework”, journal of manufacturing technology management , vol 18, issue 5 , pp513 – 537 [7] Salimi .B. [1996] "New product development in small high tech firms" M.Phil Thesis, University of Calg ary. [8] Smith, P.G. and Reinertsen, D.G., [1991], “Developi ng Products in Half the Time”, Van Norstrand Reinho ld. [9] Beatty, C., [1990], "Implementing advanced manufact uring technology", Business Quarterly , Vol. 55 No.2, pp.46-50. [10] Castrillón, I.D, Cantorna A.S, [2005], “The effect of the implementation of advanced manufacturing tec hnologies on training in the manufacturing sector”, Journal of European Industrial Training, 29, 4 [11] Cooper, R.G., [1993] Winning at New Products: Accel erating the Process from Idea to Launch, Addison We sley. [12] Beaumont, N., Schroder, R., Sohal, A., [2002], “Do Foreign Owned Firms Manage Advanced Manufacturing Technology Better?”, journal of operations and production management Vol 22, No7, pp759 – 771. [13] Zhang, O., Mark A. Vonderembse, Mei Cao, [2006], “A chieving flexible manufacturing competence: The rol es of advanced manufacturing technology and operations im provement practices”, International Journal of Operations & Production Management; Vol.26, Issue: 6, pp580-599. [14] Singh, R.K., Suresh K. Garg, Deshmukh, S.G., Kumar, M., [2007], “Modelling of critical success factors for implementation of AMTs”, Journal of Modelling in Management, Vol 2, Issue 3, pp 232-250. [15] Hutchison, J., Sidhartha R. Das, [2007], “Examining a firm's decisions with a contingency framework fo r manufacturing flexibility” Journal of Operations & Production Management , Vol 27, Issue 2, pp159 – 180. 390 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED ENGINEERING: LEAN IMPLEMENTATION IN AN SME B.M. Wahid 1 , C. Ding 1 , J.O. Ajaefobi 2 , K. Agyapong-Kodua 2 , T. Masood 2 , and R.H. Weston 1,2 1. Centre for Excellence in Customised Assembly, Lough borough University, Loughborough, UK 2. MSI Research Institute, Loughborough University, Lo ughborough, Leics., UK Abstract Popular ‘methods-based’ approaches to engineering enterprises include: BPR, Continuous Improvement, Kaizen, TQM, JIT, Lean and Agile Manufactu ring. Generally the industrial application of such methods-based ap proaches leads to long lead-times, high costs, and poorly justified engineering projec ts that do not prepare the organization for future change. These outcomes are to be expe cted because (1) invariably Manufacturing Enterprises (MEs) constitute very com plex and dynamic systems that naturally require complex design and change proces ses and (2) current methods-based approaches to organizational design and change are not analytically well founded. Therefore the authors argue that a framework and modelling toolse t are required to facilitate ongoing and integrated application of methods-based engi neering approaches, providing underlying modelling structures and concep ts to ‘systemize’ and ‘quantify’ key aspects of organization design and change. Unles s suitable decomposition, quantitative and qualitative modelling principles are used to underpin an approach such as a Lean Manufacturing, deficiencies will re main. Often, MEs adopt the “we need be lean” mindset without holistic under standings of causal and temporal impacts of such philosophies on ME processes, resour ce systems and current and possible future workflows. Enterprise Modelling (EM) p artially addresses the aforementioned problems and can support the development of rob ust understandings about current enterprise processes and potential capabiliti es of systems. However in general, current EM techniques are geared best to capturing and organizing relatively enduring knowledge and data about any given organization but are them selves deficient in respect to replicating and predicting dynamic system be haviors. 391 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME This paper presents a model driven approach to organization desi gn and change in support of methods-based engineering, applying Lean Manufacturin g principles, with a UK based bearing manufacturer. EM and various derivative Simulat ion Modelling (SM) views were generated to display system behaviors under changing scenar ios. Keywords: Lean, Takt Time, Manufacturing Enterprise , Enterprise Modelling, Simulation Modelling 1.0 Introduction In recent decades technological innovation has indu ced very significant change on the way that Manufac turing Enterprises (MEs) operate and compete. To support M Es in coping with changing business and production requirements, ‘method-based’ approaches to organiza tion design and change including: BPR, CPI, Kaizen, TQM, JIT, PPM, Lean Manufacturing, Agile Manufactur ing, etc, have been proposed and applied in many MEs. However, none of these management philosophies and production approaches could be said to be a panacea for success. With increasing pressure on ME s to remain responsive to changing market demands, whilst maintaining operational efficiency, arises t he need for robust process understandings to enable effective implementation of improvement philosophies. In general MEs are very complex entities: designed, managed and changed by people, by deploying people and technological resources in systematic, timely a nd innovative ways that generate competitive behavi ors. Prerequisites to respond such change and deal with complexity are firm understandings of; the processe s that are the focus of improvement and effects of change decisions. However, implementing change and dealing with associated effects can prove difficult when su ch an understanding is not intrinsic i.e. domains l acking structure and documentation. This paper presents a model driven approach to orga nizational design and change. An Enterprise Modelli ng and Integration (EM&I) approach was deployed to mod el aspects of a Small to Medium sized Enterprise (SME) collaborating with the Manufacturing Systems Integration (MSI) research institute. The resultant models, especially in their simulation views provid e basic frameworks for reasoning about enterprise s ystems behaviors under changing scenarios. The case instan ce reported is an SME that makes composite bearings to order, namely ComBear composite bearings. The focus objectives of the ComBear improvement project include (1) to create enterprise models (static & d ynamic) to enhance the understanding of the enterpr ise processes and systems, and (2) to bring to bear on the process and systems models created elements of Lean Manufacturing concepts and tools with a view to ach ieving improved performance in the critical areas o f lead time, process efficiency, resource utilization and consumption, and increased value generation. 2.0 Lean Manufacturing Philosophy Background The Lean Manufacturing philosophy originated from T oyota Production systems in Japan and was pioneered by Taiichi Ohno (1912 -1990). The prime purpose of Lean Manufacturing is to eliminate manufacturing 392 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME wastes (muda). Tapping [1] describes manufacturing wastes in terms of the so called seven deadly manufacturing wastes, namely: overproduction, waiti ng, transport, over processing, inventory, motion, and defects. Wastes in manufacturing are activities whi ch absorb resources but create no value in return a nd they includes: mistakes which require rectification, pro duction of items no one wants, unnecessary inventor ies, processing steps which are not needed, movement of people and transport of goods without purpose, peop le waiting because the upstream activity has not promp tly delivered, goods and services that do not satis fy customer’s requirements [2]. Lean Manufacturing is a systematic approach to identifying and eliminatin g wastes through continuous improvement, flowing the product at the pull of the customer in pursuit of perfection [3]. Hence the goal of Lean Manufacturin g is to eliminate wastes by: • Producing what the customer needs • When required by the customer • In the exact quantity needed • Using resources only when needed Womack & Jones [2] suggest five principles steps to wards achieving Lean Manufacturing benefits namely: 1. Precisely specify value by specific product 2. Identify the value stream for each product 3. Make value flow without interruption 4. Let the customer pull value from the producer 5. Pursue perfection The application and testing of some of these Lean M anufacturing principles in ComBear will be discusse d later in this paper. 3.0 Enterprise Modelling and Simulation Enterprise Modeling (EM) approaches and supporting tools provide a structured view and grounding for change decisions, enabling the systematic hierarchi cal decomposition of an ME’s processes, allowing contextual problem definition and specification. CI MOSA (Computer Integrated Manufacturing Open System s Architecture) was developed by the AMICE consortium during a series of ESPRIT projects [4]. CIMOSA aims to help companies manage change and integrate their facilities and operations. It has been emphas ized by [5], [6], and is considered by many authors to be t he most comprehensive of current public domain EM approaches [7]-[9]. CIMOSA introduced a process-bas ed approach to integrated EM, ignoring organization al boundaries, as opposed to various function or activ ity-based approaches, described in terms of their; function, information, resource and organizational aspects, a nd designed according to a structured engineering a pproach that can then be plugged into a consistent, modular and evolutionary architecture for operational use [7]. It presents a model-based approach to design, operatio nalize and manage an enterprise. The authors and th eir colleagues have been using CIMOSA in numerous resea rch and industrial projects. Sets of CIMOSA conformant models are generated during projects and presented to industrial partners for verification. This serves three purposes; (1) to enable enterprises to understand, model, analyze their processes and ope rations, (2) to provide model developers with an accurate be nchmark from which improvements can be derived, (3) to provide the management team with information to mak e effective decisions in response to change. For th e 393 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME constraints of this paper, the authors will not go in to great details and illustrations of CIMOSA mod els, but briefly introduce main modeling procedures followed and the model types produced. 3.1 Establishing a Focus Modeling Domain, the Context Diagram After the broad aims and general problems for a cas e company have been identified, the modeler must de fine a scope within which the existing modeling tools will be deployed. EM in this case uses decomposition principles to handle model complexity, this constra ins the modeler to model in abstraction and avoids modeling of the infinite complexity inherent to rea l systems. The modeling priority and emphasis is es tablished through a model depicting the global objective, whi ch is reasonably simple in structure and content. T he primary focus is central and surrounded by the most relevant domains involved in objective realization . This is termed the Context Diagram, and is the first type g enerated. Additionally, it is necessary to have spe cified an area of concern when drilling down and to demarcate immediately unconcerned domain(s) to provide succi nct models, representative of entry point and problem c oncerns. Marked domains are then treated as ‘black box’ thus not further detailed during model development nor when creating modeling scenarios for case compa nies. 3.2 Problem Domain Decomposition, Interaction Evaluation and Structure Building The next modeling step deploys a mechanism for deco mposition to break down the primary focus domain, CIMOSA modeling specifies a diagram to show relatio nship networks between those involved domains. The relations are interpreted in terms of inflow and ou tflow of; information, human resources, material an d finance. Thus when a particular domain is subjected to inter nal change, one can deduce the inter-domain effects on connected flows and responses. This outlines the pu rpose for the Interaction Diagram. A subsequent typ e of diagram, termed the Structure Diagram, is used to d ecompose and build structure. This can also be used on each of the associated domains which have been iden tified to model in CIMOSA in the Context Diagram. T he Structure Diagram takes each domain as a focus for further examination and decomposition in to a hierarchically structured set of processes. Both ty pes of diagrams can be built on a subsequent level i.e. it is possible for a particularly complex domain to have several Interaction and Structure Diagrams for the purposes of providing a sufficient level of detail as required by the modeler. 3.3 Sequential Precedence of Process Operations, the Ultimate Respond ents to Change and Where Decisions Need to be Made Procedural steps thus far have served to decompose and structure domain contents. Now a more detailed level is reached, here actual sequences of process and co nstituent operations are assessed. Complete end-to- end process networks, comprising activities with associ ated information and resource inputs and outputs ar e represented using the fourth CIMOSA diagram type, t he Activity Diagram. A numbering convention is followed to identify activities listed with their d ependencies and routings. Also, an approximate dura tion is given through means of a timeline indicating when e ach step of operation will initialize and how long they operate. 394 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME From a model developer and theoretical perspective, the concepts, methodology and technology used in CIMOSA modeling and diagrams, can usefully decompos e complex process networks into their component process segments. CIMOSA also serves to provide a m eans of documenting and visualizing associated flow s of; activities, material, information, controls and so forth. Such model diagrams can support an ME’s decision makers (i.e. company management teams and direct as sociated operators) who require increased informati on support from models. To achieve this, the models ne ed to enable; (1) appropriate presentation format a nd structure to be readily understood by users (i.e. t he decision makers) (2) efficient and equivalent in formation which can be quickly obtained from models, (3) quic k, responsive and efficient development, if origina l model data is available, as per the end users’ reference requirement, (4) a model format and building proced ure that is flexible to various model iterations, transfer, and re-use. 3.4 Systemizing Methods-Based Engineering Using CIMOSA The authors propose that the use of EM, in particul ar CIMOSA and conformant approaches developed at th e MSI Research Institute [10], in conjunction with me thods-based engineering approaches will fulfill a t wo-fold requirement, these being: (1) the provision of a st ructured route to implementation of methods-based engineering philosophies, and (2) EM based improvem ents to be informed through well defined philosophi es. Such static models can then inform the analysis of time dependant simulation models, allowing for the quantification of improvements with respect to: thr oughput, time in the system, resource efficiency an d utilization. 4.0 Case Study 4.1 Company Background ComBear is a rapidly growing SME based in the UK wi th global customers and stakeholders. Recently, ComBear completed a major enterprise engineering pr oject when it created a second production facility, similar to its UK operational base, which is now lo cated in South Korea. Further production facilities are being developed in other parts of the world including the US with a view to increasing market share. At both of its current manufacturing sites, ComBear manufactures a range of advanced composite products suitable for agricultural, marine, mechanical, pharmaceutical an d food processing applications. In general terms, composite products are manufactured from reinforced plastic laminates composed of synthetic fabrics impregnated with resins and lubricant fillers. Fina l products are delivered to customers in a variety of form factors including, but not limited to: structural b earings, washers, wear rings, wear pads, wear strip s, rollers, and bushes. 4.2 Problem Definition and Case Study Objectives ComBear’s growth in the market of composite bearing manufacture has put the company under increasing pressure to produce products in larger volumes, sho rter times, and at reduced costs. Approaching such 395 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME challenges without considering organizational desig n and change has lead ComBear to compromise metrics fundamental to the customer i.e. on time delivery, product quality, and product costs. The focus objectives of the case study were to: (1) create enterprise models to enhance the understand ing of the enterprise processes and systems, and (2) bring to bear on the process and systems models created elements of Lean Manufacturing concepts and tools w ith a view to achieving improved, quantifiable, performance in the critical areas of; lead time, pr ocess efficiency, resource utilization and consumpt ion, and increased value generation. Whilst Lean activities with ComBear are ongoing, to remain concise, the establishment of a ‘pull signal’ and reducing mater ials wastage will be discussed in this paper with p articular emphasis on the raw materials production process. 4.3 Enterprise Modelling and Simulation at ComBear CIMOSA modelling constructs and representational fo rmalism support decomposition of complex systems, i n this case ComBear’s processes, into sub systems tha t can be analyzed independently and later recompose d into a collective whole. The key processes in ComBear we re identified and encoded to enable the realization of enterprise models. These were then validated by the ir management and production teams as being representative of the enterprise processes and asso ciated resources. Exemplar diagrams follow, figure .1 shows a Context Diagram which was described in section 3. 1. Figure .2 depicts an Interaction Diagram, descri bed in section 3.2, of ComBear’s ‘plan and control product ion’ process, showing the flow of information and resources between associated processes, and also fl ow and control logic. Domain Processes (DPs) extern al to the domain under study are depicted as ‘black box’ representations. Control Production (EA5.1.1.1) rec eives the job card from the Technical process (DP4) and p roduces appropriate production schedules according to scheduling rules deployed at ComBear. The job card provides both technical product information, i.e. engineering drawings, as well as order details i.e. due date, quantity. The job card informs BP5.2.1 t hrough to BP5.2.5 along with further inputs of raw materials and human resources. BP5.2.1 produces round and fla t raw materials that are then used by BP5.2.2 and BP5.2.3 . Products are packaged and dispatched, at which po int DP4 receives notification and feeds back dispatch d etails to BP5.2.5. A number of control nodes indica te where EA5.1.1.1 will consult the production schedul e to prioritize jobs accordingly. Having developed similar static models of ComBear p rocesses, at the further detailed ‘activity’ level, the next step in the project sought to operationalize the mo dels by applying selected Lean Manufacturing measur es to the processes in a simulation environment in order to observe current ‘as-is’ process behaviors and performance of the resource systems. The initial si mulation environment was realized using Simul8® and further investigation is being carried out by anoth er software, Lean Modeller®, to establish values (t his is mentioned for completeness but is not within the sc ope of this paper). Through analysis of static and simulation models in conjunction with discussions with ComBear’s management and production teams, it was decided tha t the focus of improvements should be in the produc tion of raw materials that supply later processing shops . This was supported by the simulation model in fig ure .3 which indicates a very low percentage utilization o f resources in the histogram as well as a large dev iation in 396 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME the time round narrow raw materials spend in the ra w materials processing shop. Additionally, this par t of the manufacturing process was manually intense hence of fered the most scope for improvement. Fig. 1. Establishing a focus domain using the Context Diagr am Fig. 2. Process and resource flows within a domain, the Int eraction Diagram 397 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME Fig. 3. Initial simulation model The company’s production systems are a mix of make- to-order and engineer-to-order, despite the order b ook consisting of repeat orders. Whilst these are repre sentative of pull systems, the pull signal itself i .e. the job card, is issued to the raw materials processing sho p i.e. the start of production. This indicates that whilst production is based on firm orders, within the fact ory products are produced in a way more representat ive of a push system. ComBear had highlighted that there wer e increasing problems with late deliveries and also product quality, with products leaving site that we re then found to be of insufficient quality by the customer. Having created static CIMOSA models, to depict proc esses and provide common understandings, and simulation models to introduce time dependencies an d thus quantify current practice (in terms of outpu t, time in the system, and resource efficiency and utilizat ion) indicating where to focus lean improvements, t he next step was to deploy methods-based engineering to tar get improvements. Whilst improvements were targeted and made across flat, strip, and round raw material s, for the purposes of this paper only those made w ith respect to round narrow will be discussed. 4.4 Lean Manufacturing Principles Deployed Through use of analytical methods, guided by the mo del structures developed using EM principles and diagramming techniques, improvements were targeted to provide; effective process resourcing, product a nd tooling rationalization, reduced materials wastage, and workplace organization. 398 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME 4.4.1 Implementing Pull and Product Rationalization In order to implement a pull signal within the orga nization, it is necessary to establish takt times f or products i.e. demand driven requirements of ComBear. This ca n only be achieved if sound product classifications exist, these were established through the use of activity diagrams. It was noted that of the product classifi cations used by ComBear, processing characteristics provide d an alternative and generalized classification of round, flat, and strip products. These general form factor s exhibited distinct processing routes in the raw m aterials processing shop. This rationalization of products a llowed complexity to be minimized when establishing takt times. Analysis of historical order information allowed ta kt times to be developed and broken down in to cons tituent takt times for each of the newly defined classifica tions by key processes i.e. produce raw material, p roduce flat products, produce round products, as defined in pre viously created enterprise models. With these times understood and visible to the production team, deci sions could be exercised to enable adequate resourc ing of processes to support production of what the custome r needs, when required, in the quantity needed. 4.4.2 Materials Wastage and Tooling Rationalization The processes deployed in the production of raw mat erials require that the tooling, thus working dimen sions, be within 2mm of finished dimensions to minimize th e amount of subsequent machining effort required an d material wasted. However this must be balanced with cost, space, and demand constraints. Currently, to oling is highly product specific and whilst this reduces mat erial wasted, it does induce waste in terms of tool ing stock. Other than dimensional constraints, there is no sou nd justification for the tooling sizes stocked in t he raw materials shop. A distinction must therefore be mad e between the frequency of tooling use and a justif ication must be made for holding particular sizes and quant ities. With the aforementioned points in mind, step wise improvements have been delivered through the catego rization of products in to; those that occur on a continuous basis and are core products (runners), t hose that occur regularly (repeaters), and those th at are manufactured to specific requirements (strangers). The associated requirements for tooling sizes and q uantities were extrapolated and as a result of these grounded considerations for stocking of tooling, significan t scope for rationalization has been indicated. The authors wou ld like to state that this is still an area of cont inued work, with similar improvements targeted for the other pr oduct types produced at ComBear along with effects on scheduling of production within the organization. 4.5 Quantifying Improvements Through Simulation The quantified effects of changes to the production of round narrow raw materials, as a result of impl ementing Lean compliant improvements and resourcing strategi es, can be seen in the simulation model in figure . 4. Comparing this with figure .3 shows that the worker utilization has increased as has the overall produ ctivity of the system, this is now regularly achieved with 71% within the limit. The histogram indicates that a h igher resource utilization percentage has been achieved, this is attributable to the increased availability of tooling i.e. operators are not waiting. The implementation of a pull signal (takt time) means that operators are mo re clear 399 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME and informed as to the proceeding job, this is rein forced by the standard deviation which indicates th at the production rate is more predictable. Fig. 4. Simulation model to quantifying improvements 5.0 Conclusions The combined use of Enterprise Modelling (EM), Simu lation Modelling (SM), and methods-based engineerin g allowed the authors of this paper to conduct a quan tified, systematic and targeted Lean implementation based on; improved and verified process understandings, e stablishment of a pull signal, resource (human, too ling, raw materials) and product (establishing groupings) rationalization. Whilst it is acknowledged that the research in the areas of Lean Manufacturing is well documented, it is however lacking in a structured and integrated mean s to realizing improvements. The authors can see th e real benefits in the development of systematic support f or methods-based engineering that guides users with generalized routes to implementation. The case stud y conducted has shown that the key to successful Le an and similar improvements is largely dependant on unders tanding of current practice and the causal impacts of change. Through the use of static and simulation mo delling approaches developed at the MSI research institute, many of these impacts can be enacted and thus mitigated. 400 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 ENTERPRISE MODELLING IN SUPP O RT OF METHODS BASED EN GINEERING: LEAN IMPLEMENTATION IN AN SME Improvements and methods-based engineering conducte d were done so in abstraction from the real system, using EM and SM approaches. The project work with C omBear is ongoing and the next steps are to work on product classes to inform different stocking strate gies based on the product type i.e. runner, repeate r, or stranger. For example it may well be necessary to m ake-to-stock raw materials for less frequent produc ts. Additionally, the interfaces between the raw materi als shop and subsequent processes needs to be explo red to further inform takt times thus fully implement a pu ll signal. Whilst classifications break complexity, there is a degree of variation within them, hence further vari ables will be used in order to further inform produ ct sub- types. Further suggested improvements are to be implemente d and areas of investigation researched, the result s of these efforts will be reported on in further confer ence and journal publications. Acknowledgements ComBear (pseudonym) for their continued support of research conducted at the MSI research institute, Loughborough University. Prof K. Case and Dr. R. Harrison, Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, for their support in CECA and MSI associated project work. References [1] Tapping, D., Luyster, T., Shuker, T., Value Stream Management, Eight steps to planning, mapping, and s ustaining Lean Improvements, Productivity Press, a Division o f Productivity Organisation, Ltd, NY, USA. 2002. [2] Womack, J.P. and Jones, D.T., Lean Thinking; Banish Waste and Create Wealth in your Corporation, Free Press, A division of Simon & Schuster, Inc, NY, USA. 2003. [3] Kilpatrick, J, Lean Principles, Utah Manufacturing Extension Partnership, 2003. [online]. Available: http://www.mep.org/textfiles/LeanPrinciples.pdf [2008, February 1]. [4] ESPRIT Consortium AMICE (Eds.), “CIMOSA—Open System Architecture for CIM”, 2nd revised and extended edition, Research Report, ESPRIT Project 688/5288, Springer, 1993. [5] Kotsiopoulos, I. L., “Railway operating procedures: regulating a safety-critical enterprise”, Computers in Industry , Vol.40, Iss.2-3, pp.221-230, 1999. [6] Kosanke, K. and Zelm, M., “CIMOSA modelling process es”, Computers in Industry , Vol.40, Iss.2-3, pp.141-153, 1999. [7] Vernadat, F.B., “Enterprise Modeling and Integratio n: Principles and Applications”, Chapman & Hall, Lo ndon, 1996. [8] Monfared, R. P. and Weston, R. H., “A method to dev elop semi-generic information models of change-capa ble cell control systems”, Computers in Industry , Vol.41, Iss.3, pp.279-294, 2000. [9] Reithofer, W. and Naeger, G. “Bottom-up planning ap proaches in enterprise modeling – the need and the state of the art”, Computers in Industry , Vol.3|3, Iss.2-3, pp.223-235, 1997. [10] Monfared, R.P., A Component Based Approach to Desig n and Construction of Change Capable Manufacturing Cell Control Systems. PhD Thesis, Loughborough Universit y, UK, 2000. 401 402 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITICAL MODELS AND IMPLEMENTATION S. Tridech and K. Cheng Advanced Manufacturing and Enterprise engineering ( AMEE) Department, School of Engineering and Design, Brunel University, UK Abstract Today, the rising of carbon dioxide (CO 2 ) emissions is becoming the crucial factor for global warming especially in industrial sectors. Therefore, the research to reduce carbon intensity and enhance resources utilization in manufac turing industry is starting to be a timely topic. Low carbon manufacturing (LCM) can be referred to the manufacturing process that produces low carbon emissions inte nsity and uses energy and resources efficiently and effectively during the process as well. In this paper, the concepts of LCM are discussed and the LCM associated theoretical models, characterization and implementation perspective explore d. The paper is structured in four parts. Firstly, the conception of low carb on manufacturing is critically reviewed then the characterization of low carbon manufacturing is discussed and formulated. Third part, the theoretical models are develope d with initial models by using the theory from supply chain modeling and linear program ming solutions (LP). The models show the relationship of resource utilizat ions and related variables for LCM in two levels: shop-floor and extended supply chain. F inally, the pilot implementations of LCM are discussed with two approaches: d esktop or micro machines and devolved manufacturing. The paper is concluded wi th further discussions on the potential and application of LCM for manufacturing indu stry. Keywords: Low carbon manufacturing (LCM); Micro manufacturing system; Devolved manufacturing; Energy efficiency; CO 2 emissions 403 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION 1.0 Introduction Currently, global warming is extensively discussed as one of the most important global issues because of the rising of the amount of carbon and carbon dioxide c ontents emitted from industrial sectors. Many secto rs have been trying to develop the solutions to solve and p revent this problem e.g. carbon emission analysis, software based prediction on economic factors and physical i mplementation using micro manufacturing system and microfactory. However, most of previous solutions d oes not mention about the configuration of the proc edures to reduce carbon emission. Therefore, the effort to reduce carbon intensity and enhance resource utili zations is stated as a timely topic. The purpose of this paper is to develop an industrial feasible approach to implementing LCM including its theoretical models, methods and application perspectives. 2.0 Literature Review 2.1 Carbon Emissions Analysis In the past decade, many countries have been consci ous to develop the procedures for reducing carbon emissions. Fan et al. [1] have presented the model for prediction of carbon dioxide (CO 2 ) emissions based on the input of population, economy and urbanization. In 1996, Golove and Schipper [2] introduced the a nalysis of the tendency of energy consumption which can cau se CO 2 emissions from manufacturing sectors based on the input of the gross domestic product (GDP) chang ed to economic output and process intensity. Althou gh, these methods have been developed to deal with the global warming problem from carbon contents, the procedures to analyse is still focusing on the wide range and depending more on economic factors such as GDP. The procedures for reducing CO 2 emissions in manufacturing systems and the associa ted manufacturing processes have not been introduced yet. 2.2 Operational Model In the area of production research, most of the res earch focuses on the objective such as cost minimiz ation, quality assurance and the level of customer satisfa ction as the objectives of the process optimization according to Gugor and Gupta (1999)[3]. Carbon emissions and energy efficiency have never been a critical factor in operation optimization. However, Mouzon et al. [4] have developed the operational model by using the t heory of multi-objective mathematical programming in orde r to minimize energy consumption from equipments in manufacturing system. In the operational model, the constraints are focusing on completion time and to tal power per unit time. Even though, the production re search for reducing total energy consumption has be en introduced at this time, the operational model for reducing carbon contents from manufacturing process es need to be further developed. 2.3 Desktop and Micro Machine 404 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION The concepts of micro-factory and desktop machines for micro manufacturing purpose have been explored in the wide range. For the definition and concepts of the micro-factory and desktop machine, Yuichi [5] e xplain it as small scale manufacturing systems which can perf orm with higher throughput while resource utilizati on and energy consumption rate can be reduced simultaneous ly. In addition, Mishima [6] suggests that the conc ept of micro-factory and desktop machines should also conc entrate on low heat generation and less energy consumptions of the systems. It is concluded that t he innovation of desk-top and micro machine can be applied to the LCM by reducing the unnecessary carbon conte nts from manufacturing systems. 2.4 The Novell Approach: Devolved Manufacturing The high proportion of carbon dioxide emissions not only comes from manufacturing systems and processe s but also from the transportation while working on e xtended supply chains manufacturing. Bateman and Ch eng [7] have introduced in a novel approach called Devo lved Manufacturing (DM) which integrates main three elements together for future manufacturing systems: web based (e-manufacturing), mass customization (M C) and rapid manufacturing. The aim of this approach i s to provide “factory-less” which customers can rec eive their products at the nearest location. In other wo rds, this approach can be applied to minimize the transportation in associated with manufacturing sys tems set up. It is concluded that Devolved Manufact uring can be considered as an approach for reducing carbo n contents emissions particularly for LCM in supply chain based manufacturing systems. 3.0 Characterization of Low Carbon Manufacturing Low carbon manufacturing (LCM) can be described as the process that emits low carbon dioxide (CO 2 ) intensity from the system sources and during the ma nufacturing process. In addition, the term of LCM c an be broadly not only for environmental aspect but also the energy conservation and effective production be cause the process uses energy excess available capacity/c onstraint (low energy efficiency) simultaneously wi thout optimal algorithm to run process or system can lead to the high volume of carbon dioxide intensity to atmosphere (Figure1). Therefore, the main character ization of LCM can be categorized into specific fiv e terms as follows: Fig. 1. Characterization of Low Carbon Manufacturin g 405 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION 1) Low carbon dioxide from source: currently, almost all equipment and machines in modern industry use electricity as a main energy to operate if machines or equipment can be adjusted or improved to use le ss energy, the carbon dioxide intensity from the machi nes and equipment sources will be reduced. 2) Energy efficiency: energy efficiency can be explai ned as a percentage of output of energy from proces s (in watt or joules) divided by the input of energy to the process [8]. Hence, this parameter in LCM concept should be higher than conventional industri al processes. 3) Waste minimization: This term can be meant as how waste can be dislodged or minimized according to the reference [9]. If the third criteria above are categorized into carbon dioxide emissions due to machines and equipment, it is emission from imperfe ct operation because waste can occur in the process . For example, many wastes can appear in the turbulen t manufacturing process: idle time, waiting time an d queuing time etc. Therefore, the optimal solution a nd algorithm (for example, optimal time to run machines and equipment which can conform to operati onal constraint) for the manufacturing process should be installed into LCM in order to minimize w aste energy and thus carbon dioxide emissions. 4) Resource utilization: Sivasubramanian et al. [10] d escribed that resource utilization in today industr y can be typically observed from raw material usage and q ueue/waiting time in the process and priority rule in the process chain. These factors can become as cons traints in problem formulation in order to create optimal production algorithm. The percent of carbon contents can be reduced when percent of resource utilizations are increased because unnecessary ener gy for CO 2 emissions is also reduced. 4.0 Implementation of LCM Three implementations have been explored at Brunel University for LCM. The configuration of implementation of LCM is shown in Figure2. Fig. 2. Implemented Concepts for LCM 406 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION 1) Development of operation models: this method is de veloped for establishing suitable objective functio n which can reduce carbon content from manufacturing processes. All resources causing carbon emissions are considered as constraints in the operation mode l in order to prevent unnecessary wastes occurred i n idle and down time while the finished products can conform to customer’s demand. Therefore, it could b e described in another way that this method is specif ic for carbon minimization. 2) Using bench-top/micro machines: These kinds of mac hines have been developed in the concept of less energy consumption and small space requirement for processing. The reduction of carbon content of this method is specific on machines/equipments (location s). At Brunel University, bench-top machines have been developed for micro manufacturing purposes. Ho wever, it can be also used for LCM by taking advantage of their low energy consumption, resource efficiency and small foot print. 3) Applying of Devolved manufacturing: Bateman and Ch eng have introduced the concept of Devolved Manufacturing which aims at achieving mass customiz ed rapid manufacturing in a devolved web-based manner [11]. This method can be applied to the conc ept of LCM by minimizing carbon emission from make to order product (upstream) by customizing pro duct via Internet-based instead through the nearest location (downstream) to pick-up finished goods. It can be explained in another words that this approa ch is focused on reducing carbon emission from supply network. 5.0 O peration Models for LCM In this section, the operation models for LCM syste m are presented at two levels which concentrate on minimization of total used energy. The operational models are concerned with supplied chain level and shop- floor level respectively. 5.1 An Operational Model at Supply Chain Level The model formulation was established based on the supply network presented by Taha [12]. The objectiv e function sums up of total used energy in unit of jo ules to produce electricity of electrical flow in t he supply network operation (source: power plant to sink: spe cific shop floor). The goal of this formulation is to minimize carbon intensity in supply network by find ing the optimal electricity from (X ij) between node i and j in unit of kWh. The formulation can be described as : Min (f = ∑ Ω∈),( ji ijij XEn ) (1) Subject to ∑ ∑ Ω∈ Ω∈ =− ),( ),(kj k ji i jijjk fXX Zj ∈∀ :max:min ijijij CXC ≤≤ Ω∈∀ ji, 0≥ijX Ω∈∀ ji, where Z - set of node (location) in network = {A, B, C, D , E} Ω - set of arc (path) in network = {(A,B), (A,C), (C ,B), (C,D), (B,D), (B,E), (D,E)} En j - energy factor coefficient for flow X ij (joules) C i,j:max - maximum electrical capacity of arc (i,j) (kWh) 407 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION C i,j:min - minimum electrical capacity of arc (i,j) (kWh) f j - total net flow at node j (kwh) 5.2 An Operational Model at Shop-Floor Level This formulation is developed by using the theory o f linear programming solution (LP) [12]. The goal o f this formulation is to minimize primary energy used duri ng the manufacturing process by finding the optimal time (X ij) to produce product i on machine j. The problem fo rmulation can be described as follows: Min (f = ∑∑ = = n i j ijij XEn 1 1 φ ) (2) Subj ect to ∑ = N i ijij XS 1 jP≥ ∑∑ = = N i j ijij XC 1 1 φ E≤ ∑∑ = = N i j ijij X 1 1 φ δ L≤ AjBiX ij ∈∈≥ ;;0 where A - set of machines in the system {1, 2, …, Ф}, Ф is the maximum number of machine B - set of products {1, 2, …, N}, N is the total num ber of product type En ij - coefficient of energy used to produce product i on machine j δ - coefficient of lubricant used to produce product i on machine j C ij - coefficient of electricity consumed to produce p roduct i on machine j S ij - processing time for producing product i on machi ne j P j - demand of total finished goods on machine j L - total lubricant per period that equipment can r esist E - total electricity in specific area per period t hat shop-floor’s fuse can resist 6.0 E xperiments and Results 6.1 The System and Processes There are five machines in the system: cutting mach ine, milling machine, machine centre, inspection ma chine and packaging machine. Each machine has two basic d evices of the motor and oil tank to enable it in operation. The system starts operation at 8.00 am a nd ends at 10.00 pm. The process operates as job sh op sequences by producing two products: gear and spind le. Processes of gear are cutting, milling, machini ng, 408 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION inspect and packaging. Processes of spindle are mac hining, cutting, milling, inspect and packaging. Pr ocessing time of both two products is listed in Table 1 and energy consumption rate in Table 2. Table 1. Processing time of the gear and spindle on each machine Product P:M1 P:M2 P:M3 P:M4 P:M5 Gear 15 15 15 15 15 Spindle 15 15 15 15 15 30 30 30 30 30 Table 2. Energy consumption rate to produce the pro duct on each machine product e:m1 o:ot1 e:m2 o:ot2 e:m3 o:ot3 e:m4 o:ot4 e:m5 o:ot5 Gear 1 1 1 1 1 1 1 1 1 1 Spindle 3 2 3 2 3 2 3 2 3 2 sum 4 4 4 4 4 4 4 4 4 4 M1: cutting machine; M2: milling machine, M3; machi ne centre; M4: inspection and M5: packaging; P:Mj = processing time on machine j ( j = 1,2, …., 5); e:mj = electricity rate (kwh/cycle time) on motor j ( j = 1,2, …., 5); o:otj = oil rate (litre/cycle time) on oil tank j ( j = 1,2, …., 5). Energy is still provided to th e devices although they do not perform any work (down and idl e time) with E = 90 kWh, L = 65 litres. If total am ount used electricity and lubricant are consumed over th eir limit, all motors and oil tanks will be shut do wn for 5 hours. If total electricity and oil used are over t heir limits, the value of these two variables will be reset to 0. 6.2 Optimization Procedures Operation model aims at the optimal value by using optimization function in MATLAB programing. Optimal values can be the optimal time to turn-off each dev ice. Secondly, optimal values can be used to establ ish operational shift for each device. In this research , two systems are established with same conditions and simulated to observe energy used from the process o n ProModel simulations. The configuration of the sy stems in ProModel is illustrated in Figure 3. The first s ystem is run normally but the second system is run with LP (shop-floor) model. Operational shift for the secon d system is presented in Table 3. Table 3. Operational shift for each device Device Time Motor 1, 2, 3, 4 and Oil tank 1, 2, 3, 4 16.30 pm Motor 5 and Oil tank 5 13.00 pm 409 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION Fig. 3. The configuration of the systems in ProMode l simulations 6.3 Results Both systems are operated from 8.00 am to 1.00 am ( to get results at steady state) in the same conditi on including inter arrival time of entity and operatin g algorithm. After running system simulation by usi ng ProModel, the comparison of location states single between two systems are shown in Figure 4 and 5. Ru nning the system with shop-floor model, the second system can eliminate percent of down time from operating period. Fig. 4. Location states single of the first system 410 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION Fig. 5. Location states single of the second system Devices in the first system are down after and can not operate again until the end of operation shift. It can be described that unnecessary carbon emission occurred and thus the wasted energy. The statuses of device in the first and second system are shown in Figure 6. Fig. 6. The status of Motor1 in the first (left) sy stem and second (right) system 6.4 Carbon Emissions The amount of used energy is transformed into the u nit of joules firstly then multiplied with emission factor and fraction of carbon oxidised to get carbon conte nt in unit of Gg C according to the IPCC [13] appro ach. Energy consumption rate of motor and oil tank at do wn time & idle time are assumed to be at the rate o f 0.067 kwh/min and 0.067 litre/ min respectively (each dev ice’s capacity = 1 and it is assumed that energy is consumed every 15 minutes at down & idle time: 1/15 = 0.067). The calculation of carbon emission from the first and second system is listed in Table 4. Table 4. Carbon emissions from the first and second system 411 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LO W CARBON MANUFACTURING: CHARACTERIZATION, THEORITI CAL MODELS AND IMPLEMENTATION 7.0 Concluding Remarks In this paper, the characterization and implementat ion for low carbon manufacturing (LCM) have been explored specifically for the manufacturing system from the upstream (demand of the product) through downstream (finished goods) of the process chain. I n the simulated experiment, the results show the re duction of total used energy and carbon emissions when appl ied operation model at shop-floor level. The idle t ime and down time have been considered as main factors for the unnecessary wastes which can be reduced with en ergy constraints in mathematical model. However, the ope ration model can be improved in the future by takin g account of more queuing system constraints. For the future work, energy consumption data from the benc h-top machine developed at Brunel University will be used to further evaluate and validate the models and simulations and the analytical approach as a whole. Furthermore, the operation model will be applied t o Devolved Manufacturing scenario for reducing CO 2 emissions at the extended manufacturing supply cha in level. References [1] Fan Y., Liang Q., Wei Y and Okada N., ‘A model for China’s energy requirements and CO 2 emissions analysis’, Environmental Modelling & Software , 22, 378-393, 2007. [2] Golove W. H. and Schipper L. J., ‘Long-Term Tre nds in U.S. Manufacturing Energy Consumption and Ca rbon Dioxide Emissions’, Energy, 21, 683-692, 1996. [3] Gugor, A. and Gupta, S.M., ‘Issues in environmental ly conscious manufacturing and product recover: a s urvey’ . Computers & Industrial Engineering , 36, 811-853, 1999. [4] Mouzon G., Yildirim M. B. and Twomey J., ‘Oper ational methods for minimization of energy consumpt ion of manufacturing equipment’, International Journal of Production Research , 45, 4247-4271, 2007. [5] Okazaki Y., Mishima N. and Ashida K., ‘Microfac tory-Concept, History and Development’, Transactions of the ASME: Journal of Manufacturing Science and Engineer ing, 126, Issue 4, 837-844, 2004. [6] Mishima N., ‘A Study on a Microfactory and an E valuation Method of its System Configuration’, Proceedings of IEEE International Conference on Mechatronics and Automa tion, 25-28, 837-842, 2006. [7] Bateman J.R. and Cheng K., ‘Extending the produ ct portfolio with devolved manufacturing: methodolo gy and case studies’, International Journal of Production Research , 44, No. 16, 3325-3343, 2006 [8] Wikipedia, in link: http://en.wikipedia.org/wik i/Energy_efficiency accessed on 19/02/2008 [9] Mulholland K.L. and Dyer J.A., ‘Process Analysi s via Waste Minimization: Using Dupont’s Methodolog y to Identify Process Improvement Opportunities’, Environmental Progress, 20, No. 2, 75-79, 2001. [10] Sivasubramanian R., Selladurai V. and Gunaseka ran A., ‘Utilization of bottleneck resources for pr ofitability through a synchronized operation of marketing and manufactu ring’, Integrated Manufacturing, 14/3, 238-246, 2003. [11] Bateman J.R. and Cheng K., ‘Devolved Manufactu ring: theoretical perspectives’, Concurrent Engineering: Research and Applications, 10, No.4, 291-297, 2002. [12] Taha H.M., ‘Operations Research An Introductio n’, Prentice-Hall, Inc., USA, 1997 [13] Intergovernmental Panel on Climate Change (IPC C), CO 2 Emissions from Fuel Combustion (2006 Edition), 200 6. 412 Micro/nano and Precision Manufacturing 413 414 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS 0BMODULAR LASER INTEGRATION INTO MACHINE TOOLS A new outlook on increasing manufacturing flexibility, productivity and quality Christian Brecher, Fritz Klocke, Michael Emonts, Jörg Frank Fraunhofer Institute for Production Technology IPT, Steinbachstr. 17, 52 0 7 4 Aachen, Germany 1BAbstract In order to improve the flexibility, speed and quality of the manufacturing of parts and components with laser-treated surface areas, the Fraunhofe r IPT has developed a new technology platform from which to integrate laser modules i nto machine tools. In the course of this paper, this new concept of modular laser integration is described, using the example of a lathe/milling machine that has been eq uipped with two different laser tools for hardening and for deposition weldin g. The hybrid machine tool is the first of its kind not to need component reclamp ing between the fully- automated turning, milling, drilling, laser hardening and laser d eposition welding operations. Simultaneous 4-axis processes are possible with both the laser processes and the cutting processes. Some machining results are discu ssed. The paper concludes with a short outlook on the economic potential of the new laser integration conc ept. 1.0 Introduction The manufacture of complex components with localize d sections of optimized peripheral layers that are subject to partial or severe stress currently invol ves different manufacturing processes (turning, mil ling, drilling, hardening, cladding) being carried at var ious machining stations. Such complex, stressed com ponents, e.g. in propeller, drive and crank shafts, spindles , flanges and blanking punches) are used in the air and space industry, in the automotive industry, for machine a nd plant construction and in the tool and die makin g industry. These components are currently turned and milled before being subjected to an annealing or c ladding process. In order to meet high standards of shape a nd dimensional tolerance, the components are then subjected to further turning and milling processes after heat treatment. Each of these individual mach ining stages are currently carried out on different machi ne tools or work centers, which results in higher transportation time, storage time and rigging time [1]. The integration of laser system technology into con ventional machine tools enables the combination of turning, milling and drilling processes with the in novative laser machining processes of laser hardeni ng and laser cladding without having to reclamp the work p iece. This additional functionality not only increa ses the degree of manufacturing flexibility, but also signi ficantly reduces production flow times, which is pa rticularly 415 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS advantageous for small and medium-sized production runs or for the speedy manufacture of critical spar e parts (Fig. 1.). Fig. 1. Increased functionality, flexibility and pr oductivity of machine tools by the modular integrat ion of laser system technology Until now, there are no machine tools on the market that can flexibly carry out both mechanical manufa cturing processes such as milling and turning and laser surface treatment processes such as laser cla dding und laser hardening without compromising on the functionality and flexibility of the conventional cutting proces ses [2]. This is caused by a lack of strategies that compens ate for the negative interactions within the workin g space between the laser system technology and the cutting processes. Laser processing units that are mechani cally fixed into the machine tool’s working space can col lide with the conventional cutting tools. During th e laser process, the laser not only heats up the component but can cause thermal deformations of various machi ne elements or even of the entire machine structure wh ich, in turn, has a negative effect on machining qu ality and in particular on the component’s dimensional accura cy. The laser machining tools’ sensitive optics als o quickly become covered with coolant and material ch ips while the conventional cutting process is in pr ogress, making the level of tool wear and the likelihood of a malfunction after only a short time and at relat ively low loads unacceptably high [3]. The aim of the »KombiM asch« research project was therefore to reduce the amount of processing effort required for the indust rially manufacture of rotationally symmetrical part s by combining conventional cutting and thermal surface processing technologies into a modular machine tool . Thus, a machine tool demonstrator has been built at the Fraunhofer IPT that enables the complete manufacturing of rotationally symmetrical parts by providing both conventional cutting and laser surfa ce processing technologies in one machine tool. This d emonstrator not only provides a flexible form of conventional machining before and after heat treatm ent, but also flexible laser surface processing tec hnology based on modularly mountable laser processing units for both laser cladding and laser hardening within the same working space. This eliminates the need to rec lamp and re-reference the work piece. The hybrid ma chine tool, with its integrated modular laser systems tec hnology, can be used both for series production and for the production of individual parts via turning, millin g and drilling processes that meet all dimensional accuracy requirements and via the laser processing technologies of laser hard ening and laser cladding that can generate application-related surface properties in specific component surfaces. If post-processing operations a re still necessary after the laser processes in order to cor rect an aspect of the geometry, the work piece does not need to be reclamped but can be turned or milled within the same hybrid machine tool. Other processing stag es to 416 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS protect the component during storage and transporta tion can therefore be eliminated, e.g. greasing, de greasing and packaging processes. As some of these operation s are carried out by sub-contractors, the eliminati on of these additional operations helps to reduce environ mental pollution. The hybrid »KombiMasch« machine tool concept incorp orates the following manufacturing technologies: milling, drilling and turning before heat treatment , laser hardening, laser cladding, milling and turn ing after heat treatment. The systems engineering developed a t the Fraunhofer IPT covers the process technology investigations in industrial applications and an ev aluation of the project results. Fundamental invest igations into issues related to materials and applications a s well as specific tests of the new laser cladding and laser hardening unit were carried out in relation to the manufacture of demonstrational parts. This led to t he characterization of the process technology conditio ns needed in order to define the final specificatio ns and to the execution of process investigations on the »Kom biMasch« machine tool into combined manufacturing processes. The process data determined in the cours e of this research work was stored in a CAD/CAM sys tem [4]. 2. 2.0 Systems Engineering Fig. 2. Combination of conventional machining and l aser process technology in one machine tool In order to combine the different laser processes w ith cutting processes in only one machine, the Frau nhofer IPT has, for the first time, developed a technical platform for combined laser processing and conventi onal cutting processes. At the heart of this platform li e the modularly integrated laser processing units a nd the innovative optical and mechanical interfaces betwee n laser processing units and laser source. The desi gn of these modules makes it quick and easy to integrate the laser processing units into a conventional mach ine tool. The guidance and control of the laser processing un its are accomplished by the axes and the control sy stem of the machine tool. In order to demonstrate the poten tial of the newly developed modular laser integrati on technique, the Fraunhofer IPT built and successfull y tested the first machine tool that combines turni ng, milling, drilling, laser cladding and laser hardeni ng (Fig. 2.). Compared to other techniques used to integrate laser system technology into machine tools, the hyb rid »KombiMasch« machine tool developed by the 417 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS Fraunhofer IPT is the first that does not involve c omponent reclamping while at the same time offering a flexible and fully automated range of laser surface treatment processes. The manufacturing processes o f turning, milling and drilling can also still be car ried out as flexibly as in other conventional machi nes. Both the laser and cutting processes are capable of simultan eous 4-axis machining, making it possible to machin e almost any component geometry. The turning/milling machine was developed as an inclined bed. The main spindle, opposing spindle, turret, tailstock, 4-axi s milling spindle (B-axis spindle) and tool exchang e magazine modules are designed to be added on as ‘building bl ocks’. A list of the technical data for this machin e tool is given in the following table. Maximum turning length 900 mm Maximum pitch 280 mm Swing over bed 600 mm Main and opposed spindle Motor spindle 33 kW Speed range 25 - 4000 min -1 Maximum torque 630 Nm B-axis spindle (travel) X: +440 mm, -10 mm Y: ±100 mm Z: +100 0 mm B: ±95° (swivelling angle) Motor spindle 16 kW Speed range 25 - 1200 0 min -1 Maximum torque 76 Nm Holding torque 5000 Nm Turret 12 stations (VDI 40) Weight of machine tool 1600 0 kg 3.0 Modular Laser Processing Units Two modularly mountable laser processing units have been developed and built in order to enable the fl exible use of the laser cladding and laser hardening proce sses integrated into the machine tool. For the firs t time, these modular laser tools can be automatically exch anged into the standard tool holder in the 4-axis m illing spindle, just as the milling, turning and drilling tools are, without impairing the kinematic degrees of freedom of the 4-axis milling spindle and thus fulfilling t he relevant kinematic und mechanical requirements. This significant advantage is combined with the distinct ly multi-functional nature of the machine using jus t one 3 kilowatt high power diode laser beam source that fe eds the laser radiation with the aid of a beam swit ch via optical fibers to the laser processing unit current ly in use (either the laser cladding unit or the la ser hardening unit). The stand-by position of the laser processin g units is located outside of the working chamber o f the machine tool. Therefore the integration of the lase r processing units does not have any negative effec t on the conventional machining processes. Furthermore, the stand-by position protects the laser’s sensitive op tical components from coolant or material chip contaminat ion. The automated cladding process carried out by the modular laser cladding unit involves the processing of localized workpiece surface areas, generating f unctional sections in the peripheral layers. The application of such layers increases the functionality of the c omponent surfaces. Special peripheral layers can be generate d which are relevant to a specific application. Sub strate materials or materials with particular characterist ics can therefore be used cost-effectively. The add itional material for the cladding process, in the form of w elding wire, is fed into the laser focal point, whi ch fuses the upper layers of the substrate material during the l aser cladding process. By moving along the componen t Table I. Technical Data 418 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS surface, the laser generates a precisely defined we lding bead close to the edge of the workpiece. Comp ared to conventional deposition welding, the laser cladding process is very stable and only distorts the workp iece to a minimal extent as the localized absorption of laser energy causes only minimal heat induction. The las er cladding unit consists of the optical laser guidanc e and forming components, the mechanical connection to the optical fiber to decouple the tensile forces, and t he housing for the beam guidance unit, to which the wire feed unit and the process monitoring sensor are attached . The HSK63 interface that is attached to the housi ng around the laser cladding unit makes it possible to exchange the laser cladding unit in the B-axis spi ndle HSK tool holder, similar to the tool exchange in conven tional milling and drilling tools. In order to ensu re sufficient stiffness while keeping the weight of the laser cla dding unit low (<20 kg), the beam tube consists of a fiber reinforced material designed and manufactured by th e Fraunhofer IPT. The tensile strength of the indiv idual fibers and the direction in which the layers of fib ers are wrapped have been developed in such a way t hat the resulting stiffness of the fiber reinforced plastic (FRP) beam tube roughly corresponds to the stiffne ss of a steel tube of the same size (Fig. 3.). Fig. 3. Modular laser cladding unit The laser cladding unit contains two tilted mirrors that guide the laser radiation coupled into the cl adding unit by the optical fiber to the beam axis, coaxial to t he B-axis spindle, before being focused by the focu sing lens system. Both mirror systems have cross roller beari ngs to compensate for the rotational movement in th e optical fiber connection that arises when the B-axi s spindle rotates, thus preventing the optical fibe r from buckling. All laser optics have integrated cooling channels that dissipate the absorbed laser energy a nd prevent heat from accumulating. The optical fiber is enclos ed within a preloaded strain relief tube to lead th e forces in the power chain through the tube to the mirror hous ing in the laser cladding unit instead of through t he optical fiber itself (Fig. 4.). 419 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS Fig. 4. Laser beam Guidance system of the modular l aser cladding unit As with the laser cladding unit, the laser hardenin g unit consists of a fiber connection with revolvin g bearings on the housing and can also be exchanged via the HS K63 interface into the B-axis spindle (Fig. 5.). Th e laser hardening unit contains a laser scanner with oscill ating reflective scan optics (scanning mirrors) tha t creates the laser scanning field for flexible laser hardeni ng. This standard module can generate any scanning field geometry within the 50 x 20 mm² scanning field. The maximum laser power that can be transmitted by the oscillating scanning mirrors is 3 kW. The Fraunhofe r IPT developed the relevant mechanical and optical interfaces in order to integrate the standard scann er module into the machine tool and make it possibl e to exchange the laser scanner safely and reproducibly into the B-axis spindle as well as into the ‘park’ position. The laser scanner is enclosed within a scanner cage ; the HSK63 interface is attached to this cage. For maintenance purposes, the laser scanner module can easily be removed from the scanner cage. The scanne r cage is equipped with the same optical interface sy stem with rotational bearings for the optical fiber connection as the laser cladding unit and guides th e collimated laser radiation via a tilted mirror wi th rotational bearings to the laser scanner’s oscillating optics. The scanner cage has also been equipped with a cro ss jet system that emits a stream of purging air perpendic ular to the laser radiation to protect the F-Theta lens in the laser scanner from dust particles and dirt in the p roduction environment. Fig. 5. Modular laser hardening unit 420 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS 4.0 CAD/CAM Technology Die The newly developed hybrid machine tool with in tegrated laser system technology is not only modula r in terms of its mechanical design, but also in terms o f the control interfaces. A profibus interface is r esponsible for the communication between the laser source, the laser beam tool and the machine tool. The machine tool’s control unit takes over the main control function. Both the laser tools and the cutting tools are conf igured and administered via the machine controls. The tools pa ths are programmed and the tools (protective gas, l aser power, wire feed) are switched or set using new NC functions that have been specially developed at the Fraunhofer IPT for carrying out laser cladding and hardening processes within the lathe. The developme nt of the new NC functions are based on the results of ex tensive investigations into the process technology involved in wire-based laser cladding, laser hardening and c utting after heat treatment. Strategies were develo ped for wire-based laser cladding processes that produce hi gh quality thick films (layer thickness > 3 mm) con sisting of several layers being built up via an alternating series of cladding and cutting operations. This is the first time that laser hardening strategies have been deve loped with which to flexibly utilize laser scanner technology in machine tools. The hardening techniqu e involves a rapid oscillation (> 1 m/s) in the las er beam (galvo scanning mirror) being superimposed over the feed motion of the laser tool (machine axes). This makes it possible to closely control the introduction of energy into the component and therefore to achieve a high level of machining quality. The machining process a lso becomes very flexible in terms of geometry. The NC functions have been transferred into different CAM machining strategies and integrated as a software m odule into a CAD/CAM programming system for turning/milli ng to make it easier for the operator to work with the new NC functions and to link the different cutting and laser operations to create high performance pro cesses. The operator therefore has comprehensive NC program ming for all the machining operations needed to perform all necessary processes on the component. N C cutting machine operators who are generally train ed to use one specific process, as opposed to a whole ran ge of cutting and laser processes, will therefore f ind it easier to work with the new machine. 5.0 Process Technology and Technological Potentials The new machine concept, with its considerable pote ntial in terms of automation and accelerated compon ent manufacture, can now be applied to a wide range of components that are currently produced via a combin ation of turning, milling, drilling, cladding/ deposition welding (TIG, PTA) and/or hardening (e.g. inductio n). The stop valves and control valves, pistons and baskets used in power plant fittings are typical examples of the components that this machine can be used to produce . The conventional manufacture of these components starts with a mechanical processing followed by dif ferent thermal processing and a final finishing pro cessing (hard machining). The overall process also includes various quality assurance tests, such as tear or h ardness tests. These tests are needed because the component s must fulfill important safety functions; sudden component failure would result in unacceptable dama ge. The components pass through a total of at least four different stations before they are finished (cuttin g before heat treatment, peripheral layer modificat ion, cutting after heat treatment, quality inspection). By substituting the laser cladding and laser harden ing processes for the conventional cladding and har dening processes, the new hybrid machine tool can process the component parts from start to finish. The combi nation of laser processing and cutting in quick succession without the time-consuming reclamping processes ma kes 421 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS this system both technologically and economically v aluable. Using the laser to process the workpiece i nstead of conventionally welding and hardening it reduces component warping and, in turn, minimizes the amoun t of hard machining processes needed. The ability to int roduce energy precisely and locally into the compon ent also makes it possible to produce complex and delic ate near net geometries. The use of laser technolog ies means that the surface treatment processes can be f ully automated with the aid of conventional CAD/CAM /NC technology. Figure 6 shows the control valve from a high-pressure stop valve made using the KombiMasc h at two different stages of the machining process (prep ared for laser cladding and completely finished) as well as a longitudinal cross section through the finished c omponent. The bond between the layers (1.4923/Stell it 6) meets all the manufacturer’s requirements in terms of the amount of pores, cavities and cracks. The mi xing required to form a metallurgical bond between the s ubstrate and the cladding material takes place in a relatively small transition zone. A metallurgical p erfect bond is generated between the layer and the substrate material (close-up, Fig. 6.). The profile of the ha rdened grain structure in the pan-head also fulfill s the manufacturer’s standards in terms of the properties needed for this specific application. Fig. 6. Laser clad cone spindle (left), after mach ining (right) Fig. 7. Hardness of pan-head after laser cladding The cladding is generated in a two-stage process. F irst, the laser heats up the pre-machined pan-head to 250 °C (the workpiece rotates while it is heated up). The cladding is then welded on, layer by layer. Before the next layer is welded on, the layer that was previously w elded on is milled or lathed back to a defined heig ht to remove any impurities and create the smooth surface topography (removal of ripples/undulations) needed for the wire-based laser cladding process. Compared to other cladding strategies, the ability of this auto mated machine to quickly change between laser tools and c utting tools means that any additional cutting is k ept to an 422 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS absolute minimum. After laser cladding, the compone nt is subjected to the final machining stage in ord er to eliminate any warping. The pan-head is turned after heat treatment to give it its final geometry. 6.0 Economic Potentials The first field tests in an industrial pilot applic ation have shown that the newly developed machine g enerates added technological, economic and ecological value in the product value creation process (Fig. 8.). Th e use of this hybrid machine tool would lead to significant improvements in terms of throughput time and proces sing quality for a wide range of component parts that ar e currently produced in a series of conventional tu rning, milling and drilling, cladding and hardening operat ions with additional finishing (grinding, cutting a fter heat treatment). Typical examples of this type of parts are valves and control valves, pistons and shafts t hat are subject to high mechanical and/ or thermal load cyc les. The sections of a component part that are subj ected to particularly high loads (e.g. valve seats) are clad ded or hardened with protective wear and corrosion resistant layers to prolong the service life of the entire co mponent. The combination of conventional turning, m illing and drilling operations with laser surface treatmen t in one hybrid machine tool makes it possible to p erform all machining operations on the work piece without havi ng to reclamp the work piece. The modular concept o f the hybrid machine tool means that it can be customized to meet specific client requirements, safe in the knowledge that the laser processing units can not i mpair the conventional machining processes [5]. The »KombiMasch« machine tool offers the following tech nical and economical benefits: • Shortening of production flow time as a result of t he reductions in machining time, rigging time and storage time (Fig. 8.), • Increase of machining flexibility of machine tools, • Reduction of logistic efforts, • Increase of product quality as a result of the inte gration of reliable and capable laser processing technologies, • High amount of usability as a result of automated p rocedures. Fig. 8. Economic potentials 423 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODULAR LASER INTEGRATION INTO MACHINE TOOLS 7.0 Conclusion The combination of conventional turning, milling an d drilling operations with laser surface treatment in one hybrid machine tool enables the complete machining of work pieces within a single clamping. For the fi rst time two modularly mountable laser processing units for laser hardening and laser deposition welding b ased on a new combination of mechanical and optical inte rface to the machine tool have successfully been developed, built and tested. In comparison to forme r machine tools with integrated laser system techno logy, the »KombiMasch« system accomplishes the automated tool change of the laser processing units into the 4- axis milling spindle. By this, simultaneous 4-axis processes are possible with both the laser processe s and the cutting processes. During mechanical processing, th e two laser processing units are ‘parked’ outside o f the machine’s working area to protect the sensitive las er optics from the machine’s vibrations, from mater ial shavings or cooling lubricants. This layout also pr events the lasers from hindering the mechanical pro cesses. The laser processing units, the laser source and th e machine tool communicate via a profibus interface . The machine tool control is responsible for the central control function. With the new machine tool components with laser tre ated surface areas have been manufactured in a full y- autometed process without any reclamping for the fi rst time. The quality of the components in terms of form, shape and the characteristic of the laser welded an d hardened rim zones is corresponding to the conven tionally manufactured components. The rigging, machining and storage time have significantly been reduced. 7B References [1] Brecher, C.; Kordt, M.; Groll, K.; Glasmacher, L.; Bichmann, S.; Emonts, M.: Automatisierte Reparaturz elle »OptoRep« – Komplettbearbeitung für den Werkzeug- u nd Formenbau. In: wt Werkstattstechnik online, 2005, Heft 11/12, Internet: www.werkstattstechnik.de. Düsseldorf: Spr inger-VDI- Verlag, Pages 1-8. [2] Brecher, C.; Emonts, M.; Frank, J.; Wenzel, C.: Hyb rides Bearbeitungszentrum für die Dreh-, Fräs- und Laserbearbeitung. In: VDI-Z Integrierte Produktion, VDI Verlag, Heft 9, 2007, Pages 34-38 [3] Kasperowski, S.: Integration von Diodenlasern in Pr äzisionsdrehmaschinen zur laserunterstützten Keramikbearbeitung. Shaker, Aachen, 2000, ISBN 3-82 65-7670-5 [4] Brecher, C.; Deutges, D.; Emonts, M.; Frank, J.; La nge, S.: Maschinenbau-Verbundprojekt »KombiMasch« - Fräsen, Drehen und Lasern in einer Maschine. In: Maschinenb au und Metallverarbeitung, Kuhn-Verlag, August 2006 , Pages 48-49 [5] Brecher, C.; Emonts, M.; Frank, J.: Kürzere Prozess ketten durch Verfahrenskombination. In: VDI-Z Integ rierte Produktion, VDI Verlag, Heft 9, 2006, Pages 77-78 424 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW LASER PROCESS MONITORING: A CRITICAL REVIEW Stournaras, A., P. Stavropoulos, K. Salonitis and G. Chryssolouris * Laboratory for Manufacturing Systems and Automation , Director, Prof. George Chryssolouris, Department of Mechanical Engineering and Aeronautic s, University of Patras, Greece * xrisol@mech.upatras.gr Abstract Laser processes are nowadays well established sheet metal pr ocessing methods. Due to their high potential for industrial applications, the good pr oduct quality, resulting also in cost savings and enhanced productivity, laser processes ar e of high importance. The development of a reliable monitoring system evaluating quality on-line , provides a solution to this problem. Thus, it is essential that advanced monitoring systems be deve loped in order for the requirements of laser processing to be fulfilled. The cu rrent work reviews the laser process monitoring systems that have been developed so far. Req uirements and characteristics of laser processing that influence product quality are described. Sensorial information from several sensors is compared and analyze d. Methods of the most important monitoring of laser processes, applications and st rategies are reviewed and evaluated in the current work. Keywords: Laser processes, Monitoring, On-line qual ity evaluation 1.0 Introduction Laser processes are thermal in which a focused lase r beam of high energy intensity is used for heating and even melting the material in order to be machined. Due to their unique characteristics laser processe s have met great response in industry ([1]):  The effectiveness of laser machining depends upon t he thermal properties and, to a certain extent, the optical rather than the mechanical properties of th e material to be machined. Therefore, materials th at are difficult to be machined, due to their high degree of hardeness or brittleness, can be processed easil y with laser manufacturing techniques if their thermal pro perties are favorable.  Since energy transfer between the laser and the mat erial occurs through irradiation, no cutting forces are generated by the laser, leading to the absence of m echanically-induced material damage, tool wear and 425 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW machine vibration. Moreover, the material removal rate for laser machining is not limited by constrai nts such as maximum tool force, built-up edge formation or tool chatter.  When combined with a multi-axis workpiece positioni ng system or robot, the laser beam can be used for drilling, cutting, grooving, welding, and heat trea ting processes on a single machine. This flexibili ty eliminates the transportation necessary for process ing parts with a set of specialized machines. In a ddition, laser machining can result in higher precision and smaller kerf widths or hole diameters than other comparable mechanical techniques do. Despite the very important advantages of laser proc esses, the result of this one could be influenced b y a number of factors such as, defects in the microstru cture of the material, contaminants on the workpiec e surface, alteration on laser beam properties, etc. resulting in a non-acceptable product ( Fig. 1 ). For that reason, the surveilance of the process is very important in order for the requested quality of the product to be assured, as well as for the efficiency of the process itself , especially if it is part of a mass production seq uence. Fig. 1. possible defects in laser welding, lase r cutting and laser drilling processes. The accurate control of a laser process can be acco mplished by ensuring that the process parameters ar e in a ceftified range of specifications. Furthermore, th e ability to detect defects that occur during proce ssing and the valid change of the process parameters values, play important role in assuring the pre-defined quality standards. Monitoring of a process, and consequent ly its control, can be categorized, based on the ti me accomplished, into three stages (Fig. 2): • Pre-process • In-process • Post-process Fig. 2. Classification of monitoring stages based in the time accomplished [2]. The in-process monitoring, is very important due to the fact that it enables the control of the proces s without stopping it and uses destructive methods for the ev aluation of the product quality, a fact that is of high importance to industrial applications. Taper Burring Spatter Cutting edge roughness Dross attachment Taper Heat affected zone Kerf width Pores ι lack of welding Cracks Pre-process Post-process Direction of movement Laser beam Processing head In - process 426 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW 2.0 Laser Welding, Cutting and Drilling Monitoring Techniques In-line monitoring of the process is accomplished b y the acquisition and evaluation of various indicat ors that can be detected during the process. Optical and ac oustical emissions as well as temperature related d ata can be measured with the use of the appropriate sensors . These signals are a result of the laser-material interaction and thus, carry information of the process and can be used for monitoring purposes. Methods and syste ms that have been developed, for laser process monitoring, can be classified based on the signal type that is acquired: • Monitoring using optical signals utilizing CMOS, CC D cameras and photodetectors. • Monitoring using acoustical signals utilizing micro phones and acoustic emission (AE) sensors • Monitoring using temperature-based, electrical and other type of data. Fig. 3. Typical process emissions during laser mate rial processing [2]. 2.1 Process Monitoring using Optical Signals An optical detector, particularly ultraviolet (UV), visible or infrared (IR) detectors, have been wide ly used for converting the flux density of the radiation, emitt ed by the welding process into an electrical signal . An optical filter is often placed in front of the dete ctor to confine the spectral ranges of the whole se nsor system. Typical setups using co-axial and off-axial arrange ments are illustrated in (Fig. 4) Fig. 4. Typical setups for optical detectors using co-axial and off-axial arrangements [3]. In ([4]), the measured signals were used for the de velopment of prototypes, used for the detection of geometrical defects in laser welding area. In ([5] , [8], [9]) the optical signals acquired from the i nteraction 427 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW zone were analyzed in order to be correlated with t hose of the welding depth and its defects. Further more, in studies ([10], [11]), the acquired photodiode signa ls were used in order for the relationship among th e plasma, the spatter and the bead shape to be clarified acco rding to the welding variables. Through a correlati on between these signals and the weld quality, there w ere a multiple regression analysis and a neural net work developed for the estimation of the weld bead’s pen etration depth and width. In laser processing syst ems, using a fiber optical fiber for the delivery of the laser beam, the reflected energy is collected from the output lenses and from that re-launched back into the fibr e. In ([24]), a method of monitoring the laser dri lling and cutting processes, based on the energy reflected an d re-launched back into the fiber, has been present ed. It was observed that the amount of the radiation refle cted could be used in order for the difference betw een successful and unsuccessful machining to be detecte d. In studies ([25], [26]), the monitoring of the surface plasma and measurement of the emitted light from th e cutting front, respectively, were used for the determination of the cutting quality and the morpho logy of the cutting edge. It was observed that the wave frequency of the signal carried information concern ing the striations formed in the cutting edge. In study ([27]), a technique, using a simple confocal sensor arrangement, based on a single-mode optical fibre for measuring the hole’s diameter at high speed, was pr esented. Light from a low-power He-Ne laser source is launched into a single-mode fibre and guided to a 5 0:50, coupler, where its amplitude is divided. In order for the diameter to be measured, the focused spot from the He-Ne laser is traversed across the hole and a dropping out of the signal is observed. This is converted i nto a diameter measurement provided that the spot i s traversing at a constant speed. (a) (b) Fig. 5. (a) The fibre-optic based measurement syste m and (b) measurement of hole formed by a fibre be am delivery system [27]. Apart from the photodiodes, various types of camera s have been used for capturing images from the processing zone. In [12], a monitoring procedure of the laser welding process was outlined, as an elem ent of an integrated automotive panel production system. The system developed measures each welding bead on – line and in – process, in the following procedure. Immediately after the butt – welding, with the use of CO 2 laser, the welding bead is irradiated at a right an gle to the bead direction and with a slit light of a semiconductor laser, in an oblique position above t he bead. The light reflected from the bead is phot ographed by a TV camera, positioned directly above the bead and two – dimensional monochromatic image data, is obtained for the section of the welding bead. The slit light cutting line is extracted from this imag e data to convert it into waveform data. This data is proces sed in a computer to obtain five bead configuration al data (swell, step, width, inclination and dent). Finall y, this data is compared with specified reference o nes in order for the welding quality to be assessed. 428 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW A real – time system was developed to process the i mage of the weld pool in [13] . In this study, a p olar coordinate model was constructed to characterize th e weld pool geometrically. The identification of t he weld pool parameters was made with the use of a neural n etwork. An accurate boundary of the weld pool was acquired with the use of a camera viewing the weld pool region at a 45o angle from the rear of the wel d pool and a pulsed laser illumination. A fast image – pr ocessing algorithm was also developed to extract th e boundary of the weld pool in real time. Additionally, in ([14], [15]) there was a system pr esented for laser welding and laser cutting process monitoring with a CMOS camera, which observed the p rocess, in a coaxial and centered to the laser beam position. The software of the camera evaluates onl ine different failures and process parameters with the help of characteristic regions within the image of the c amera and the recorded film is also used for analyz ing the process and improving the process parameters. The systems have been used in industrial applications ( [15], [16]). (a) Laser welding (b) Laser cutting Fig. 6. Monitoring system for (a) laser welding and (b) laser cutting ([14]- [16]) In ([21], [22]) a CCD camera was used for recording the magnitude of the irradiance emitted from the cut front and the shape of the erupted sparks in laser cutting and the images captured were used for the e valuation of the cutting quality and the on-line control of t he process. A CCD camera was also used for monitor ing the laser cutting process under study [23]. The imagin g of the IR radiation, obtained with the CCD camera of the Coaxial Process Control, can be interpreted by comp aring it with the thermal emission as calculated fr om the dynamical model and correlated with quality charact eristics, such as the dross attachment and cutting edge's striations. Fig. 7. (a) CCD image as recorded and (b) Simulatio n of the model [23]. 429 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW 2.2 Process Monitoring using Acoustic Signals Process monitoring through acoustic signals involve s a sensor that converts, in a measurable variable, the process sounds into electrical output. In -Process monitoring through acoustic emissions involves a microphone, placed nearby the processing zone to me asure air-borne emissions or a piezoelectric transd ucer, mounted at an acoustic mirror, acoustic nozzle and workpiece for measuring structure born emissions. Fig. 8. Typical setup for acoustic emissions [3]. In [17], an attempt to determine a link between air -borne acoustic emission signals and weld character istics was accomplished. Acoustic emissions when spectral ly resolved into a power spectrum of 20 1 kHz sub-b ands between 20 Hz and 20 kHz showed a distinctive frequ ency distribution under different welding condition s. It was shown that a comparison between the sum of a sq uared standard deviation over these 20 sub-bands wi th a standard result is capable of indicating weld quali ty. Furthermore, in studies ([18], [19]), optical and air-borne acoustic emissions during laser welding were acquir ed with the use of a photodiode and microphone. Re sults indicated that the keyhole and its surrounding liqu id layer acted as a frequency selective amplifier f or pressure fluctuations induced by changes in the interaction of the laser radiation with the walls of the keyhol e ([18]). The optical emission was predicted to be increasing almost linearly with initial vapour flow rate and it was of the same magnitude as the measured emissions. The p lume acoustic emission was predicted by considering the volume of ambient air displacement, required by the outflow of the vapour coming out from the laser we ld keyhole. The analysis predicted that the plume acou stic emission should vary, as the time derivative o f the vapour flow rate, and hence, as the time derivative of the optical signal ([19]). One of the first studies of using air-borne acousti c emissions for monitoring the laser drilling proce ss, was presented in [28] and results have shown that the a coustic signal is proportional to the thermal energ y liberated during the combustion process and consequently, it depends on the amount of the combustible material ablated. An experimental investigation on laser dr illing and laser grooving processes in ([29] - [31] ), utilized the acoustic waves, generated during the impingemen t of the gas jet on the erosion front. Results ind icated an inverse relationship between the depth of cut and t he resonant frequency of grooving, cutting and dril ling. 430 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW Fig. 9. Experimental setup for laser drilling and laser grooving monitoring using air-borne acoustic emissions ([30], [31]). In study [32], an attempt was made to clarify the f requency characteristics of the laser processing so und by calculation, in which experimental data of the lase r drilling with a single pulse laser beam was used, when a continuous pulse laser beam was assumed to be used for laser grooving, and for comparing the calculate d frequency characteristics with the experimental cha racteristics. In this research, sound frequencies o f up to only 20 kHz were monitored with a microphone. Furt hermore, in [33], an attempt was presented to deter mine a link between the strength of processing sound and the groove cross-sectional area per pulse, when a processing condition and a work material is changed . Results indicated that the cross-section of mate rial removed can be calculated by measuring the strength of the processing sound. Finally, in [34], the influence of laser pulse shap e on the laser drilling process and on the optoacou stic effect was studied for two laser applications: laser drill ing in aluminium and laser drilling in hard dental tissues are presented. Results confirm that the laser pulse sh ape might have a significant influence on the inter action processes and can be used for increasing machining efficiencies whilst the optoacoustic signals are fo und to be depending on the laser pulse shape 2.3 Process Monitoring using Temperature and other Measurements In study [6], temperature measurements, inside and near the weld pool during the laser welding process , were presented. The temperature fields inside the mater ial are determined with thermocouples and on its su rface are determined by the use of a charge coupled device (C CD) camera with infrared filters. The laser weldin g experiments were performed on austenitic stainless steel with the purpose of using the determined ther mal fields as a database to calibrate a finite element simulation of the process. In [7], a sensor system is presented for monitoring the misalignment of the edges and un dercuts during the laser welding. The principle tha t the sensor is based on is the measurement of the distri bution of the heat radiation from the weld pool. T his system consists of three sensors, one positioned to the le ft of the weld pool, one to the right and one as a reference sensor directly above the weld pool, which detects the heat radiation from the weld pool. From these data the power received can be calculated by each sensor, th e misalignment as well as the undercut. 431 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW Fig. 10. Sensor arrangement for monitoring system i n [7]. In study [35], a monitoring system for the laser we lding process that utilizes a plasma charge sensor, was presented. The plasma behaviour has been observed during welding through measuring the space charge voltage induced on an electrically insulated weldin g nozzle, the plasma charge sensor (PCS). It was sh own both theoretically and experimentally that the indu ced voltage is a measure of plasma temperature and thus, of the welding performance. In particular, the results have indicated, under laboratory conditions, that the PCS signal can measure the weld penetration and detect a wide range of weld defects. Fig. 11. Basic configuration of Plasma Charge Senor (PCS) [35]. 2.4 Process Monitoring using Multiple Sensors It has been demonstrated that combinations of diffe rent types of sensors, such as the photodiode and t he electronic camera, or the photodiode and the microp hone, or combinations of photodiode and filter arrangements being sensitive either to the IR, VIS or UV spectral region, sometimes aiming at differen t parts of the interaction zone, give an increased signific ance for the detection and classification of treatm ent faults, and at the same time, reduce the false alarm probab ility by, e.g., correlation-based signal assessment methods [36]. In [37], an approach with additional sensors and advanced evaluation methods has been investiga ted. The goal was to create flexible algorithms that can be used for the observation of laser welding as we ll as laser cutting. The signals from the optical and acoustic detector are processed by statistical methods imme diately after they have been acquired. Short time mean valu es, standard deviations, derivatives and histograms are calculated from the data of both sources separately . After this first analysis, there is a statistical set of data for each method and sensor. In the second stage, the f uzzy part of the analysis was designed to process t he 432 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW statistical results. This concept allows for a faul t classification and a highly reliable and easy to interpret evaluation. The outcome is reference free and is no t based on measurement units. a) Laser cutting b) Laser welding Fig. 12. Cutting and welding example with correspon ding signals and overall fuzzy ratings [37]. The relationship between airborne acoustic and opti cal emissions from the laser welding process is des cribed in [38]. The emissions were measured with the use of a microphone and a photodiode during laser weldi ng experiments. A thorough investigation of the signa ls recorded has revealed them to be highly related at a phase shift corresponding to the delay time for sou nd to propagate from the weld area to the microphon e. A model was constructed which predicted the acoustic signal from light signal measurements. It was show n that the acoustic signals predicted, corresponded fairly well to the experimental acoustic data. The relat ionship between the signals has revealed that the sound pre ssure predictions were proportional to the time der ivative of the light signal samples. Fig. 13. Experimental setup for laser welding monit oring using acoustic and optical signals [38]. 3.0 Conclusion In the present work, a review of the surveillance t echniques and methodologies for the most potentiall y laser processing techniques, including laser welding, las er cutting and laser drilling, has been presented. As it was shown, process information can be obtained by measu ring and characterizing the various forms of energy that propagates from the laser material interaction site . Electromagnetic radiation and acoustic waves, ai r-borne or structure-borne, are two of the most significant si gnals that can be used for extracting information r egarding process’ evolution and characteristics and can be a cquired using co-axial of even off-axial sensors' 433 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW arrangement. Optical signals can be acquired using appropriate sensors, such as photodetectors or cam eras, while acoustical signal using microphones or piezoe lectric sensors. To keep pace with the demands of the production ind ustry, such as cost, quality and functionality, fut ure systems of the laser process monitoring and control have to provide a further increasing level of spee d, reliability and flexibility. Something that could be accomplished by real-time signal assessment syst ems, fed by extended reference and calibration data files, a -priori defined and/or self-generated for specific treatment situations, and by the on-line acquired signals of integrated multi-sensor fusion setups. Future moni toring systems need also to be more user-friendly, incorpo rating standardized user interfaces and increased functionality (self-testing, automated fault detect ion and correction actions, etc.), in order to be e asily accommodated from industrial users. Acknowledgement This paper is part of the 03ED757. research project , implemented within the framework of the “Reinforc ement Programme of Human Research Manpower” (PENED) and c o-financed by National and Community Funds (25% from the Greek Ministry of Development-General Secretariat of Research and Technology and 75% from E.U.-European Social Fund). References [1] G. Chryssolouris., “Laser Machining: Theory and Pra ctice”, Springer-Verlag, 1991. [2] M. Kogel-Hollacher, C. Dietz, T. Nicolay, J. Schmid , M. Schmidt, J. Bahnmüller, B. Kessler, B. Schürma nn, M. G. Müller, “Process monitoring or process control in l aser materials processing”, International Congress on Applications of Lasers and Electro-Optics (ICALEO) , 2001. [3] Y. Shao, Y. Yan, “Review techniques for On-line mon itoring and inspection of Laser welding”, Journal Of Physics: Conference series , Vol. 15, pp. 101-107, 2005. [4] N.E. Rudiger, M. Houlot, "Integration of real time quality control systems in a welding process", Journal of Laser Applications, Vol. 9, 1997, 95 – 102. [5] P.G. Sanders, K.H. Leong, J.S. Keske, and G. Korne cki, "Real – time monitoring of laser beam welding using infrared weld emissions", Journal of Laser Applications , Vol. 10, Number 5, October 1998, 205 – 211. [6] S. L’hospitalier, P. Bourges, A. Bert, J. Quesada, and M. Lambertin, "Temperature measurement inside a nd near the weld pool during laser welding", Journal of Laser Applications , Vol. 11, Number 1, February 1991, 32 – 37. [7] C. Kirby, P. Abels, A. Drenker, H.Nitsch, E. Beyer, “Process monitoring for Laser Beam Welding and App lications”, Laser Assisted Net shape Engineering, Proceeding of the LANE’94 , Vol.1, 125 – 134. [8] G. Riciardi, M. Cantello, F. Mariotti, P. Castelli, M. Penasa,"On – line process control used in the L aser welding of metal sheet lightweight structural Panels," Annals of the CIRP , Vol. 48, 1999, 159 – 162. [9] K. Mori, and I. Miyamoto, "In – process monitoring of laser welding by the analysis of ripples in the plasma emmision," Journal of Laser Applications , Vol. 9, 1997, 155 – 159. [10] H. Park, S. Rhee, "Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized stee l," Optics & Laser Technology , Vol. 31, Issue 2, March 1999, 119 – 126. [11] H. Park, S. Rhee, “Estimation of weld bead size in CO2 laser welding by using multiple regression and neural network”, Journal of Laser Applications , Vol. 11, No 3, 143 – 150. [12] T. Iwai., K. Sadamura, F. Natumi, “Integrated Autom otive Panel Production System using Laser Welding”, Laser Assisted Net shape Engineering, Proceeding of the LANE’94, Vol.1, 135 – 144. [13] Y.M. Zhang, R. Kovacevic, L.Li, “Characterization a nd Real – time measurement of geometrical appearanc e of the weld pool”, International Journal of Machining Tools and Manufacturing, Vol. 36, No. 7, 1996 , pp. 799 – 816. [14] J. Beersiek, “A CMOS camera as a tool for process a nalysis not only for laser beam welding”, International Congress on Applications of Lasers and Electro-Opti cs (ICALEO) , 2001. 434 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 LASER PRO CESS MONITORING: A CRITICAL REVIEW [15] J. Beersiek, “New aspects of monitoring with a CMOS camera for laser material processing”, Proceedings of International Congress on Applications of Lasers an d Electro-Optics (ICALEO) , 2002. [16] J. Beersiek, T. Deverman, K. Behler, “Practical app lications of in-process monitoring for laser proces ses-not only for single welds and common materials”, Proceedings of the 23rd International Congress on A pplications of Lasers and Electro-Optics (ICALEO) , 2004. [17] H. Gu, W.W. Duley, “A statistical approach to acous tic monitoring of laser welding”, Journal of Applied Physics, Vol. 29, 556-560, 1996. [18] H. Gu, Duley. W. W., “Resonant acoustic emission du ring laser welding of metals”, J Phys.D: Appl. Phys. , Vol 29 550-555, 1996. [19] D. Farson, F and Kim, K, R, “Generation of optical and acoustic emissions in laser weld plumes”, Journal of applied physics, Vol. 85, 1329-1336, 1999. [20] P. Di Pietro, Yao Y.L., “An investigation in charac terizing and optimizing laser cutting quality”, International Journal of Machine Tools Manufacture, Vol. 34/2, pp. 225-243, 1994. [21] M.Y. Huang, C.R. Chatwin, “A knowledge-based adapti ve control environment for an industrial laser cutt ing system”, Optics and Laser in Engineering , Vol. 21, pp. 273-295, 1994. [22] G. Roggero, Scotti F., Sessa, F.S., Piuri V., “Qual ity analysis measurement for laser cutting”, IEEE International Workshop in virtual and intelligent measurement sys tems, Budapest, Hungary, May 19-20, pp. 41-44, 2001. [23] P. Poprawe, “Modelling, monitoring and control in h igh quality laser cutting”, Annals of the CIRP , Vol. 50/1, 2001. [24] D. Su, I. Norris, C. Peters, D.R. Hall, J.D.C. Jone s, “In-situ laser material process monitoring using cladding power detection technique”, Optics and Laser in Engineering , pp. 371-376, 1993. [25] B.S. Yilbas, “Experimental investigation into CO2 l aser cutting parameters”, Journal of Materials Processing Technology, Vol. 58, pp. 323-330, 1996. [26] S.L. Chen, “In-process monitoring of the cutting fr ont of CO2 laser cutting with of-axis optical fibre ”, International Journal of Advanced Manufacturing Technology, Vol.13, pp. 685-691, 1997. [27] D. P. Hand, C. Peters, F. Haran, J. D.C. Jones, “A fibre-optic-based sensor for optimization and evalu ation of the laser percussion drilling process”, Material Science Technology, Vol. 8, pp. 587-592, 1997. [28] C.E Yeack, R.L. Melcher, H.E Klauser, “Transient ph otoacoustic monitoring of pulsed lased drilling”, Applied Physics Letters , Vol. 41/11, 1982. [29] G. Chryssolouris, P. Sheng, “Process control of las er grooving using acoustic sensing”, Transactions of ASME, Vol. 113, pp. 268-275, 1991. [30] G. Cryssolouris, P. Sheng, “Investigation of acoust ic sensing for laser machining processes-Part 1: La ser Drilling”, Journal of Materials Processing Technology, Vol. 43, pp. 125-143, 1994. [31] G. Cryssolouris, P. Sheng, “Investigation of acoust ic sensing for laser machining processes-Part 2: La ser grooving and cutting”, Journal of Materials Processing Technology, Vol. 43, pp. 145-163, 1994. [32] T. Kurita, T. Ono, N. Morita, “ Study on numerical analysis of the frequency characteristic of laser p rocessing sound”, Journal of Materials Processing Technology, Vol. 101, pp. 193-197, 2000. [33] T. Kurita, T. Ono, T. Nakai, “A study of processes area monitoring using the strength of YAG laser pro cessing sound”, Journal of Materials Processing Technology, Vol. 112, pp. 37-42, 2001. [34] L. Grad, J. Mozina, “Laser pulse influence shape on optically induced dynamic processes”, Applied Surface Science, Vol. 127-129, pp. 999-1004, 1998. [35] L. Li,, D.J. Brookfield, W.M. Steen, "Plasma charge sensor for in-proces, non-contact monitoring of th e welding process", Measurement Science & Technology , Vol. 7, pp. 615-626, 1996. [36] R. Poprawe, H. Weber, G. Herziger: Editors, “Laser Physics and Applications”, (Chapter 2.8: Process mo nitoring and closed-loop control”, W. Weismann), Sub-volume C: Laser Applications”, Springer-Verlag Berlin Heid elberg New York, ISBN: 3-540-00105, (2004). [37] H.K Tonshoff, A Ostendorf, K. Korber, O. Hillers, “ Principles of reference-free process monitoring for laser material processing based on a multiple sensor system”, Proceedings of International Congress on Applicatio ns of Lasers and Electro-Optics (ICALEO) , 2001. [38] D. Farson, Y. Sang, A. Ali, “Relationship between a irborne acoustic and optical emissions during laser welding”, Journal of Laser Applications , Vol. 9, pp. 87-94, 1997. [39] D. Farson, K.R. Rim, “Generation of optical and aco ustic emissions in laser weld plumes”, Journal of Applied Physics, Vol. 85/3, pp. 1329-1336, 1999. 435 436 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L A NOVEL ARCHITECTURE FOR A RECONFIGURABLE MICRO MACHINING CELL R. Al-Sharif , C. Makatsoris *, S. Sadik School of Engineering and Design, Brunel University , West London, UK im06rra@brunel.ac.uk , *Harris.Makatsoris@brunel.ac.uk , im06sss@brunel.ac.uk Abstract There is a growing demand for machine tools that are specific ally designed for the manufacture of micro-scale components. Such machine tools ar e integrated into flexible micro-manufacturing systems. Design objectives for su ch tools include energy efficiency, small footprint and importantly flexibility, with t he ability to easily reconfigure the manufacturing system in response to process requirements and product demands. Such systems find application in medical, p hotonics, automotive and electronic industries. In this paper, a new architecture for a reconfigurable micr o manufacturing system is presented. The proposed architecture comprises a micro man ufacturing cell with the key design feature being a hexagonal-base on which three tool head s can be attached to three of its sides. Each such machine-tool head, or process ing module, is able to perform a different manufacturing process. These tool heads are interchangeable, enabling the cell to be configured to process a wide range of components wi th different materials, dimensions, tolerances and specification. Additional components of the cell include manipulation robots and an automated buffer unit. Such cells can be integrated into a manufacturing system via a modular conveyor belt to transfer parts from one cell to another and into assembly. A key consideration of the architecture is a control system that is also modular and reconfigurable; such t hat when new processing modules are introduced the control system is aware of the change and adjusts accordingly. Further to this coordination, issues betwe en modules and machining cells are also considered. Other design considerat ions include work-piece holding and manipulation. This paper provides an overview of the architecture, the key d esign and implementation challenges as well as a high level operational p erformance assessment by means of a discrete event simulation model of the micro factory cell. Keywords: micro manufacturing, reconfigurable manuf acturing, flexible manufacturing system. 1.0 Introduction 437 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L The idea of producing a flexible Microfactory cell started to arise due to the needs of producing smal l machined parts using more efficient manufacturing s ystems and techniques. Therefore, it was necessary to come up with platforms that were suitable for produ cing micro-size parts. This idea, which developed d uring the 1990s, had several advantages, such as better u se of resources including time, energy and space [1 ]. Besides this, micro-factories can be represented as fully automated units, which will result in higher precision and productivity levels, and less human involvement during any process. The concept of micro-factory h as become essential in a number of industries that req uire a high level of precision and detail such as semiconductors, microprocessors medical parts (hear ing aids) and electronics based industries. Due to the increasing dependence on consumer electronics and I T peripherals, these sectors are expected to contin ue to dominate micro-technology while other areas like au tomotives will see a significant decrease. This concept considers the possibility of offering more than one process or function to be conducted u sing a micro-factory unit, and since manufacturing systems usually face changes in functions and production methods, it was necessary to develop the current co ncept of micro-factory to satisfy these changes and configurations by creating reconfigurable micro-fac tory platforms and modules that could be adjusted i n order to perform more than one functionality and producti on capacity [2]. These recent developments in the m icro- factory concept can provide a wider range of produc ts that can be produced by only one platform, which is also a cost effective process because fewer resourc es will be consumed during each process. In this pa per, an overview of the architecture will be presented, fol lowed by a detailed description of the system parts . Then, an operational performance assessment will be addresse d in order to validate the design properties. 2.0 Architecture Overview A new concept of microfactory is presented in this paper, based on satisfying certain objectives, incl uding designing a novel architecture of an easily-reconfi gure machining cell that is capable of processing a nd handling a wide range of micro-component materials within a small footprint and with more energy efficiency. The purpose of designing such a syste m is to increase the productivity level by performi ng several machining processes simultaneously, reducin g the set-up time of machining tools, and reducing the material handling process as well. 2.1 Proposed Architecture The proposed architecture consists of three key com ponents: a machining module, a material handling platform and the control system (Figure.1). These i ndividual components will work in collaboration wit hin the system to deliver the final product. With regards t o the machining module, this was designed to have a hexagonal-shaped body and three tool-heads attached to each side. The body is based on a similar shap e rotating base which has three workpiece holding fix tures fixed to all three of its sides. This module is responsible for holding raw materials and performin g the required machining processes on them. The mat erial handling platform consists of a number of units suc h as: the cylindrical robot-arm, a buffer unit, and a material transfer belt. This platform transfers micro-compon ents to the machining modules as raw materials, and places them using the robot-arm, onto the holding fixtures , enabling the tool-heads to perform the required machining processes on them. Afterward, the same ar m will pick up the workpiece, as finished goods, in order to place them on the transfer belt which will move these materials to the buffer unit. All these activ ities will be managed by a dedicated control system in order t o obtain an improved work environment. This archite cture has a footprint of 2300 mm (w), 1190 mm (h), which means that all the system’s components will fit wit hin this area. 438 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L Fig.1. General view of the microfactory cell, showi ng main components and dimensions in millimetres (m m). 3.0 Key Design and Implementation Challenges During the design stage of this system, several des ign and implementation issues have been taken into consideration including: the effect of operating th ree tool-heads simultaneously, maintaining the lowe st level of vibration during each machining stage, and chang ing tool-heads smoothly. Each one of these issues h as been resolved using designing, operating and contro lling approaches. Fig.2. General and detailed views of the proposed m achining module, including dimensions, in millimetr es (mm). 439 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L 3.1 Hexagonal Body Fig.3. Top and general views of the module’s hexa gonal body. Starting with the machining module, which represent s the core of this system (Figure.2), the main body is designed to have a hexagonal shape (Figure.3). More over, the reason for choosing a hexagonal-shape is because it allows the fitting of three tool-heads o n its outer sides. Also, it provides increased weig ht distribution and balance between the module’s parts , since the module’s body has a symmetric design, w hich means that all tool-heads will have the same design and properties, and the distance between each tool -head and the other will be precisely the same. Therefor e, any physical contact between the tool-heads will be avoided. Moreover, this body contains vibration iso lation gaps which separate the three tool heads fro m each other. The purpose of this feature is to increase t he isolation level between tool-heads in order to i mprove the level of accuracy in the system. 3.2 The Module’s Base The previous hexagonal body is based on a similar h exagonal base with three fixtures attached to its s ides. This base represents the moving part of the module which rotates in order to place each fixture under one of the tool-heads (Figure.4). Also, the hexagonal part of the base contains a damping system which is use d to reduce the vibration of the module’s body during ea ch machining stage. The attached three fixtures hav e been designed based on a modular concept which increases the system’s flexibility and reduces the required set-up time. The design of these fixtures allows jaws and clamps to move automatically to hold the workpiece. This technique is crucial in this cell due to the variet y of the workpiece design, size and material. Fig.4. Views of the base unit and material holding fixtures 3.3 Tool-Heads The three tool-heads in this module share an identi cal BASE structure including design, dimensions and material. However, the machining tools will vary d ue to the machining nature of each process. Accordi ng to the module’s re-configurability, this design should allow the operator to change each one of these too ls with another new tool. The module’s body has been design ed to provide better physical and electrical contac t with each tool, it contains two rectangular cavities tha t allow the tools base unit to slide and connect to the module’s body. The circular connector between the t wo cavities is an air-suction unit, while the third part is a rectangular power connector. The mechanism of this assembly process is simple; first, the tool will sl ide into the body's two assembly paths that have a similar s tructure which matches the body’s features. Then, a s soon 440 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L as the power connector in the module is connected t o the power socket in the tool, the suction unit wi ll work using a vacuum-mechanism to guarantee that both par ts are well-connected and they can perform the designated process. The same procedures can be appl ied on the other two tools in order to have the fin al completed module. 3.4 Machining Processes In order to present a wide range of produced parts, three different processes have been chosen to be p erformed in this Microfactory system: Micro EDM-Milling, Mic ro EDM-Drilling, and Laser Machining. The reason fo r choosing these processes can be justified; Electric al Discharge Machining can be miniaturized and fitt ed to Microsystems due to their simple mechanical setup a nd design [3], which can provide high efficiency an d space saving opportunities. Moreover, Micro-EDMs ha ve a number of advantages over other machining processes, such as cutting and drilling, because th ey are a non-contact machining technique using ther mal energy like plasma, allowing it the capability to p roduce high precision products with much less tool breakage problems. The main concept of this technique is mac hining complex shapes using high speed rotation of a simple shaped electrode [4]. Laser technology has always been capable of providing top-class machinin g on small scale, due to the wide range of its applicati on, such as engraving and surface finishing [5]. Th is technology can also be used with several types of m aterial such as metals, ceramics, polymers and sili con. 4.0 Design Analysis of Machining Module An initial dynamic FEA model has been developed to examine the dynamics of the machining module. Several inputs have been assumed at this stage of d esign: Motor speed (3000 rpm or 50 Hz), module’s material (Granite body, Steel tool-heads, and cast iron base), and Base Damping (2%). At this stage, we assumed that the damping level of the base is equal to natural damping in order to examine the limits of this design. Based on this analysis, six natural frequen cies have been observed: 121.6 Hz, 125.2 Hz, 128.3 Hz, 210 Hz, 212 Hz and 234.7 Hz (values higher than 234.7 H z have been ignored due to their insignificance to the design). Figure 5 shows the reaction of the module during each one of the above natural frequencies. Fig.5. The module’s reaction to six natural frequen cies. Moreover, based on these natural frequencies, both dynamic displacement and stress have been calculate d. Additionally, according to these calculations (figu re.6), two peaks have generated on frequencies (125 .2 Hz) and (234.7 Hz). However, the module has an acceptab le dynamic structure since both levels are low in 441 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L considering the damping level of the hexagonal base . This is an important step in the design process due to the level of accuracy required during the machining stage. 5.0 Material Handling Platform The other main part of the system, which is the pla tform, is in charge of material handling within thi s system. Furthermore, the control system will be considered as part of this platform due to its position in the system’s layout. 5.1 Control Unit The control unit will be mainly responsible for con trolling the overall movement and function of the m achine. This includes: activating the different parts on th e cell that include the pumps and motors, controlli ng the mechanical parts in the system including fixtures, Grippers, and movement/operation of the tools. Conn ecting this unit to a database will be essential in monito ring the system’s performance and use of the collec ted data to improve the system in the future. Fig.6. Dynamic displacement and stress Vs. Frequenc y. 5.2 Material Transfer Module An automated mechanism is needed to be designed to transport material from one machine to another. Fro m the mechanism’s viewpoint, the material will be tra nsported using a standard pallet. Thus, the simples t method would be the use of conveyor belts with appropriate width to accommodate the pallets. The module (Figu re.7) would be built in standard sizes that could then be connected to each other in the desired layout. 5.3 Buffer Module This module is aimed to temporarily store work-in-p rogress (WIP) by acting as a buffer between cells. The cell layout is supposed to be linked with a number of other similar cells that will provide the comple te process line for mass manufacturing. It is expected that so me machines will perform slower than others, or tha t there will be different components that will require diff erent processes. This means that one cell would nee d to handle more than one pallet at a time in order to f ree the conveyor belt for other passing pallets. Fo r this 442 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L Ce l l Pa l l e t Arri v e s a t Tr ue F a l s e Ce l l ? Re q u i re s Pro c e s s i n g i n M o v e d t o Ne x t Ce l l W IP Sto ra g e Pro c e s s i n g T o o l 2 T i m e Pa l l e t P ro c e s s i n g Va ri a b l e s At tri b u te s a n d T o o l s F re e ? Tr ue F a l s e T o o l s Bu s y T o o l s Av a i l a b l e Ro b o t PP 1 Ro b o t PP 2 T o o l 1 Re q u i re d ? Tr ue False Pro c e s s i n g T o o l 3 Pro c e s s i n g T o o l 1 T o o l 2 Re q u i re d ? Tr ue False T o o l 3 Re q u i re d ? Tr ue False Be l t o n to Co n v e y o r Pa l l e t Pu s h e d Be l t f ro m Co n v e y o r Pa l l e t Pi c k u p Bu f fe r Ba s e o r Dro p p e d o n fo r Pro c e s s i n g Pa l l e t Re l e a s e d 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 reason this module has been designed (Figure.8) to temporarily store WIP and retrieve it when the proc essing module is available, then transfer finished pallets back onto the conveyor belt. Fig.7. Conveyor Belt Module Fig.8. Complete Buffering Module 5.4 Cylindrical Robot-Arm A cylindrical robot is employed, as shown in Figure .9, to pick and place components from the pallet to the holding fixtures, and vice versa. The robot chosen for this application is a simple ‘cylindrical robot ’ which is usually a custom made item. The one shown in the f igure below is merely a structure to show its posit ion and the robot’s main components. Fig.9. General and Top view of the Robotic-arm posi tion 6.0 Operational Performance Assessment Fig.10. Simple model of the material handling platf orm using ARENA simulation. 443 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L Following the completion of the design it was impor tant to perform an initial test in order to examine operational performance of the cell. A simulation model using ARENA [6] has been developed (Figure.10) to offer a simple demonstration of the machine’s behav iour, since all timings are estimated values, as th e actual machine is just a design that hasn’t been manufactu red yet (Table. I). However, the simulation provide s some idea of what parameters the machine will run on and most importantly an estimated pallet processing ti me. The aim of the simulation is mainly to understand t he operational performance of the cell. This is mea sured in terms of queuing and utilization of resources. In figure 11, the work in progress (WIP) buffer is sho wing the highest utilization. Also, queues were only observe d by the simulation model in the buffer and no othe r queues were observed in the system. This initial re sult confirms that there is no other build-up of qu eues anywhere else in the system, and enables simpler op erational control of the cell as scheduling needs o nly to be concerned with controlling the queue in the WIP buf fer. The second most utilized are the tools as the y are also always in operation, but only handle one palle t at a time. The other processes show much less uti lization since they are used much less relative to the WIP s torage unit and tools. Results from the simulation also show that on average a pallet takes 6.5 minutes to exit the cell, recording a maximum time of 10.2 minutes and a minimum of 3.9 minutes. Results are in the expected range given the total lengths of time for processi ng, storage and transportation entered. Moreover, this run also shows that WIP storage is the only unit in the system that stores pallets. TABLE I. Input Parameters Resource Input Units Pallet arrives at cell Constant distribution {0.7} Min per Pallet Pallet pickup from belt Triangular distribution {0.05,0.08,0.08} Min per Pallet WIP storage Constant distribution {0.2} Min per Pallet Pallet released for processing Constant distribution {0.3} Min per Pallet Component placed on fixtures by Robot-arm Triangular distribution {0.3,0.4,0.5} Min per Pallet Total process time Triangular distribution {2.5,3.1,3.7} Min per Pallet Tool 1(Micro EDM-Drilling) Triangular distribution {0.3,0.5,0.7} Min per component Tool 2 (Micro EDM-Milling) Triangular distribution {0.7,0.9,1.1} Min per component Tool 3 (Laser Machining) Triangular distribution {1.5,1.7,1.9} Min per component Component picked up from fixtures by Robot-arm Tria ngular distribution {0.3,0.4,0.5} Min per Pallet Pallet pushed onto belt by pushing unit Constant distribution {0.5} Min per Pallet Fig.11. Resource utilization graph 7.0 Conclusion and Future Work In this paper, we presented the initial design of a reconfigurable micromachining cell and discussed t he design considerations of its key components. These include a hexagonal-base with modular fixtures, interchang eable 444 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel university, UK, 9-11 th September 2008 A NOVEL ARCHITECTURE FOR A RECO NFIGURABLE MICRO MAC HINING CEL L toll-heads for performing a variety of processes, w ork-piece manipulation and work in progress storage components, and an adaptable control system. Flexib ility, re-configurability and a small footprint hav e been the key design goals considered in this development . This entails that various complements of our cell to be assembled together in a variety of combinations and different layouts into a complete micro-factory, a nd reconfigure these cells by adding or replacing proc esses to accommodate production demands as these change. Our approach in this development is iterative. In t his paper, focus was on the presentation of the con ceptual architecture design of our micro-manufacturing cell , including the results of our preliminary dynamic FEA analysis of the processing module and the operation al performance of the whole cell with simulation. T hese results confirmed several of our design goals and a llowed further progress of this development. The ne xt step in our development is to elaborate this architectur e more with detailed design specifications, includi ng detailed FEA modelling of the cell. Then, we aim to build a prototype based on that specification. Fur ther considerations that must still be addressed include the control system, which itself must also be dyna mic and flexible entailing a “plug and play” approach, and the design of fixtures. References [1] Okazaki, Y., Mishima, N., Ashida, K., (2004) “Micro factory – Concept, history and developments”, Journal of Manufacturing Science and Engineering, Vol. 126, No. 4, pp 837-844 [2] Ashida, K., Mishima, N., Maekawa, H., Tanikawa, T., Kaneko, K., and Tanaka, M., 2000, Development of d esktop machining Microfactory, Proc. J-USA Symposium on Fl exible Automation, pp. 175–178. [3] Beltrami, C., Joseph, R., Clavel, J.-P., Bacher., S . Bottinelli, (2004) “Microand nanoelectric-dischar ge machining,” Journal of Materials Processing Technology, Vol. 149, issue 1-3, pp. 263-265,. [4] Lim, H.S. Wong, Y. S. Rahman, M. Edwin Lee, M.K. (2 003) A study on the machining of high-aspect ratio micro- structures using micro-EDM, Journal of Material Process Technology. 1 40 , pp. 318–325 [5] Pham, D. T., Dimov, S. S., Ji, C., Petkov, P. V., D obrev, T. (2004) “Laser milling as a ‘rapid’ micro manufacturing process”. Proc. Inst. Mechanical Engineers, Journal of Engineering Manufacture, 21 8 Part B: p. 1 – 7. [6] Rockwell Automation, (2005) Available: http://www.a renasimulation.com/products/basic_edition.asp. [Cit ed on 20 June 2005]. 445 446 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING IMPACT OF HEAT TRANSFER ON FEMTOSECOND LASER MICROMACHINING H. Song 1, E. Miao 1,2, Y.L. Tu 1* 1. Department of Mechanical and Manufacturing Enginee ring, University of Calgary, Calgary, Alberta, Canada T2N 1N4 *.Tel. 1 (40 3 ) 22 0 41 4 2 ; fax: 1 (40 3 ) 28 2 84 0 6 . paultu@ucalgary.ca 2. School of Instrument Science and Opto-Electronic En gineering, Hefei University of Technology, Hefei, Anhui, China Abstract Femtosecond laser cutting on glass is investigated using 150 fs p ulses with a center wavelength of 800 nm. Experiments have revealed that heat transfer in the stuff resulted from laser irradiation has to been considered to im prove the manufacturing quality of femtosecond laser cutting. In this research, multi ple surfaces were manufactured by cutting grooves side by side on glass with a ser ies of pulse energy levels, respectively. By analyzing the characteristics of diffe rent surfaces, it has been found that there exists a minimum spacing between two adjace nt groove cuttings in the manufacturing of a surface. If the spacing is less than t he minimum one, the sidewall between two groove cuttings will be melted by the heat coming from laser energy. The minimum spacing between two groove cuttings for pr eventing groove sidewalls from being melted is investigated as a function of pulse energy level and feed rate. Keywords: Femtosecond laser micromachining, Groove cutting, Minimum spacing, Heat transfer. 1.0 Introduction Laser-based machining of materials with femtosecond light pulses has attracted considerable attention in recent years. The development of ultrafast laser sources has simulated interest in the application of sub- nanosecond lasers for modification of semiconductors [1], [2]. With rapid technological advances in ultrafast laser technology, the tremendous potential of femtosecond laser (hereinafter, FS laser for short) pulses as a material processing tool has been demonstrated by many research works [3]-[8]. Nowadays, ultrafast laser technology is entering the industrial market and there is an increasing need for systematic analysis and characterization of a various materials machined by FS laser. 447 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING The research on application of FS laser micromachining falls into two categories. First, functional relationships between the ablation rate and laser parameters, typically including pulse energy, feed rate, spot size, pulse duration, and laser wavelength are established for different kinds of materials. Second, micromachining methods for different form 2D or 3D features, such as surface, hole, blind hole, counter bore, slot, step, and so on, are to be developed. The research in this paper is concerning the second aspect. One of the advantages of FS laser pulse is that the thermal diffusion is limited. The energy of FS laser is deposited at a time scale much shorter than both the heat transfer and the electron-phonon coupling (typical heat diffusion time is in the order of nanosecond to microsecond time scale whereas the electron-phonon coupling time of most materials is in the picosecond to nanosecond) so that the light-matter interaction process is essentially frozen in time and the affected zone altered from solid to vapor phase and to plasma formation almost instantaneously. As a result, FS laser machining reduces collateral damages to the surroundings and the quality of the machined surface is significantly increases. In our experiments, it has been found that stuff melting caused by thermal diffusion has considerable impact on the manufacturing quality of surface under some condition in FS laser machining. To the best of our knowledge, there is no research work concerning that phenomenon and its physical mechanism so far. In this research, we found that there exists a minimum feasible spacing between two neighboring grooves when manufacturing a surface. The sidewalls of a groove will not melt until the spacing between two neighboring grooves is less than the minimum one when given other manufacturing parameters. The numerical relationship between the minimum spacing and laser energy, pulse frequency, and feedrate is investigated in this research. 2.0 E xperimental Setup A femtosecond laser micromachining system has been built in University of Calgary. Fig. 1 shows the architecture of the system. It includes two modules: laser system (Spectra Physics, Spitfire), micromachining workstation and its machining controllers (AeroTech). The micromachining workstation has to be installed on the heavy granite platform because of its very high resolution stage. The laser is a commercial 1~ 100 0 Hz amplified Ti: sapphire laser system which produces polarized ~ 130 fs pulses with a peak wavelength at 800 nm. The mechanical shutter is controlled by a computer to select the number of laser pulses. The maximum power energy is 500 mw. The micromachining workstation includes a stage with computer-controlled x and y translation, and x , y, and z rotation (≤ 5o) stage and a z translation axis on which four objectives are mounted: 5×, 10×, 50×, and 100×. The step resolution on xy translation of the stage is 10 nm. The far-field intensity distribution after the objective is a good approximation followed a Gaussian distribution. The Gauss laser irradiates from a lens that is settled on Z axis wh ose step resolution is less than 0.1 um. A vacuum chamber is installed on the stage. The controller is used for adjusting power energy and driving the stage and z axis manually or by CNC programming. The machining process is monitored on-line with a confocal CCD camera. 448 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING Glass samples were placed inside a chamber which is mounted on a stage, respectively. The machining experiments on glass were performed with the femtosecond laser system. The laser was focused on the sample by a 10× and 50× microscope objectives, respectivel y. The spot size ω0 (10 µm, spatial radius at 1/e2 of peak intensity) on each sample was determined by measuring the diameters of single-pulse craters as a function of pulse energy, then by fitting the data to the appropriate equations as described previously. Due to the laser energy fluctuations and uncertainty in the spot size measurements, an uncertainty in the fluence measurement of approximately ±20% is caused. Unless otherwise specified, the laser polarization was chosen to be perpendicular to the translation direction. After machining, the sample was cleaved perpendicular to the grooves near the middle of the grooves. The sample was then analyzed using scanning electron microscopy (SEM). 3.0 Pulse Energy The single pulse has a Gaussian spatial and temporal profile (diffraction and Fourier transform limited) [9], )/2exp()/2exp(),( 222020 τω trItrI −−= (1) Fig.1. Femtosecond laser micromachining system Fs Laser Laser control software Laser controller Micromachining workstation CCD cameraObjectives Z axis Stage Chamber Machining controller Machining control software Granite platform 449 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING where 0I denotes the peak intensity of the single pulse; 0ω signifies the spatial radius at the (1/e2) intensity contour; τ expresses temporal radius (150 fs in this research); r represents the radial coordinate of distance from the propagation axis; t is the time variable. The spatial distribution of the energy fluence is formulated as follows )/2exp()/2exp( 2 ),()( 20202020 ωωτ pi rrIdttrIr −Φ=−==Φ ∫+∞ ∞− (2) where Φ0is the peak fluence at the center of the single pulse beam. By comparing formula (5) to Gaussian distribution, the following formula can be obtained d 6 120 ≈= σω (3) where d represents the diameter of a laser spot, σ is the standard deviation of Gaussian distribution. In this research, the diameter of the laser spot is about 10 µm, so ω0 is about 1.67µm. By integrating Equation (2) on the spherical domain of laser spot shown in Fig. 3, the relationships between the peak intensity, I0, and the pulse energy, denoted by E, is obtained as follows 2/33 00 2/33 00 )2/()2/(2 piωpiωτ pi Φ== IE (4) 4.0 E xperimental Procedure The surface is machined by cutting grooves side by side with the same feeding direction. The spacing between two adjacent grooves has significant impact on the micromachining quality. The following phenomenon was found under given power energy and feed speed: if the spacing between two adjacent grooves is less than a fix value, the sidewall between them is melted obviously and the molten material flows into the sibling groove so that the groove is partly filled in. To reveal the relationships between the minimum spacing of two adjacent cuttings and the laser parameters in surface micromachining, different groups of grooves were cut with different power energy and feed rate to find out the corresponding minimum spacing. To analyze the experimental results, the relationship between power energy and the diameter of the craters created by laser pulses are also investigated. 5.0 E xperimental Results and Discussion 5.1 Analysis of Mechanism Fig. 2 shows the front view and top view of the sample which depicts the aforementioned phenomenon in groove cutting. The sample surfaces are cleaned by nitrogen gun to remove the debris resulted from the cutting. In Fig.2a, the sidewalls between grooves can be seen. The front view of the surface is in shape of 450 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING scallops and the manufacturing quality is relative planar. In Fig. 2b, the surface is smoother than that shown in Fig. 2a, but it slopes along the cutting direction due to the melting of the sidewalls. Fig. 2. Groove cuttings a. Set of grooves with sidewalls Top view F ro nt view b. Set of grooves with melted sidewalls The energy of the laser beam is distributed spatially in accord with Gaussian rule. There exists an ablation threshold of a laser beam for a given material. This is illustrated in Fig. 3. The ablation threshold was calculated in this research. When the laser beam illuminates on the sample surface, the ablation energy, which is distributed in the ablation sphere with the density higher than ablation threshold, denoted by Φabl, deposits into the sample and contributes to removing the material which results in a crater. The energy under the ablation threshold, which is the energy of laser beam except the ablation energy, is absorbed by the sample and diffuses by means of heat transfer. Energy intensity r Ablation Iabl Laser pulse 3s 3s 0 Ablation threshhold I0 Peak intensity Fig. 3. Gaussian distribution of laser beam energy 451 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING Three cases may happen when cutting the two adjacent grooves, depending on their spacing, as illustrated in Fig. 4. In Fig. 4a, the spacing is big enough so that the impact of the first cutting on the second one can be neglected. When the spacing is getting smaller during the cutting, the heat energy resulted from the first cutting transfers to the location where the second cutting is performed and starts to impact on the second cutting so that the second groove is deeper and wider than the first one, as shown in Fig. 4b. When the spacing gets small enough during the cutting, besides the phenomenon shown in Fig. 4b, the sidewall between the two cuttings is melted and the melted material flows into the first groove. This is shown in Fig. 4c. 12 12 12 a b c d Transverse feed direction Balance of heat accumulation and dissipation 1: first cutting 2: second cutting Fig. 4. Two groove cuttings with different spacing When cutting multiple grooves with the spacing described in Fig. 4b or Fig. 4c, the accumulation and dissipation of the heat energy along the transversal direction (perpendicular to the cutting direction) will ultimately reach a balance at a groove cutting under given power energy and feed rate. As a result, the following cuttings bear the same depth. This is shown in Fig. 4d. Three cases of two adjacent craters, which are similar to the three cases about two adjacent grooves along transverse direction, also happen along the longitude direction (cutting direction). They can be analyzed similarly to the above analysis for transverse direction. In this analysis, the spacing between two adjacent craters which are resulted from pulse shots plays the role like the spacing between two adjacent cuttings in transverse direction. This analysis process is omitted in this paper. After this analysis, we can see that the right view of a groove is like that described in Fig. 4d when the spacing is small enough. By integrating the analyses along two directions, the shape of cutting surface with small transverse and longitude spacing is illustrated in Fig. 5. Accordingly, the effective way to improve the micromachining quality of surface is to reduce area 1 and area 2 shown in Fig. 5. Based on above analyses, area 1 (2) can be reduced by looking for the minimum spacing between two neighboring grooves (craters) along transverse (longitude) direction to avoid the impact of heat transfer. Obviously, the spacing between grooves should increase with the increase of pulse energy. 452 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING 1 2 x y z Transverse Longitude Start point Fig. 5. Shape of cutting surface with small spacing Actually, both the transverse minimum spacing and the longitude minimum spacing are related to the parameters of laser beam when the material is given. In the next section, the study is focused on the transverse minimum spacing. 5.2 Quantitative Analysis In order to determine the relationship between parameters of laser pulse and the minimum spacing for avoiding melting the sidewalls of grooves, multiple groups of grooves were cut by a single pass of the translation stage at multiple pulse energy levels available in the power controller module of the system. The same feed rate was kept for each energy level within the corresponding set of grooves. Five feed rate levels are used, including 50µm/s, 100µm/s, 200µm/s, 400µm/s, and 800µm/s. Pulse energy is specified at 1.5µJ, 2.5µJ, 5µJ, 10µJ, 15µJ, and 20 µJ, respectively. Table 1 shows the minimum spacing between adjacent grooves under different pulse energy levels and feed rate levels (part of the experiment data). 6.0800 20.4 18.0 15.2 14.3 12.8 12.7 9.9 6.98.9 6.2400 20.1 17.5 14.8 14.1 12.6 12.6 10.1 6.88.6 6.1200 20.1 17.9 15.1 13.9 12.5 12.6 10.3 7.18.5 6.2100 20.3 17.8 15.0 14.0 13.2 13.1 10.5 7.28.6 6.150 20.1 17.8 15.4 14.2 13.2 13.1 10.2 7.08.8 1.560 50 40 30 20 15 10 2.55P (µJ/s)F (µm/s) Table 1 Minimum spacing (µm) versus pulse energy and feed rate The frequency of the laser beam is 1kHz When a pulse energy level is fixed, the minimum spacing varies little with the feed speed. For example, the minimum spacing for the pulse energy level 10mw under different feed speed is around 10µm. The minimum spacing between grooves increases when the pulse energy level is increased. 453 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING For revealing the relationships between the minimum spacing and laser pulse energy, first, we made experiments to identify the relationships between the diameter of the craters on the sample and the laser pulse energy. Table 2 shows the part of the experimental data corresponding to the data in Table 1. As can be seen by comparing the data in Table 1 to those in Table 2, the value of the minimum spacing between two adjacent grooves is approximately equal to the value of crater diameter. Two extreme cases are illustrated in Fig. 6. In Fig. 6, the diameter of the laser spot is about 10µm. 5.920.3 17.6 15.5 14.0 13.5 13.1 10.0 7.28.7 1.560 50 40 30 20 15 10 2.55P (µJ/s) Diameter Table 2 Crater diameter (µm) versus pulse energy Laser spot 2 10µm 20.3µm Pulse energy: 60µJ Laser spot 1 10µm 20.3µm ≈20.3µm Crater 2 10µm 5.9µm Pulse energy: 1.5µJ ≈5.9µm Laser spot 1 Crater 1 10µm 5.9µm Laser spot 2 Crater 2 Crater 1 Fig. 6. Shape of cutting surfaces with small spacing Fig. 7 shows the relationship between the natural logarithm of the peak intensity, ln(Φ0), and the cube of crater diameter, d3, (which is also equal to the minimum spacing between two adjacent grooves). The solid lines are semi-logarithmic fitting lines according to the experimental data. By extending the fitting line to a point, where the cube of crater diameter equals to 0, the value of ablation threshold is calculated: Φabl=0.143µJ/µm3 (I0=7.6×10 11J/cm3). It can be seen that the crater diameter increases sharply when the peak energy fluence is more than Φ0=0.469µJ/µm3 (I0=2.5×10 12J/cm3). It means that much more energy of a laser pulse goes beyond the ablation threshold when the peak energy fluence is more than Φ0=0.469µJ/µm3 than that of a pulse with the peak fluence less than Φ0=0.469µJ/µm3. 6.0 Conclusion The obvious melting phenomenon caused by heat transfer has been observed in the experiments of groove cutting by femtosecond laser. There exists a definite minimum spacing between two adjacent groove cuttings for avoiding the sidewall between two grooves from melting under a certain laser energy level. The experimental data revealed that the minimum spacing approximates to the crater diameter shot by laser pulses. The relationship between the minimum spacing and peak fluence of laser pulse were also analyzed. The result shows that the crater diameter increases sharply when the peak energy fluence is more than Φ0=0.469µJ/µm3 (I0=2.5×1012J/cm3) because much more energy of a la ser pulse goes beyond the ablation threshold. 454 The 6th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11th September 2008 IMPACT OF HEAT TRANSFER ON FEMTOSECO ND LASER MICROM ACHINING -2 0 2 0 2000 4000 6000 8000 10000 12000 d3 (µm 3 ) ln (Φ 0 ) (µJ/µm3) Fig. 7. Peak fluence of laser pulse vs. crater diameter References [1] Borowiec, A., Bruce, D. M., Cassidy, D. T., and Haugen, H. K.,2003, “ Imaging the strain fields result ing from laser micromachining of semiconductors” Appl. Phys. Lett. , Vol.83, No.2, pp. 225-227. [2] Trelenberg, T.W. (Lawrence Livermore Nat. Lab., California Univ., Livermore, USA); Dinh, L.N.; Stuart, B.C.; Balooch, S.M., 2004, “Femtosecond pulsed laser abla tion of metal alloy and semiconductor targets,” App lied Surface Science, v 229, n 1-4, pp. 268-74. [3] Borowiec, A., Haugen, H.K., 2003, “Subwavelength ri pple formation on the surfaces of compound semiconductors irradiated with femtosecond laser pulses,” Applied Physics Letters, v 82, n 25, pp. 4462-4464. [4] Borowiec, A., Haugen, H. K., 2004, “Femtosecond las er micromachining of grooves in indium phosphide,” Applied Physics A (Materials Science Processing), v A79, n 3, pp. 521-9. [5] Breitling, D., Ruf, A., Dausinger, F., 2004, “Funda mental aspects in machining of metals with short and ultrashort laser pulses,” Proc. SPIE 5339, pp. 49-63. [6] Ostendorf, A. (Laser Zentrum Hannover e. V., German y); Kulik, C.; Bauer, T.; Baersch, N., 2004, “Ablat ion of metals and semiconductors with ultrashort pulsed lasers: improving surface qualities of microcuts and grooves,” Proceedings of the SPIE - The International Society for Optical Engineering, v 5340, n 1, pp. 153-63. [7] Jiang, L., Tsai, H. L., 2004, “Prediction of crater shape in femtosecond laser ablation of dielectrics,” Journal of Physics D (Applied Physics), v 37, n 10, pp. 1492-6. [8] Crawford, T.H.R., Borowiec, A., Haugen, H. K., 2005, “Femtosecond laser micromachining of grooves in silicon with 800 nm pulses,” Applied Physics A: Materials S cience and Processing, v 80, n 8, pp.1717-1724. [9] Liu J. M. Liu, 1982, “Simple technique for measurem ents of pulsed Gaussian beam spot sizes”, Opt. Lett ., vol. 7, pp. 196−198. 455 456 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHINING OF HARDENED TOOL STEEL A Aramcharoen, P T Mativenga Manufacturing and Laser Processing Group, School of Mechanical, Aerospace and Civil Engineering, The University of Manchester, UK Abstract Micro mechanical milling plays a significant role in fabrication of miniature features in a variety of materials with capability for producing three di mensional (3D) freeform surfaces. A major challenge for micro machining of hard ened material is a high tool wear rate and unpredictable tool life. In addition, su rface roughness and burr formation are factors to be closely controlled. In this pap er, the statistical analysis of critical parameters for micro milling of hardened t ool steel is presented. The parameters included spindle speed, depth of cut, rati o of undeformed chip thickness to cutting edge radius and lubrication/environment conditions. The significance of these parameters on surface finish, burr for mation and tool wear are reported. The study shows that machining environment is the most significant factor in controlling surface finish and tool wear. While, selection of appropriate spindle speed and the ratio of undeformed chip thickness to cutti ng edge radius is more critical in limiting burr size. The work reported in this paper is important, because, industry needs to know key process variables (KPVs) that are critical in the control of micro machining performance. Additionally, the methodology enabled identification of optimum cutting values for these cutting parameters withi n the identified process window. Keywords: micro milling, wear, tool steel. 1.0 Introduction Mechanical micro machining is now recognised as a k ey technology for the manufacture of micro devices and features, especially 3D free form features. The tec hnology can be feasible for a wide variety of mater ials compared to other micro fabrication technologies. T he manufacture of micro dies from hardened steel materials is a particular case where micro milling finds application. However, in using sub-millimetre diameter milling tools, tool wear and fracture are the limit ing factors. In addition, it is also recognised tha t the 457 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL sharpness of the cutting tool in relation to the un deformed chip thickness affects the material remova l process. Very small undeformed chip thickness compared to cu tting edge radius can imply negative rake angles an d promote material ploughing instead of the tradition al shearing mechanism. The above case contributes t o one aspect of the so called “size effect”. Due to the s ize effect it is possible for the mechanism of mate rial removal to be different between macro and microscale cuttin g [1]. Given the possible departure in cutting mech anism, the fragile size of the tool and the difficult to c ut hardened workpiece material, it is essential to determine the key process variables for successful micro machinin g. It is now recognised that at a certain level of und eformed chip thickness (minimum chip thickness), th e limit for micro machining is established since this value indicates the chip formation process. Chea et al [ 2] proposed basic micro machining mechanisms based on the relative size of the undeformed chip thickness to the minimum chip thickness. If the chip load is les s than the minimum chip thickness, chips will not b e formed since the workpiece material will be compressed by the cutting tool and then recovers back after the t ool passes. When the undeformed chip thickness is large r than the minimum chip thickness, then workpiece material will be removed in the form of chips throu gh shearing mechanism. Thus, the ratio of undeforme d chip thickness to cutting edge radius needs be known in order to predict the material removal mechanism. Th is ratio influences the effective rake angle, chip for mation and chip thickness as well as the specific c utting energy [3]. One of the product attributes required for a micro device is a very good surface finish. In micro mach ining of aluminium and steel, surface finish was reported to be influenced by minimum chip thickness [1], cutti ng edge radius [4], workpiece material [5, 6] and feedrate [1]. When chip thickness is less that the cutting e dge radius, the surface finish increases due to the ploughing e ffect [5, 7, 8]. Liu et al [1] further observed tha t in microscale machining, surface finish was affected b y the trade-off between minimum chip thickness and traditional effect of feedrate. Larger cutting edge radius induce more ploughing effect and lead to po or surface finish [4, 8]. Statistical studies on the effect of tool diameter, spindle speed, depth of cut and fee d rate on surface roughness in micro machining of brass [9] a nd aluminium [10] concluded that tool diameter was the most influential factor on surface roughness of mic ro milling. However, in most applications the tool diameter may be limited by the feature size or the need to r emove material as fast as possible. Other than surface finish, in micro milling, burr f ormation is another critical aspect for micro compo nents since de-burring is more challenging and can damage micro features on the component. In slot milling, burrs occur along the side wall of the slot. These are kn own as the top burr. The mechanism of burr formatio n was proposed as the interaction between cutting edge ra dius and feed per tooth [3, 7, 8, 11]. This is due to the fact that a decrease in the ratio of undeformed chip thi ckness to the cutting edge radius results in a more negative rake angle. The material ahead the tool is pushed, bent and moved in the axial direction of the tool t o form a burr [3, 11]. Additionally it is known that down mi lling leads to larger burr size as compared to up t he milling mode [8, 12]. Tool wear is also a critical factor in micro machin ing of hard materials. Since tool wear can result i n an increase of the cutting edge radius, the ratio of u ndeformed chip thickness to cutting edge radius dec reases and this promotes burr formation [3, 8, 11, 12]. Thus, rapid wear progression can have a significant influ ence in formation of larger and undesirable burrs. 458 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL The motivation for this study was to investigate th rough analysis of variance, the dominant of key pro cess variables with respect to influencing surface finis h, burr size and tool wear when micro slotting hard ened H13 tool steel (45 HRC). More specifically, the critica l ratio of undeformed chip thickness to cutting edg e radius, which in micro machining, defines the “size effect” has in the past been neglected in the statistical evaluation of key process variables. This factor is included i n this study. 2.0 E xperimental Details 2.1 E xperimental Setup A hardened H13 tool steel was selected as a workpie ce material because it is widely used for die and m ould applications. A mid range workpiece hardness of 45 HRc was selected. This is a fine grained tool steel . Hard homogeneous microstructure is ideal choice for micr o machining as it minimises the effects of material elastic spring back after cutting and reduces the complexit y of the different mechanism of removal that may be formed if different material phase were present. Ul tra fine tungsten carbide micro tools were selected for the cutting tests. The ultra fine grain carbides provid ed the toughest shank for the milling process. The geometry was selected as two-flute micro with a flat end mil l for easy of tool wear assessment. A 500 µ m tool diameter in the mid range (1 to 999 µ m) of the micro tool size was selected. Slots with a volume of 0.18 mm 3 of material removed were machined on hardened tool steel. Before machining, new micro tools were inspected using a scanning electron micr oscope (SEM) to examine the acceptability of cuttin g edge geometry and measure the cutting edge radius. The r atio of the maximum undeformed chip thickness to cutting edge radius, rr, was then calculated for each tool. This is the on e factor in the investigation, and it relates to the size effect. 2.2 Orthogonal Array and Experimental Design The parameters chosen for study were, spindle speed (rpm), depth of cut (µ m), ratio of undeformed chi p thickness to cutting edge radius, rr, and the cutting environment. As discussed earlier the parameter rr is more critical in micro machining combining the effect of chip load and tool edge radius. The cutting parame ters of spindle speed and depth of cut needs to be set when machining. These are also traditionally considered in process optimisation. In addition the spindle speed mirrors the cutting speed effect. When H13 tool st eel was machined in macroscale milling, Ghani et al [13] re ported after Taguchi experiments that cutting speed is the most significant factor on surface finish. The cutt ing environment was selected as another factor due to the fact that friction and temperature could influence the surface finish and wear progression. In this paper the, Taguchi method was implemented b ased on an orthogonal array. The Taguchi method hel ps to economise time and cost of experimentation. The orthogonal array has been reported to be the most f lexible method for designing experiments [14]. Table I show the summary of factors and levels set for the L 9 (3 4 ) array used in this study. After the cutting test, surface finish at the bottom of micro slot and top burr wi dth along the 459 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL slot wall were measured using a Wyko optical profil er. Tool diameter reduction and tool wear were eval uated after taking measurements using a scanning electron microscope. Table I: Four factors and three levels of the exper iment Level Factors 1 2 3 Spindle Speed (rpm) A 2000 0 300 0 0 400 0 0 Depth of Cut (µm) B 15 20 25 The ratio of undeformed chip thickness to cutting edge radius C 0.4 1 2 Environment D Dry MQL Flood 3.0 Results and Discussions In the Taguchi method, the signal to noise ratio (s /n) is a measurable value that determines the varia tion in the quality of the output. The determination of the s/n ratio is based upon the aim of an investigation. I n this study, the aim is to minimise surface roughness, burr size , diameter reduction and tool wear. Thus, the small er the better was selected as the strategy to determine th e factor effect. Table II shows the s/n ratio outpu ts of surface finish (Ra), burr size, diameter reduction and tool wear from each test run. Factor effect graphs can be then plotted and the optimum level for each factor can b e found at the maximum value of s/n ratio. The anal ysis of variance (ANOVA) was used to assess the influence o f the designed parameters on the machining process. The results show the significance (in percentage) o f each factor in dominating the output objective fu nction. Table II: Signal to noise ratio results s/n ratio Tool wear Test Run Ra Burr Diameter reduction Flank Chipping 1 15.70 -34.14 -24.52 -25.63 -21.40 2 16.29 -39.27 -5.75 -23.12 -18.12 3 14.86 -28.99 -18.14 -24.86 -18.06 4 11.50 -29.05 -23.34 -27.48 -21.99 5 9.35 -27.55 -23.62 -29.06 -22.29 6 16.90 -28.60 -4.11 -21.03 -11.77 7 15.16 -27.92 -9.94 -25.15 -18.09 8 13.85 -28.11 -23.85 -27.06 -21.16 9 11.74 -30.26 -26.28 -30.33 -21.81 3.1 Surface Roughness Fig. 1 shows the factor effects on surface finish. The optimum value levels for the best surface finis h are found to be at N = 20000 rpm, ap= 25µ m, rr = 0.4 and using MQL condition. It is also interesti ng to note that the best surface finish for the H13 workpiece material occur s in the ploughing mode (at undeformed chip thickne ss less than tool edge radius). This could be expected since elastic spring back for the hardened materia l is expected to be at a minimum. Table III shows the AN OVA for surface finish. The largest value of varian ce (F) represents the most influential factor on surface f inish. Thus machining environment is the most domin ant on surface finish with at 45.56 % and followed by spin dle speed (27.75%), the ratio of undeformed chip th ickness to cutting edge radius (21.18%) and depth of cut (5 .5%) respectively. Thus in micro milling of hardene d steel the use of minimum quantity lubrication gives the b est surface finish compared to dry or machining und er flood coolant. The dominant effect of the more lubr icant oil mist on surface finish supports the need for more lubricant tool surfaces. Once this is set then the selection of appropriate spindle speed becomes the next critical process variable. 460 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL 400003000020000 16 15 14 13 12 252015 210.4 16 15 14 13 12 CoolantMistDry A s/ n ra tio B C D Fig. 1.Factor effect on surface finish Table III; ANOVA for surface finish Source Df SS MS F % A 2 14.31 7.16 1.15 27.75 B 2 2.84 1.42 0.17 5.50 C 2 10.92 5.46 0.81 21.18 D 2 23.50 11.75 2.51 45.56 Error 0 0.00 - - - Total 8 51.57 25.79 4.64 100.00 3.2 Burr Formation The size of the burrs formed on the workpiece were evaluated and Fig. 2(a) and (b) show an example of a workpiece after dry cutting at N = 20000 rpm, ap = 15µ m, rr= 0.4 as analyses by an SEM and Wyko optical profiler respectively. The burr width was measured from the Wyko images. The factor effect on burr formation, Fig. 3, displays that the optimum level in minimising burr width is at N = 30000 rpm, ap = 25µ m, rr= 2 and using flood coolant condition. The ANOVA fo r burr formation is summarised in Table IV. Spindle speed is the most effective on burr formation (51.8 2%) and followed by the ratio of undeformed chip thickness to cutting edge radius (27.96%), machinin g environment (13.18%) and depth of cut (7.04%) respectively. The results show that the higher spin dle speeds are better for reducing burr size. The h igher the rpm, the higher the cutting speed and shear rate an d these factors influence material shear and hence burr formation. In line with the expected theory the siz e effect as minored by the ratio of uncut thickness to cutting edge radius is statistically significant in control ling burr size. This is because the factor can set the threshold for minimum chip thickness and hence burr size. (a) (b) 400003000020000 -28 -30 -32 -34 252015 210.4 -28 -30 -32 -34 CoolantMistDry A s/ n ra tio B C D Fig. 2. Burrs on workpiece after dry cutting at N= 20000 rpm, ap= 15µm, rr= 0.4 (a) SEM image and (b) Wyko image Fig. 3. Factor effect on burr formation 461 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL Table IV: ANOVA for burr formation Source Df SS MS F % A 2 61.80 30.90 3.23 51.82 B 2 8.39 4.20 0.23 7.04 C 2 33.35 16.68 1.16 27.96 D 2 15.71 7.86 0.46 13.18 Error 0 0.00 - - - Total 8 119.26 59.63 5.08 100.00 3.3 Diameter Reduction The factor effect on tool diameter reduction is sho wn in Fig. 4 and the optimum level are N= 20000 rpm, ap= 25µ m, rr= 2 and MQL condition. ANOVA of tool diameter reduc tion in Table V indicates that machining environment is a most significant factor in control ling reduction of tool diameter by tool wear at 93. 20%. Again this could support the dominance of friction aspects in controlling wear performance of micro to ols. 400003000020000 -5 -10 -15 -20 -25 252015 210.4 -5 -10 -15 -20 -25 CoolantMistDry A s/ n ra tio B C D Fig. 4. Factor effect on tool diameter reduction Table V: ANOVA for tool diameter reduction Source Df SS MS F % A 2 24.88 12.44 0.13 4.06 B 2 14.32 7.16 0.07 2.34 C 2 2.49 1.24 0.01 0.41 D 2 570.97 285.49 41.09 93.20 Error 0 0.00 - - - Total 8 612.66 306.33 41.30 100.00 3.4 Tool Wear Fig. 5 shows an example image for tool flank wear a nd chipping. Edge chamfering was observed. Fig. 6 and 7 present the factor effect on flank wear and chippin g respectively. The optimum level for minimisation of flank wear and chipping are different being 20000 rpm for flank wear and 30000 rpm for chipping. The results are expected since for flank wear a lower spindle and c utting speed helps reduce wear rates. While an incr ease in spindle speed may reduce cutting forces (and can re sult in lower chip loads) and hence promote less ch ipping. The optimum level for other parameters are the simi lar being ap= 25 µ m, rr= 0.4 and MQL condition. ANOVA results for flank wear and chipping are shown in Table VI and VII respectively. The most influen tial factor on tool wear is the cutting environment with 63.5 % and 60.87 % for flank wear and chipping respectively. Again these results are in agreement with the drivers for tool diameter reduction as di scussed in section 3.3. 462 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL Fig. 5 Tool wear (at N= 30000 rpm, ap= 20 µm, rr= 2 and dry condition) 400003000020000 -24.0 -25.5 -27.0 -28.5 252015 210.4 -24.0 -25.5 -27.0 -28.5 CoolantMistDry A s/ n ra tio B C D 400003000020000 -16 -18 -20 -22 252015 210.4 -16 -18 -20 -22 CoolantMistDry A s/ n ra tio B C D Fig. 6. Factor effect on flank wear Fig. 7. Factor effect on chipping Table VI: ANOVA for flank wear Source Df SS MS F % A 2 13.372 6.686 0.750 20.08 B 2 1.576 0.788 0.070 2.37 C 2 9.362 4.681 0.490 14.06 D 2 42.295 21.148 5.220 63.50 Error 0 - - - - Total 8 67 33 7 100 Table VII: ANOVA for chipping Source Df SS MS F % A 2 4.405 2.203 0.150 4.82 B 2 21.716 10.858 0.940 23.78 C 2 9.623 4.811 0.350 10.54 D 2 55.592 27.796 4.670 60.87 Error 0 - - - - Total 8 91 46 6 100 4.0 Conclusions • The use of statistical analysis helps to identify k ey process variables in micro machining as well as establishing best process performance. • The choice of machining environment is the most sig nificant factor in improving surface finish and reducing tool wear (flank wear, edge chipping and t ool diameter reduction). • The optimum for all the investigations shows MQL to be better than dry milling or using flood coolant. The low friction coefficient vegetable oil mist use d in the MQL helps in improving machining performance. • The results show that when machining the hardened t ool steel, surface finish was best when undeformed chip thickness is less than the cutting edge radius . This is the case for the hard and homogeneous mat erial as material spring back is less prevalent compared to micro machining of ductile multiphase material. • Since the machining environment and hence friction coefficient dominate surface finish and wear progression, the idea of using soft lubricant coati ngs over a hard coating could be implemented in ord er to obtain better surface finish and also to reduce too l wear. 463 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EVALUATION OF CRITICAL PARAMETERS IN MICRO MACHININ G OF HARDENED TOOL STEEL • Burr formation can be controlled more significantly by selection of spindle speed and the ratio of undeformed chip thickness to cutting edge radius. In general increasing the spindle speed leads to lo wer burr size. The spindle speed and hence cutting spee d has a bearing on the shear rate and chip temperat ure. These parameters influence the mechanism of materia l flow and extrusion that drive burr formation. • The significant effect of the ratio of undeformed c hip thickness to cutting edge radius on burr size c an be accounted for by the size effect. References [1] X. Liu, R.E. Devor, S.G. Kapoor, K.F. Ehmann, T he mechanics of machining at the microscale: Assess ment of the current state of the science., Transaction of ASME, Journal of Manufacturing Science Engineering 126(4 ) (2004) 666- 678. [2] J. Chae, S.S. Park, T. Freiheit, Investigation of micro-cutting operations, International Journal of Machine Tools Manufacture 46(2006) 313-332. [3] K. Lee, D. Dornfeld, Micro-burr formation and m inimization through process control, Precision Engi neering 29(2005) 246-252. [4] X. Liu, R.E. DeVor, S.G. Kapoor, Model-based an alysis of the surface generation in microendmilling -Part II: Experimental validation and analysis, Transaction o f ASME, Journal of Manufacturing Science Engineerin g 129(2007) 461-469. [5] H. Weule, V. Hüntrup, H. Tritschler, Micro-cutt ing of steel to meet new requirements in miniaturiz ation, Annals of the CIRP 50(2001) 61-64. [6] M.P. Vogler, R.E. DeVor, S.G. Kapoor, On the mo delling and analysis of machining performance in mi cro end- milling, Part I: Surface generation, Transaction of ASME, Journal of Manufacturing Science Engineering 126(2004) 685- 694. [7] G. Bissacco, H.N. Hansen, L.D. Chiffre, Size ef fects on surface generation in micro milling of har dened tool steel, Annals of the CIRP 55(2006) [8] S. Filiz, C.M. Conley, M.B. Wasserman, O.B. Ozd oganlar, An experimental investigation of micro-mac hinability of copper 101 using tungsten carbide micro-endmills, I nternational Journal of Machine Tools Manufacture 4 7(2007) 1088- 1100. [9] W. Wang, S.H. Kweon, S.H. Yang, A study on roug hness of the micro-end-milled surface produced by a miniatured machine tool, Journal of Material Processing Techno logy 162-163(2005) 702-708. [10] P. Stavropoulos, K. Salonitis, A. Stournaras, J. Pandremenos, J. Paralikas, G. Chryssolouris, Exp erimental investigation of micro-milling process quality, Pro c 40th CIRP International Manufacturing Systems Sem inar (2007) [11] T. Schaller, L. Bohn, J. Mayer, K. Schubert, M icrostructure grooves with a width of less than 50 µm cut with ground hard metal micro end mills, Precision Engineering 2 3(1999) 229-235. [12] J. Schmidt, D. Spath, J. Elsner, V. Hüntrup, H . Tritschler, Requirements of an industrially appli cable microcutting process for steel micro-structures, Microsystem Tec hnology 8(2002) 402-408. [13] J.A. Ghani, I.A. Choudhury, H.H. Hassan, Appli cation of taguchi method in the optimization of end milling parameters, Journal of Material Processing Technolo gy 145(2004) 84-92. [14] P.J. Ross, Taguchi techniques for quality engi neering, 1996. 464 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT G RINDING KNOWLEDG E SHARING THROUG H A KNOWLEDG E WAREHOUSE DEVELOPMENT Asmaa Alabed, Xun Chen School of Computing and Engineering, University of Huddersfield, UK Abstract This paper reports upon a literature review of current de velopment in knowledge management (KM) and grinding technology. As a result, it sets out a framework for a research into the development of a manufacturing knowledge war ehouse that enables grinding knowledge to be shared in a manageable way. In practi ce the success of grinding is highly depending on the level of expertise and knowledge of the machinist and engineer. The existing techniques for selection of grind ing conditions and control variables are empirical methods, historical data retrieval method s, model applications and artificial intelligence (AI). Different approaches have b een implemented to select grinding conditions using Case-Based reasoning, Rule-Based reas oning, and hybrid approaches. In most cases, these methods are applied separatel y and mainly focusing on explicit knowledge processing. It is noticed that knowledge has an interesting characteri stic that the value of knowledge grows when it is shared. Knowledge sharing is a cr itical issue in any KM program. The community of practice (CoP) and the electronic network of practice (EnoP) are methods for more effective knowledge sharing. Altho ugh the shared artefact does not solve the problem of tacit knowledge sharing in a broad working environment, it can be of a real benefit and can play a variety of useful roles to support the sharing of tacit knowledge. To this end, this p aper examines feasible structures for a knowledge warehouse development regarding bot h explicit and tacit knowledges. A case study within manufacturing environment, wh ich covers grinding knowledge management, is presented in the paper. Keywords: Knowledge management (KM), knowledge shar ing, grinding, knowledge warehouse 1.0 Introduction Many factors have contributed to the growth of know ledge management such as downsizing; outsourcing; revolution in information technology and deskilling [1], [2]. According to these factors, organization s and firms feel more pressure than ever to maintain a we ll-informed workforce, boost productivity and gain competitive advantage [3]. Downsizing and outsourci ng mean a reduction in personnel. As people leave, organizations realize that they take valuable knowl edge with them [4], [5]. These factors will have a great impact on the operations that depends on the skills of people such as grinding technology. The main ob jective 465 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT of Knowledge Management (KM) is to mange knowledge process, whereas the knowledge itself cannot be managed; what can be managed is the knowledge gathe ring, storing and organizing, retrieving, and shari ng. It was recommended that an organization should have an effective and efficient information system to faci litate knowledge management process [1], [6] - [12]. The k nowledge management is often facilitated by information technology; information technology by i tself is not KM. It can be argued that the knowledg e management is more than the acquisition, analysis, storage and manipulation of data and information. I t is an attempt to recognise the human assets within the mi nds of individuals and to leverage them as organisa tional assets that can be accessed and used by boarder set of individuals. 2.0 Knowledge Creation, Sharing and Their Techniques Creating knowledge is not simply a matter of learni ng from others or acquiring knowledge from the outs ide [13]. Knowledge must be built on its own, frequentl y requiring intensive interaction between individua ls and the organisation. The most powerful learning comes from direct experience and trial and error. For exa mple, a child learns to eat, walk through trial and error. According to Nonaka and Takeuchi [13], knowledge co nversion is about the distinction between tacit knowledge and explicit knowledge and how tacit and explicit knowledge interact and interchange into ea ch other. Tacit knowledge and explicit knowledge are n ot totally separated but mutually complementary ent ities. It is important to note that the interaction betwee n tacit and explicit knowledge is performed by indi viduals but not the organisation. Etienne Wenger [14] descr ibes the “negotiation of meaning” as how people experience the world and their engagement in it as meaningful. The negotiation of meaning involves the interaction of two processes, participation and rei fication, which form a duality. An important aspect of the participation/reification duality is balance betwee n each of the constituent processes. Each needs to be in its proper proportion so that each remains in equilibri um with the other. Regarding knowledge duality theo ry, Wenger [14] consider that hard knowledge is the par t of what people know that can be articulated, capt ured, and stored and soft knowledge is the part of what p eople know that can’t be articulated. According to the view of knowledge as duality then by implication, all kn owledge is to some extent both hard and soft: it is simply that the balance between the two varieties. The explicit knowledge such as working procedures a nd related data can be easily collected, stored, re trieved and shared [5],[6],[12]. Explicit knowledge in an organization is represented in the form of databases, documents, memos, reports, best practices, or proce ss in the organization. The knowledge repository ca n be accessed at the convenience of employees and well s uited even for busy employees. The tacit knowledge is found in people’s-heads or e xperience and it is developed from direct experienc e of action. It could be shared through highly interacti ve conversion, story telling, and sharing experienc e [4], [10], [13], [14]. The most critical part of knowledge is the tacit kn owledge, such as employee skill and experience, which is in their minds. Difficulty of managing tac it knowledge is an important challenge for KM progr am [4], [5], [17] – [20]. The challenge inherent with tacit knowledge is figuring out how to recognize, g enerate, share and manage it. While Information Technology i n the form of e-mail, groupware, instant messaging, electronic database, video and audio recording, mul timedia presentations, and related technologies can help facilitate the dissemination of tacit knowledge; id entifying tacit knowledge in the first place is a m ajor hurdle for most organizations [7], [20]. The reach of taci t knowledge and experience possessed by individuals can be greatly extended once it is captured and explicated so that others can easily find it and use it [11], [20]. The problem with current technology was that ignored th e tacit aspects of knowledge [7], [11], [20]. 466 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT Globalization affects most organizations that many organizations are now undergoing some forms of structural changes and looking for new technology o f knowledge sharing to cope with the internationali zation of business [1], [2], [5]. Knowledge sharing is a c ritical issue in any KM program. New knowledge coul d be created by the generation of novel concepts through knowledge sharing. Coleman [21] described culture evolving from networ ked cultures, network applications, collaborative applications and groupware, knowledge management an d to online communications cultures. Many leading companies start with the technology networks and in frastructures and build up to collaborative applica tions [21], [22]. By using a computer network or communic ation network, people from different geographical a reas can communicate and share efforts across time and s pace and collaborate on their common targets. After network applications, many companies moved forward to use groupware to share their knowledge. Groupwar e is an umbrella term describing technologies that su pport person-to-person collaboration. Groupware inc ludes e-mail, electronic meeting, desktop video conferenc ing as well as systems for workflow and business pr ocess reengineering [5], [20], [21], [23]. Recent advance s in information and communication technologies hav e led to the emergence of online structures on which the primary purpose is knowledge exchange. These electr onic networks enable individuals to interact around a sp ecific practice, regardless of physical proximity o r prior personal acquaintance, thus eliminate the need for face-to-face meeting. [24], [25]. The community of practice (CoP) and electronic network of practice (EnoP) are methods for more effective knowledge sharing. It m akes some inroads in tackling the complexities and chall enges in the new business environment and it can be integrated with both the physical and electronic en vironment. Although the EnoP and CoP do not solve t he problem of tacit knowledge sharing in a distributed international environment, it can be of a real ben efit and can play a variety of useful roles to support the s haring of tacit knowledge [10], [14], [20] [26]. Many companies are migrating from e-mail to collabo rative culture that ensue will cause a huge increas e of the generation of knowledge. This increase in knowl edge generation will require corporate culture to e volve to the next step of managing this knowledge [22]. The table 1 shows common knowledge management technologies that have been adopted by knowledge in tensive firms [27]: Table 1. Knowledge management technology adopted [2 7] Knowledge Management Technology Adoption E-mail 100% Internet 100% Videoconferencing 100% Project management systems 91% Groupware 91% Knowledge-based systems 82% Intranet 82% Yellow Pages for knowledge 44% 3.0 Grinding Technology Grinding is a material removal process that utilise s a grinding wheel with a large number of randomly bonded grains [28]. Grinding is a widely used machining pr ocess in applications requiring high production rat es and very good dimensional accuracy and surface finish. It is also one of the complex metal machining proce sses and one of the less understood [29]. Consequently, successful of grinding in practice is highly depen ding on the level of expertise of the machinist and enginee r [30]. Grinding is also viewed as an unpredictable process because of the large number of the relationships be tween those variables of grinding and grinding proc ess performance [31]. The variables are summarised in t able 2 [29], [31]. 467 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT Table 2: Grinding Variables Variable Definition Specific Power The coefficient of friction could be measured from tF and nF relationship. So tangential force and power can be used to measure specific power. Specific Energy Power/volume of metal removed G-Ratio Volume of metal removed /volume of wheel speed Equivalent Diameter the diameter of a wheel that would be used for surface grinding Wheel dressing For non rotating diamond dressing, either with a single point or multipoint diamond, the dressing lead and depth are two important factors Wheel grade Topography of a grinding wheel Coolant effect Grinding fluids Sparkout The end of most precision grinding cycle Wheel wear Parameters The sum of attritional (grain-flat-wear) and grain-fracture wears Surface finish The surface finish of a workpiece Surface Integrity The quality of the surface on the finished The existing techniques for the selection of grindi ng variables are data retrieval method, empirical m ethod, and artificial intelligence (AI). Although the data retrieval methods are simple and practicable, the data apply only to a particular machine situation [28], [32]. Empirical model methods tend to be limited value si nce a small change in a variable that is uncontrolled can have a large effect on the models [33], [34]. Rowe et al. [32] summarized the application of Artificial Intel ligence that used in grinding technology: knowledge -based expert systems, fuzzy logic, neural networks, genet ic algorithms and adaptive control for optimization [33]. AI includes rule-based, case based, neural networks , fuzzy logic, and hybrid methods [32]. Rowe [32] presented a conceptual framework for an i ntelligent grinding system. The intelligent grindin g machine has the potential to include adaptive contr ol optimization (ACO) and the features indicted bel ow: • To remember optimized conditions for future operati on in learning database. • To provide intelligent selection of grinding parame ters from a learning database. Li et al. [34], [35] developed an approach for sele ction of grinding process condition using the black board approach. The approach has shown the ability to int egrate different intelligent technologies into one system. The knowledge agents consist of case-based reasonin g, neural network reasoning and rule-based reasonin g. Chen [36] proposed a tow-layer structure for the de velopment of a grinding knowledge management system ; it was focusing on the knowledge structures for man ufacturing technology [36]. Morgan et al [37] developed an intelligent grinding assistant system. A decisio n-making system was integrated in the database to s elect the initial grinding parameters. Although different te chniques have been investigated to select grinding conditions [20], [31] - [35 ] by many researchers and some initial experiments c ompleted, none of them has been focused on the tacit knowledge, which is an im portant element for knowledge management. 4.0 Methodology of Knowledge Warehouse Development for Grinding From the Literature review of KM, One of main KM ob jectives is to mange organisational knowledge to create new knowledge. The new knowledge is created by combining existing knowledge pieces or by generation of novel concepts through knowledge shar ing. The problem with earlier technology was that ignored the tacit aspects of knowledge. The other r ole of IT in KM is to make the implicit visible. According to the issue of the XPERTS investigation, knowledge management tools are not currently used within the European machinery-engineering domain. I t should be noticed that in this sector, the indust ries are not really aware of the benefits of using such a to ol [38] . The grinding process depends highly on skilled engineers or technicians that they can leave their jobs at anytime taking with them their knowledge an d 468 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT experience. The organisations need to find a way to keep their knowledge available when it is needed n ot only the explicit knowledge but even though the tacit kn owledge. On the other hand, grinding is still an unpredictable process if there is not suitable know ledge available and it heavily relays on experience because of a large number of variables involved and inadequ ate understandings of the relationships between the se variables and grinding process performance [29], [3 1], [32], [34], [35]. While groupware and intranet facilitate the knowled ge exchange and the variation of tremendous amount of knowledge, it is very difficult to extract the exac t knowledge efficiently from them [24] - [27]. The messages EnoP technology are not stored in a single reposito ry that can be accessed by new comers or searched f or historical information [21] - [27]. A flexible and easy to use web-knowledge based (war ehouse) is proposed here, which could manage the explicit knowledge and facilitate transferring taci t knowledge into explicit knowledge. Upon this, the new knowledge base will support the decision making pro cess for selecting grinding conditions. The new knowledge based system will encourage and facilitat e the sharing of explicit and tacit knowledge by bu ilding problem solving and question-answer modules. Data Collection Create/revise Refine Update Knowledge Acquisition Create/revise Refine Update Database Evaluate Index Store Knowledge Warehouse (RBR,CBR,MBR) Evaluate, Index & Store Problem-Solving (CBR, RBR, & data) Retrieve knowledge Consolidate knowledge Knowledge Analysis Modifications Validations Knowledge Discovery Pattern and rules Discovery Modifications CoP Emails Grinding cases Knowledge Creation Knowledge Data Securing Knowledge dissemination and retrieval Database Decision Support Questions and answers Statistical Analysis Query/Reporting Knowledge Acquisition Knowledge Analysis Cases Collection Knowledge Warehouse Learning Knowledge Discovery External Sources Grinding Cases Cases collection interface Knowledge Engineers CoPs Knowledge acquisition interface Problem-Solving Knowledge Discovery Fig. 1. Knowledge management in GKW Fig. 2. GKW Module structure A Grinding Knowledge Warehouse (GKW) is designed to facilitate and support knowledge management process for grinding technology. Figure 1 shows the knowledge management process in GKW. The GKW will include six modules as shown in Figure 2. The Grind ing Knowledge Warehouse (GKW) includes five modules as shown in Figure 1. Data Input Module col lects the data through the data collection interfac e and Community of Practices (CoP’s) members. Database Mo dule stores, retrieves, and shares data in various formats. These data will then transfer to Problem S olving Module and Learning Knowledge Discovery Module, which will provide the user guidance for se lecting grinding conditions. Knowledge Acquisition Module facilitates knowledge conversion from tacit to explicit knowledge by directly acquiring the tac it knowledge from knowledge engineers or CoP members. Also, Knowledge Acquisition Module provides the interface for CoP members and knowledge engineers. The core of Problem Solving Module includes Case Based Reasoning and Rule Based Reasoning. Learning Knowledge Discovery Module extracts implicit, previously unknown and potential useful rules and p atterns to modify and update existing rules and pat terns. The GKW will be integrated into an internet based f ramework for a wide accessibility. 469 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT 5.0 Comparison of Communications within Two Groups in Rolls-Royce The structure of grinding knowledge has been analyz ed and a framework of a grinding knowledge warehous e is constructed to facilitate knowledge management. In order to acquire an overview and awareness of th e grinding knowledge sharing activities in Rolls Royc e, several meetings and discussions were carried ou t. Community of practice is used as a sharing knowledg e tool between the employees. Statistical analysis between two different communities was carried out t o evaluate their performance. From the CoP emails of group A and B between March 2005 until November 2006, it can be noticed that th e CoP tool has been mainly used for questions and ans wers, passing documents, calling for conference or event, passing web links, sending NewsBox and others. Figu re 3 represents the percentage of different categor y activities in each group. Depending on the work int erests of the CoP members, the usage of the groupwa re is different. It has been found that 54% of the e-mail s sent by the employees in group A are used to shar e their knowledge through sending questions and answers, wh ile only 49% of the e-mails in Group B serves this purpose. Nevertheless, it confirmed that the highes t percent was for question and answer category, whi ch meant the CoP could be considered as a good tool fo r knowledge sharing. CoP in Group A Passing Files 10% NewsBox 5% others 5% Web links 3% Questions and answers 54%Calling for events 23 % Questions and answers Passing Files Calling for events Web links NewsBox others CoP in Group B Questions and answ ers 49% Passing Files 33% Calling for events 12% Web links 6% others 0% Questions and answ ers Passing Files Calling for events Web links others 10% 5 % 5 % 3 % 5 4 % 2 3 % 4 9 % 33 % 12% 6% 0% Fig. 3. Percentage of different activities within t wo communities of practice The number of emails in group A is much more than t he emails in group B, that means the CoP was used more frequently in group A than B as shown in Table 3. For group A, the number of asked questions is 4 0 questions. From these questions, 10 questions have not been answered. That means 75% of these question s have been answered. Thirty-six of these questions ( 90%) were about the manufacturing process. The tota l number of answers is forty-six answers. The answer s were classified into three categories; answers co ntain knowledge, contacts, or others. It was found that 7 5% of these answers contain knowledge, 17% of answe rs does not. There are some 8% of answers are not rele vant. It can be seen that e-mail is becoming a popu lar tool for knowledge dissemination. However the usage of s uch tool depends on difference working environment and the specialised fields. 470 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT Table 3. Number of emails communicated in group A a nd B Questions and answers Passing Files Calling for events Web links NewsBox others Total Group A 86 15 36 4 7 7 155 Group B 16 11 4 2 0 0 33 6.0 Conclusive Remark Many companies have used file servers, email and gr oupware as a collaborative tool [31], [35], [39], [ 40]. However, none of these tools are fundamentally desi gned to share knowledge. While intranet and groupwa re facilitate the creation of a tremendous amount of k nowledge, it is very difficult to extract the exact knowledge efficiently from it [39]. By using the most advance research technology, finding the right knowledge i s still very difficult. On the other hand, from the statistical analysis of this research, it has been noticed that a knowledg e warehouse could make knowledge more accessible for users, could retrieve knowledge more efficiently an d quickly, and should facilitate extracting knowledge for CoP. The CoP has been used as a tool of collab orative among the Rolls-Royce workers. It been mainly used to send questions/answers, call for a meeting or se minar, and forward files or web links. From the meeting wi th the key representative of Rolls-Royce and the an alysis of the emails, it has been convinced the importance of knowledge ware house development. The next stag e will be how to extract the knowledge effectively fr om the CoP. The new KW will encourage and facilitate the sharin g of explicate and tacit knowledge. The grinding ca ses will be kept in the knowledge warehouse that will s upport the decision making process for selecting gr inding conditions for new processes and optimization. As a result, it will save the time for CoP members by providing them with most relative answer to their q uestions. It also helps them sharing up-to-date kno wledge. The further work will develop and integrate all the modules into a system which allow users to access the knowledge warehouse via the intranet. Acknowledgement The authors wish to acknowledge the assistance and support of Rolls-Royce plc. References [1 ] Davenport, T., Prusak, L (1 9 98 ), “Working Knowledge: How Organisations Manage What They Know”, Harvard Business School Press . [2 ] O’Dell C. and Jackson Grayson C. (1 9 9 8 ), “ If we only knew what we know: identification and transfer of internal best practices”. California Management Review 4 0 (3), 154-174. [3 ] Stuart, A. (19 96 ), 5 “Uneasy Pieces”, CIO magazine, Available at http://www.cio.com/archive/060196_uneasy_1.html [Ac cessed 2005]. [4 ] Kimble C, Hildreth P., Wright P (199 9 ), Knowledge Management: are we missing something?, In books and Kimble C, Information system,the next generation pr oceeding of 4 th UKAIS conference York UK pp. 347-356, London, MC Grow Hill1. [5 ] Kimble C, Hildreth P., Wright P (2 00 1 ), Communities of Practice: Going Virtua”, In: Knowle dge Management and Business Innovation, edited by Y. Ma lhota,216-230. Hershey, PA: idea Group Publishing. [6] Denning, S. (1 9 98 ), What is knowledge management? (A background paper to the World Development Repor t 1998). www.stevedenning.com Available at: http://ww w.stevedenning.com/knowledge.htm [Accessed 2005]. [7 ] Wiig, K. (1 99 7 ), Knowledge Management: Where Did It Come From And W here Will It Go?, Expert Systems with Applications, Pergamon Press/Elsevier, 1 4 . [8] Prusak, L. (2 0 0 1 ), Where did knowledge management come from? IBM Systems Journal , 40 (4), 1002-1007. Available at http://www.research.ibm.com/journal/sj /404/prusak.html [Accessed 2005] [9] Davenport, T.H., Prusak, L., Wilson, J.H. (2 00 3 ), “Who’s bringing you hot ideas and how are you responding?”, Harvard Business Review , 8 1 (2), 59-64. 471 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 GRINDING KNO W LE D G E SHARING THRO GH A KNOWL ED GE WAREHO U S E DEVELOPEMENT [1 0 ] Bhirud S., Rodrigues L., Desai P (20 05 ), “Knowledge Sharing Practices in KM: A Case Study i n Indian Software Subsidiary”, Journal of Knowledge Management Practice., Available at www.tlainc.com/articl103.htm - 60k [Accessed April 2007]. [11 ] Parlby, D. and Taylor, R. (2 00 0 ), The power of knowledge: a business guide to knowle dge management, available at: www.kpmgconsulting.com/index.html [Ac cessed 2006]. [12 ] Greco, J. (1 9 9 9 ), Knowledge is power, March-April, available at: http://web7.infotrac.london.galegroup.com/itw/infom ark/482/356/72858976w3/purl ¼ rc1_GBIM_0_A54293713&dyn ¼ 30!xrn_10_0_A54293713?sw _aep ¼ up_itw [Accessed 2006]. [13 ] Nonaka I. and Takeuchi H.(1 99 5 ), The Knowledge-Creating Company: how Japanese compa nies create the dynamics of innovation, Oxford University Press, IS BN 0-19-509269-4. [14 ] Wenger E (19 9 8 ) , Communities of Practice learning as a Social syst em, Published in the Systems Thinker. [15 ] Hildreth P. M., Kimble C. (2 0 02 ), “The Duality of Knowledge”, Information Research , 8 (1). [1 6 ] Gillingham, H., Roberts, B. (2 00 6 )” Implementing Knowledge Management: A Practical Appr oach”. Journal of Knowledge Management Practice, 7 (1). [17 ] K PMG Consulting (200 0 ), Knowledge management research report 2000, availab le at: www.kpmgconsulting.com [Accessed in April 2007]. [1 8 ] Du Plessis, M. (20 07 ), ” Knowledge Management: What Makes Complex implementa tion Successful”, Journal of Knowledge Management, 11 (2), 91-101. [19 ] Gurteen D., (1 99 9 ), Creating Knowledge Sharing Culture, Knowledge Movement Magazine , 2 (5). [2 0 ] Santosus M.,Surmacz J. (20 01 ), the ABCs of Knowledge Management, CIO magazine,. A vailable at http://www.cio.com/research/knowledge/edit/kmabcs.h tml [ Accessed 2005]. [21 ] Coleman, D (19 9 7 ) , Groupware: Collaborative Strategies for Corporat e LANs and Intranet”, Prentice Hall. [22 ] Coleman, D . (1 9 99 ), Groupware: Collaboration and Knowledge Sharing” Ch 12 in Knowledge Management Handbook, edited by Liebowitz, Jby CRC Press LLC. [2 3 ] Allee V., (19 9 7 ) , The knowledge evolution: expanding organisational intelligence. Butterworth-Heinemann, p.231. [2 4 ] Teigland, R., Wasko, M. (200 4) , Extending Richness with Reach: Participation and Knowledge Exchange in Electronic Networks of Practice. Chapter 19 in Hildreth, P & Kimble, C (eds.), Knowledge Networks: Innovation Through Communities of Practice , London: Idea Group Inc., 2004. [25 ] Teigland, R. (2 0 03 ). Knowledge Networking: Structure and Performance in Networks of Practice. Published doctoral dissertation. Stockholm: Stockholm School of Economics. [2 6 ] Kimble, c., Li F., Barlow, A. (20 0 0) , Effective Virtual Teams Through Communities of Pr actice, University of Strathclyde, department of Management and Science. [27 ] Sena, J., Shani, A.B. (1 9 9 9 ), Intellectual Capital and Knolwedge Creation:Towrds an Alternative Framework, Ch 8 [2 8 ] Kalpakjian, S. (1 99 1 ), Manufacturing Processes for Engineering Materials, 2 nd edition, Wesley Publishing company, ISBN 0-201-60702-6. [29 ] Kegg, R. L.(1 98 3 ), Industrial Problems in Grinding, Anna’s of the CIR P, 32 (2),559- [3 0 ] Row ,W. B ., Li,Y., Inasaki,I. and Malkin, S. (1 99 4 Application of artificial intelligence in grinding. Keynotes paper,Ann. CIRP, 43(2),52 1-531 . . [3 1 ] King, R. I., Hahn, R. S. (1 98 6 ), Handbook of Modern Grinding technology , Chapamn and hall, New York . [3 2 ] Rowe, W. B ., Li, Y., Inasaki, I. and Malkin, S. (19 94 ), Applications of artificial intelligence in grinding . Keynote paper. Ann. CIRP , 43 (2), 521–531. [3 3 ] Rowe, W.B. , Allanson, D.R. , Pettit,J.A., Mouruzzi, J.L. snd Kelly,S. (19 91 ), ’Intelligent CNC for Grinding’, Proceedings of the Institute of Mechanical Engineer s, Vol. 205, pp.233-239. [34 ] Li Y., Rowe, W. B ., Mills B. (1 99 9 ), “Study and Selection of Grinding Conditions Part 1 : Grinding Conditions and Selection Strategy”, Proc Instn Mech Engrs, 21 3 (B). [3 5 ] Li Y., Rowe, W. B ., Chen, X., Mills B. (19 99 ), “Study and Selection of Grinding Conditions Part 2 :A Hybrid Intelligent System for Selection of Grinding Condit ions” Proc Instn Mech Engrs, 2 13 (B). [3 6 ] Chen, X (20 0 2 ), Systematic Consideration of Grinding Process Monit oring, Proceedings of the 8 th Chinese Automation and Computing Socety Conference in the U K, 175-178. [3 7 ] Morgan, M.N., Cai, R., Guidotti, A., Allanson,D.R. , Moruzzi, J.L. , and Rowe, W.B . , (2 00 7 ) ‘Design and implementation of intelligent grinding assistant sy stem’, Int. J Abrasive Technology, Vol.1, No. 1, pp . 106-135. [3 8 ] Azarian, A., Bueno, R., Alzaga, A (20 02 ) , Knowledge Management Requirements within Machiner y Engineering Domain, XPERTS (N 11961). [3 9 ] Baek S., Liebowitz, J., Prasad, S. Y., Granger, M . (19 99 ), Intelligent Agents for Knowledge Management, Ch 11 in Knowledge Management Handbook, edited by Lieb owitz, J, by CRC Press LLC. [4 0 ] Sharratt M., Usoro A. (20 03 ), “Understanding Knowledge Sharing in Online communi ties of Practice”, Academic conference Limited 472 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PROCESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING Lei Zhou 1, 2 , Kai Cheng 1 1. School of Engineering and Design, Brunel Universit y, Uxbridge, UK, UB 8 3PH 2. School of Mechatronics Engineering, Harbin Institu te of Technology, Harbin, China, 150 0 0 1 Abstract In nano/micro cutting process, the surface quality is heavily d ependent on all the dynamic factors in machining including those from the material, tooling, cutting parameters, servo accuracy, mechanical structure deformation, and non-linear factors as well. The machined surfaces are generated based on the tool profile and the real tool path combining with the various external and internal disturbances . To bridge the gap between the machining conditions and the surface quality , the integrated simulation system presented involves the dynamic cutting proc ess, control/drive system and surface generation module. It takes account all the intricate aspects of the cutting process, such as material heterogeneity, regenerative c hatter, built-up edge (BUE), spindle run-out, environmental vibration, and tool interf erence, etc. The frequency ratio method is used to interpret the surface t opography and texture formation. The proposed systematic modelling approach is verified by the cutting experiment. Keywords: Dynamic cutting force, non-linear factors , surface topography, tool interference. 1.0 Introduction Metal cutting, especially in nano/micro scale, is a complex process which comprises workpiece material , tooling geometry, cutting parameters, servo drive a ccuracy, static and dynamic deformations of structu re in machining, etc. The machined surfaces are generated based on the tool profile and real tool path consi der with various disturbances. It is of great benefit to bui ld a systematic model and analyze the effects from the dynamic cutting process. Many linear/nonlinear fact ors in the whole system will contribute to the surf ace quality. The dynamic cutting force joints all the f actors together, and produce the tool-workpiece rel ative displacement in the stiffness loop. 473 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING Take 3-Axis turning machine as a case in point, sho wn as Fig. 1, the stiffness loop in face cutting pr ocess involves: spindle axial runout, workpiece clamp err or, workpiece material property, cutting tool geome try, tool holder stiffness, Z axis servo stiffness in axial d irection, X axis slideway stiffness in side directi on, and mounting stiffness. In this stiffness loop, all the elements are guaranteed by physical construction l ike bearings, mounting screws and other mechanical stru ctures except the linear motor in axial direction. Since employ the direct driven linear motor, the stiffnes s in Z axial motion is directly provided by electri cal servo motor. The tuning of Z axis linear motor, in the er ror sensitive direction, is one of the most critica l ingredients of ultra-precision machine tool. Fig. 1 Stiffness Loop of 3-Axis Turning Machine Fi g. 2 Machining Process Integrated System To get the connection of the tool path and the mach ined surface, several literatures focuses the surfa ce topography generation process. Cheung and Lee [1]-[2] used the digitizing oscilloscope to record the rel ative displacement between the tool and the workpiece. Af ter been analyzed by Fast Fourier Transform (FFT), the main frequency and amplitude of vibration reconstru cted the tool deviation from ideal position. The to ol profile superimposed onto the tool path to generate the surface topography. Lee and Cheung [3]-[4] built a dynamic cutting system to predict the surface topog raphy in face turning. However the cutting force mo del only contained the linear factors and the whole cut ting system was simplified into a second-order syst em. The servo control system, material hard grain effect, c hatter [5] , tool wear, spindle runout and environmental vibration were not taken into account. Luo et al. [6]-[7] introduced some non-linear factors to the cutting process and studied the effects of machining process variab les and tooling characterizations on the surface te xture, yet the tool interference effect in nano/micro scale cu tting process had not been involved. In short, for micro/nano cutting process, the machi ned surface performance is all about how successful ly the tool-workpiece relative position meets its desired objective. The topography and texture of cutting su rface will be generated by the dynamic feed response under the multifarious cutting factors. In this paper, the dynamic cutting process system i s established. The machined surface topography is p redicted though the real tool path and tooling geometry. Fro m analyzing the cutting force, the tool-workpiece r elative displacement and the surface profile, the origins o f surface error are located to the dynamic cutting factors. The preliminary cutting trials certify the validity of the dynamic cutting system modelling. 2.0 Systematic Modelling of the Dynamic Nano/micro Cutting The generation of workpiece surface is a very compl ex material removal process as shown in Fig.2 inclu ding:  servo control dynamics, which is joined by control algorithm, servo amplifiers, actuators and inspecti on sensors; Machining Process Dynamics Modeling Surface Generation Tooling G eometry Machine Tool Performance Workpiece Property O peration Factors Control Algorithms Machining Optimization A ctuator Inspection Sensor Surface Finish / Texture / Functionality Servo Drive Environment Trajectory Generator Tool Holder X Y Z S t i f f n e s s L o o p o f M a c h i n e T o o l s 474 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING  machining process dynamics, which is highly affecte d by six main aspects: machine tool performance, cutting tool geometry, workpiece material propertie s, operation factors, servo motion and environment; and  surface generation and surface texture, integrity a nd functionality analysis. To get a better understanding of the complex proces s, a systematic model is developed in the MATLAB Simulink module as Fig. 3. The modelling approach b ridges the gap between the determinative factors an d the surface generation. Based on a thorough theoretical analysis of servo motions and cutting mechanics/dy namics, the integrated model produces a scientific methodol ogy to simulate the precision surface generation in nano/micro turning processes. Fig. 3 Dynamic Machining Process Integrated Modelli ng 2.1 Modelling of Servo Control Dynamics Most of the cutting dynamics simulations are simpli fied model the control and drive system as a second -order system. However, the advanced control algorithms ar e always applied on the ultra precision machine too ls to achieve higher servo accuracy. Modelling the contro l system can also help to modify the controller construction and finally improve the surface qualit y. In this paper, the ultra precision turning machi ne employs the UMAC motion controller who provides a PID posit ion/velocity servo control scheme with velocity and 475 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING acceleration feedforward loops [8] . The discrete model uses Z-transforms whose sampli ng period is the real servo-interrupt cycle. The position quantification based on the encoder resolution is 4.88 nm. The output from the UMAC controller is lead to serv o amplifier to continue the current loop. The direc t-drive linear motor model is built on the resistance and i nductance of the motor windings, force constant and back EMF constant from the specifications. In this case, the static air bearing slideway is applied to redu ce the friction and unexpected load in the motion directio n. Therefore, the plant of the slide carriage is si mplified represent as a pure mass module. From servo control dynamics Model, the actual motio n of each slideway is calculated for the next stage to generate the real tool path. In fact, there are som e additional disturbances in the drive system: elec trical noise in the current loop and cutting force disturbance s uperimposed on the sideways which is generated by d ynamic cutting process model. 2.2 Dynamic Cutting Process The quality of the machined surface is mainly deter mined by the relative motions between the cutting t ool edge and the workpiece. In practice, the tool path deviates from the ideal tool path as a result of ki nematic or dynamic factors from the dynamic cutting process, m achine tool motion errors and environmental disturbances. For instance, the deformation of the slideway bearings under the action of dynamic cutti ng forces will deflect the cutting tool nose point; environme ntal vibrations will make the motion of the cutting tool away from the designed tool path. Since the surface topography model basically follow s the reflection of cutting tool edge profile, to c alculate the real tool path becomes the critical part in the modelling approach. Table I lists the linear and n onlinear factors from the structural deformation and motion errors of the machine tool and cutting tool which w ill contribute the deviation of the tool path. Table I: Linear and Nonlinear Factors in the Cuttin g Process Source Influence Factor Mathematical Function BUE Rake Angle ∆α = αA Rse( ωαt) (1) Hard Grain Shear Stress ∆τ1 = τA Pul( t) (2) Coolant Friction Angle ∆u=r (t) (3) Flank Wear Tool Wear )exp( f s r t w RT EBV Vf F H Al −+= [9-10] (4) Regenerative Vibration Feed and Depth of Cut Chatter ∆ct = Z (t) - Z (t-T) (5) ∆cw= X (t) - X (t-T) Spindle Runout Es = As sin( ωt+ φ) (6) Slideway Stiffness es= Fs / ks (7) Environmental Vibration Displacement between Cutting Tool and Workpiece Ev = Aev sin(2 pifevt) (8) where ∆α is the variation of the rake angle due to BUE, αA and ωα are the amplitude and frequency of the variation of the rake angle due to BUE, Rse is a f unction to generate an arbitrarily shaped period si gnal; ∆τ1 is the increment of the shear stress due to the hard g rain, τA is the amplitude of the increment of the shear str ess due to the hard grain; Pul is the Pulse function to generate square wave; ∆u is the change of the friction angle between the tool rake face and chip, r is the step function; lw is the tool flank wear width, Fr is the resultant 476 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING cutting force, Ht is the hardness of the cutting tool material, V is the cutting speed, V s is the sliding speed, E is the process activation energy, R is the universal gas constant, Tf is the cutting temperature in the tool flank zone, A and B are the constants; ∆ct is the variation of cutting thickness, ∆cw is the variation of cutting width; Es is the spindle synchronous error, As, ω and φ is the amplitude, spindle angular speed and phase shift of the spindle axial runout; es is the error of the slideway in X/Y/Z direction, ks is the stiffness of the slideway in X/Y/Z direction, Fs is the cutting forced in X/Y/Z direction; Ev is the environmental vibration, fev and Aev are the frequency and amplitude of the environmental vi bration. The demonstrated workpiece material is the Aluminum alloy, thus the dynamic cutting force model will f ollow the elastic-plastic mechanics model. Based on the e lastic-plastic deformation principle, the forces ac ting on the rake face can be acquired by the coordinate transfo rmation of the shear plane force based on the shear plane cutting model. The force acting on the cutting edge and flank face can be deduced based on the empiric al formula of the contact stress and elastic recovery. Accumulating the forces action on the three zones, there will be the dynamic cutting forces in the three directio ns following the Cartesian coordinate system in Fig . 1 expressed as [11] :             +−+−−−+−−+−−+−= −⋅+−⋅++ −+−−−+−−+−−+⋅= +⋅++⋅+−++ −+−−−+−−+−−+−+= rrteccrrcfcz tftfrte cctcy ffffrrrfe ccrrfcx RKTtztzdRRTtxtxhTtztzdKKtF KKRK TtztzdRRTtxtxhTtztzdKtF KKRK TtztzdRRTtxtxhTtztzdKtF θθθ βµββµβθ βµββµβθθθ θθ sin})]()([)]()()]{[()([sin)()( )cos(sin) 2 cos 2 (sin })]()([)]()()]{[()([)( )sin(cos) 2 sin 2 (cos)cos1sin( })]()([)]()()]{[()()[cos1sin()( 0 2 0 2 0 210 2 0 2 0 210 2 0 2 0 (9) where h is the undeformed chip thickness, f is the feed rate, µ is the friction angle coefficient, R 0 is the tool nose radius and dc is the depth of cut, β is the side clearance angle, θr is the intersection angle of two continuous tool paths, T is the spindle revolution period, Ktc, Krc, Kfc, Kte, Kre, Kfe, Ktf1 , Ktf2 , Kff1 and Kff2 are the cutting constants at rake face, cutting edge and fl ank face in X , Y and Z directions, which can be acquired by the transformation from the orthogonal cutting expe riments and an empirical tool force model. As has been stated, the real tool path ( X tp, Z tp) is created by all the above factors: servo motion s influenced by dynamic cutting disturbance ( X sm, Z sm), machine tool deformations (spindle radial runout Esr, spindle axial runout Esa, Z slideway side-stiffness Kzx , X slideway side-stiffness Kxz ) and environmental vibrations ( Eevx , Eevz ) shown as:    −++= −++= xzzsaevzsmtp zxxsrevxsmtp KFEEZZ KFEEXX / / (10) 3.0 Simulation of Nano/micro Machined Surfaces 3.1 Prediction Algorithms for a Machined Surface The last stage of machining system is to renders th e surface topography. The cutting tool will follow the real tool path to reproduce the tool profile on the mach ined surface in the form of feed marks. 477 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING Cheung et al. [12] and Kim et al. [13] calculate the intersection of two adjacent tool pro files to get the boundary of each feed marks like Fig.8 (a). However, in the ultra-precision machining, whose feedrate is so sma ll that the next several cuts will clean up the previous to ol marks. As shown in Fig.4 (b), the ( i+1) th and ( i+2) th tool profiles which are over cut by the ( i+3) th can not affect the workpiece surface generat ion. The machined surface is contoured by the i th, ( i+3) th, ( i+4) th, etc. tool profiles as a result of tool inte rference from vibrations. Frankly speaking, tool path vibration i s not the entire reason to form the surface topogra phy in this case. The tool interference is normal in the small feedrate cutting process, and it always help to fla t the tool- workpiece vibration in this case. All the 3D surface topography prediction procedures are described in Fig. 5 as a block flow diagram. T he principal advantage of this approach is considering the multiple tool interference which is more coinc ident with the real practice. 0. 1 0.11 0. 12 0. 13 0.14 0.15 0. 16 0. 17 0.18 0. 19 0. 2 -5 0 5 10 15 20 x 10 -5 radius of the workpiece (mm) de pt h of cu t (m m ) 0 0. 01 0. 02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 -5 0 5 10 15 20 x 10 -5 radius of the workpiece (mm) de pt h o f c ut (m m ) i i+ 1 i+ 2 i+ 3 i+ 4 (a) Intersections of two adjacent tool profiles ( b) Intersections in consideration of all tool profi les Fig. 4 2D Machined Surface Profile simulation Fig. 5 The Flow Chart of 3D Surface Topography Simu lation Model Simulation Resolution ? ? , ? f Process Parameters n , f , d c Workpiece Dimention R w Calculate the Simulation Parameters N s , N r , N p Real Tool Path x p , z t p Convert the Tool Path into Rectangular Coordinate System x c , y c , z c Real Tool Path Plot Sample the Tool Path to R adial Sections r a , ? a , h a Tooling G eometry R 0 , L t Superpose the Tool Profile onto each Tool Nose Point in j th Section r t , ? t , h t Select the Minimize Tool Profile Point at each Radius in j th Section r , ? , h 2 D Surface Profiles Grid the 2 D Surface Profile Points to Construct the 3 D Surface Data x , y , z 2 D Surface Texture R t , R a , ... 3 D Surface Topography 3 D Surface Analysis S t , S a , ... 478 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING 3.2 Simulation on the Surface Generation The relative displacement between the tool and the workpiece in Z direction is mainly decided by workp iece material, machine tool performance, tooling geometr y and cutting parameters. The simulated surface und er the dynamic cutting process, cut from edge to centre, i s a 1.4 mm × 1.4 mm square located at the spindle c entral line as Fig. 6. The impressive features on the surface are the ring s and flutes which are much bigger than the tool fe ed mark size. To analysis the waviness structure regulation , it is better to work out the connection of the vi brations and the machined surface topography. Since the tool pat hs in one section are the discrete points by feed s pacing per spindle revolution, the surface generation is l ike the process of cutting tool edge in feed direct ion sampling the vibratory tool path. The machined surface topog raphy under vibration is directly decided by the re lation between the vibration frequency fv and the spindle rotational frequency fs. Define the frequency ratio fr as: ba n f f ff v s v r +=== 60/ (11) where a is the integer part of the ratio and b is the fractional part in the rage of -0.5 to 0.5. In this case namely, fr is the frequency that the tool tip traverses the v ibration within one spindle rotation period. The nu mber of the flutes is identical to the integer part a, which means how many times the vibration is compl etely undergone per spindle revolution. The absolute value | b|, which can not cover the full vibration cycle, wi ll make the phase offset in each spindle rotation. The orientation of the flutes is decided by the sign of b. Positive sign means counter-clockwise (CCW); negative sign is clockwise (CW). In other words, the radial section presents the fractional part of the frequency ratio b, while the integer part a will be reflected in the circumferential direction. Fig. 6 Flat Turning Surface Simulation Fig.7 Ident ification of the Surface Prediction System (a) Measurement of the Actual Machined Surface (b) Simulation of the Machined Surface 479 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING 4.0 E xperimental Verification In order to evaluate the effectiveness of the dynam ic cutting process model and the surface prediction method, the surface descript so far were verified through t he experimental work. Fig.7 (a) shows the measureme nt result of an actual Al specimen machined at 1000 RP M spindle speed, 0.01 mm/rev feedrate, 0.01 mm dept h of cut, and 0.508 mm tool nose radius using Zygo NV 5000 system. The measured surface locates on the 2 8- 29 mm radius area where the tool marks are nearly t he straight lines. The surface prediction is simula ted as Fig.7 (b) whose form is accordant with the experime nt result. 5.0 Conclusion • Metal cutting process is combined with the material property, cutting chatter, tooling geometry, servo capability, mechanical performance, and environment al vibration, etc. The integrated simulation system can help to better understand the connection of the dynamic cutting process and the surface generation. The non-linear factors will simulate th e real cutting condition, such as hard grain of material, BUE, regenerative vibration, tool wear an d spindle runout. Building the drive and control system in the model gives the flexibility to evalua te and improve the control system performance. • Surface prediction should take consider of tool int erference effect. The tool interference commonly appears when the low feedrate or big rake radius cu tting tool is adopted. The valleys on the machined surface may be not caused by the tool interference, but from the tool path vibration. • The frequency ratio method can interpret the surfac e topography formation. In this case, a states the number of the flutes; | b| is the phase offset in one section and the sign o f b present the flutes direction. • The computer simulation and preliminary experimenta l results have proved that the approach is able to identify any existence of tool vibrations in nan o/micro cutting process and their effect on the surface generation. Currently, the authors are unde rtaking well-designed substantial cutting trials an d simulations to further verify the approach develope d potentially applied to free-form surfaces in multi-axis nano/micro machining, the results will b e presented in other papers in the near future. Acknowledgement The authors wish to acknowledge the assistance and support of the EU sixth Framework IP MASMICRO project (contract NMP2-CT-2004-500095-2). In partic ular we wish to thank our partners in the RTD 5 subgroup. References [1] C.F. Cheung, W.B. Lee. Characterisation of nanosurf ace generation in single-point diamond turning. International Journal of Machine Tools & Manufacture . 2001, 41, 851~875 [2] C.F. Cheung, W.B. Lee. A theoretical and experiment al investigation of surface roughness formation in ultra- precision diamond turning. International Journal of Machine Tools & Manufactur e. 2000, 40, 979~1002 [3] W.B. Lee, C.F. Cheung. A dynamic surface topography model for the prediction of nano-surface generatio n in ultra- precision machining. International Journal of Mechanical Sciences, 2001, 43, 961~991 [4] W.B. Lee, C.F. Cheung, etc. Develop of a virtual ma chining and inspection system for ultra-precision d iamond turning. Proc. IMechE. Part B: Journal of Engineering Manufa cture, 2007, 221, 1153~1174 [5] A.A. Cardi, et al. Workpiece dynamic analysis and p rediction during chatter of turning process, Mechanical Systems and Signal Processing, 2007 [6] X. Luo, K. Cheng and R. Ward. The effects of machin ing process variables and tooling characterizations on the surface generation, International Journal of Advanced Manufacturing Technology. 2005, 25, 1089~1097 480 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MODELLING AND SIMULATION OF THE DYNAMIC CUTTING PRO CESS AND SURFACE TOPOGRAPHY GENERATION IN NANO/MICRO CUTTING [7] X. Luo and K. Cheng. Nonlinear effects in precision machining of engineering materials, American Society of Precision Engineering Annual Meeting, 2003, 489~493 [8] Delta Tau Data Systems, Inc, Turbo PMAC User Manual , 2006 [9] T.H.C. Childs, K. Maekawa, T. Obikawa and Y. Yamane . Metal Cutting: Theory and Applications, Arnold, L ondon, 2000 [10] C. Schmidt, P. Frank, H. Weule, J. Schmidt, Y.C. Ye n and T. Altan. Tool wear prediction and verificati on in orthogonal cutting. 6th CIRP International Workshop on Modeling of Mach ining Operations . Hamilton, Canada, May 20, 2003 [11] Yusuf Altintas. Manufacturing automation: metal cut ting mechanics, machine tool vibrations, and CNC de sign. Cambridge University Press, Cambridge, 2000 [12] C.F. Cheung, W.B. Lee. Modelling and simulation of surface topography in ultra-precision diamond turni ng. Proc. IMechE. Part B: Journal of Engineering Manufacture , 2000, 214, 463~480 [13] Dong-Sik Kim, In-Cheol Chang, Seung-Woo Kim. Micros copic topographical analysis of tool vibration effe cts on diamond turned optical surfaces, Precision Engineering. 2002, 26, 168~174 481 482 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MICRO-MILLING MACHINES P. Stavropoulos, A. Stournaras and G. Chryssolouris * Laboratory for Manufacturing Systems and Automation , Director, Prof. George Chryssolouris, Department of Mechanical Engineering and Aeronautic s, University of Patras, Greece * xrisol@mech.upatras.gr Abstract A critical issue in micro-milling is the unpredictabili ty of tool life and the premature tool failure. Micro-end-milling is emerging as an important fab rication process. Its benefits include the ability to fabricate micro and meso-sc ale parts out of a greater range of materials and with more varied geometry than it is possi ble with lithography and etching. A variety of sensors can be used for capturing th e necessary information on micro-machining process. These sensors may vary from encoders , load cells, accelerometers to acoustic emission sensors. Each sensor type has a main field of application, which depends both on the desired level of pre cision and the control parameter that must be measured. This paper presents the design philosophy, restrictions and considerations, of a combinational method for de veloping a process monitoring system that is capable of monitoring simultaneously th e spindle’s and tool’s condition during micro-milling operations. The design of the monitoring system is based on vibration and acoustic emissions caused by the micro-milling pr ocess. Keywords: Micro-Milling, Condition Monitoring, Vibr ation Emissions, Acoustic Emission 1.0 Introduction The importance of maximizing the tool’s working tim e and doing the utmost to keep them from breaking, is directly related to the cutting-process optimizatio n in modern manufacturing [1]. One of the main goa ls is to find the appropriate balance among the tool-wear, s urface quality and productivity regarding the tool’ s cost, its replacement cost, the cost of writing off the machi ne’s idle time and so forth [2] . In recent years, a trend for integrating several mo nitoring approaches, through sensor fusion, exists in developing the Tool Condition Monitoring Systems (T CM). Such solutions are very important for the development of new more effective systems for monit oring the micro-milling process. However, before investigating the modern TCM, the breakage mechanis ms should be fully understood in order to be monito red effectively. Three breakage mechanisms may occur i n the micro-milling operation [3]. The first one i nvolves the chip clogging that could take place when chips are not removed fast from the machining area. Cutt ing 483 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES forces and stresses on the tool are increased beyon d its endurance limits. Consequently, a breakage i s an unavoidable happening. Chip clogging is difficult or even impossible to be predicted. Secondly, a fa tigue- related breakage may occur due to tool wear. The r eason for this is that the cutting forces and stres ses increase with the increase of the tool wear and stay high fo r an extended period. Thirdly, a tool breakage may occur if the tool deflection during the cutting is beyond th e endurance limit of the tool. This may result bec ause of the decreased sharpness of the cutting edges, their par tial damage, or the deposition of material in the t iny grooves of the cutters. Unfortunately, in micro-milling ma chines, the breakage of the tool is not visually de tectable, due to their small sizes, and the generated chips a s well as the cooling mist in the machining area [3 ]. Thus, the only solution in micro milling is the use of TC M in order for the process effectiveness to be impr oved. The tool condition monitoring systems (TCM) is a ke y issue for ensuring a better use of the machine-to ol’s capabilities. Sensorial information from several s ensors (i.e. accelerometer, dynamometer, acoustic-e mission AE sensor, strain force sensors, etc) is gathered d uring the process, analyzed and compared, by assess ing the deviations, in representative variables, in the tim e and frequency domain [1]. The time and frequency domain analysis confirms the relevance of the sensor signa ls’ signatures of the TCM systems in the HSM proces ses. The major factors affecting the TCM systems, the wo rkpiece and the machine tool in an HSM process, are cutting force, tool wear, tool deflection and spind le vibration. In most engineering problems, feedin g the process with inputs could lead to getting useful ou tputs. From the TCM system’s viewpoint, there are some variables and parameters involved in the process. These can be considered as inputs to a TCM problem, and are; the spatial position of the cutting tool (Cart esian coordinate axes), the spindle speed (Vsp in r pm), the relative feed speed between tool and worktable (fee d rate – f in mm/min), the cutting speed (Vc in m/m in), the radial depth of cut (DoC in mm) as well as the cutt ing-tool diameter (d in mm). Regarding these varia bles (inputs), the experiments could be executed for var ious (in theory infinite) values and in different combinations, by collecting in this way, the sensor ial signals and analyzing the tool wear condition. The monitoring operation, as described by Artis in [4], is based on the comparative measuring methods. During the entire machining process, the curve of m easured characteristics is compared with the stored characteristics of an identical machining operation having been carried out previously with a sharp to ol. Whenever the characteristics of the current machini ng operation deviate from the comparison stored dat a by more than a specified tolerance range, the type and amount of deviation can be used in order for speci fic information on various errors to be obtained. Micro-end milling enables the manufacturing of func tional micro-components, in three dimensions, with a high material removal rate, with the use of tools u nder 50 µm in diameter [5][6]. Furthermore, tool parameters such as the edge geometry, grain size and orientati on that do not affect considerably the machined com ponent at larger scales, are of great importance; whilst t he unit removal size decreases because these parame ters affect dramatically the accuracy, surface quality and inte grity of the component [6]. These characteristics set a challenging environment for the development of a re liable tool condition monitoring system. 2.0 Monitoring System Requirements Nowadays, in order for products with a high degree of economic efficiency, productivity and quality to be manufactured, the use of tool monitoring systems is indispensable for many production processes [7]. In order for the basic function of a TCM system to be guarantied, the following requirements for a mon itoring system are being specified and are as follows: 484 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES • Few or no sensor systems on the machine (if possibl e) • No cut-out required in the switch cabinet • No space required in the switch cabinet • No additional operator panel on the machine • Quick installation • Network capability • Automatic set-up function, simplified operation • Low-cost • Modular and expandable • Process visualization The above requirements show that a high degree of i ntegration is necessary in order for an efficient a nd flexible monitoring system to be built. However, t his integration is not only limited to operation an d visualization, but it is also necessary in order fo r further information and signals from the control system to be used. Moreover, the monitoring systems should reco gnize the defects and report them as a fault messag e or system state [8]. This result can be used as input for a diagnostic process in which the reasons for fault are found and located. In the field of manufacturing p rocesses, the main tasks of these systems are: • Function control and fault location of the machine components • Process control for recognizing any process failure • Recognition of machine inaccuracy, leading to the l ack of quality • Support to the operating and maintenance staff 3.0 Tool Condition Monitoring Methods Current methods of tool condition monitoring can be classified into two main categories [6] [9]: i) th e direct method, in which sensors gather data directly from the cutting edge of the tool and ii) the indirect method, in which sensors gather data that can be correlated wi th the tool condition. Direct Tool Condition Monitoring Method Optical microscopes are used for gathering detailed images and for inspecting the geometry of the cutt ing edges as well as the surface condition of the tool. Flank wear can be detected with the use of a CCD camera, whilst the crater wear requires the projection of a structured light pattern onto the tool, in order t o acquire depth information from within the crater. In struc tured light sensing, with the use of laser interfer ometers, the distortion of parallel lines of the laser light giv es a measure of the crater’s depth. Due to the sma ll size of the tools, used in micro machining, this method is rath er difficult to be applied. Proximity sensors esti mate the tool wear by measuring the change in the distance b etween the tool edge and the workpiece. This dista nce is affected by the tool’s thermal distortions, the def lections or vibrations of the workpiece and the too l. Capacitive displacement sensors are capable of meas uring a minimum distance of 0.2mm with a frequency up to 20 kHz [10]. Such sensors can also be used for measuring the unbalance, due to the loss of materia l from the tool cutting edges. The variation of the measu red distance, up to certain limits, from a relative ly fixed point to the cutting edge, can show the condition o f the cutting tool. Direct tool condition monitori ng strategies have been proposed and reviewed in liter ature. The main advantage of the methods proposed is that they do not introduce any restrictions on the cutti ng tool movements/operations. 485 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES Indirect Tool Condition Monitoring Method A variety of sensors can be used for capturing the necessary information on the machining process. Th ese sensors may vary from encoders, load cells, acceler ometers to acoustic emission sensors. Fig. 1. Sensor application vs. level of precession and control parameters [11] Each sensor type has a main field of application, w hich depends both on the desired level of precision and the control parameter that must be measured. However, due to the exponential growth of the sensor technol ogy, these borders are flexible, since each sensor type may be utilized successfully in several application s. Load cells are mainly adequate to perform conventional m achining, but they do not offer the necessary signa l to noise ratio (S/N) and sensitivity that are required in precession machining [5][11]. Cutting forces a nd power consumption in micro end milling are extremely low, so the application of the load cells is reliable f or the measured range of frequency, at several times highe r than the rotational frequency of each tool’s cutt ing edge. As shown in Fig. 1, acoustic emission (AE) sensors are rather adequate for high precision machining, d ue to the sensitivity ratio of high noise. Moreover, a m ajor advantage of the AE sensors is their ability t o deal with frequencies higher than the characteristic ones of the machining process, thus, limiting the introduct ion of noise into the generated signals[11][12]. The sign als are classified into continuous and burst, havin g distinctly different characteristics. The continuous signal i s associated with shear in the primary zone and wea r on the tool face and flank, whilst burst signals result ei ther from tool failure or from chip breakage [12]. The above advantages come to terms with the results of [13] t hat monitored a milling process successfully with t he use of an AE sensor, placed on the work piece. Their moni toring set up was capable of holding an indication of the tool wear and surface quality and of measuring tool breakage. AE signals were proven as the best choi ce between signals from a microphone and an accelerome ter. Vibrations in a machining process may be produced by cyclic variations on the dynamic compon ents of the cutting forces [14]. An important poin t of a monitoring system is the measurement of vibrations of the cutting tool’s rotational shaft. These meas urements can be made with the use of accelerometers [15]. T he sensors of this type can typically measure the acceleration in three dimensions (tri-axial), in a frequency greater than 10 kHz [16]. The most impor tant point of an indirect tool condition monitoring system is the fusion [17][18][19] of different types of senso r signals 486 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES and the conclusions concerning the tool condition. The mounting location of each sensor, related to t he machining point, plays a major role to the desired characteristics of the indirect monitoring system. 4.0 TCM System Setup The possible-preferred physical quantities and sens or types that can be used for detecting tool failur e are: force and torque, vibration via acceleration, acoustic em ission and ultrasound sensors and laser sensors. D ue to the nature of the process (high speed machining) the ma ximum frequency is in the range of 9.5-10 kHz (90.0 00 rev/min) and of the cutting forces around 5Nt. The measurement of the cutting forces has several d rawbacks. Firstly, these forces are very low and a re difficult to be measured accurately. Also the inst allation of the force-torque sensor has negative in fluence on the dynamics and stiffness of the machine. So ther e is a need for an ultra precise force sensor in or der to meet the above force requirements. Most promising techniques are the vibration and aco ustic measurements (including acoustic emission). Acoustic emission (AE) measurements are high freque ncy acoustic signals that originate from the deform ation of the work piece material. When the tool breaks t he sensor it also measures the AE energy resulting from the fracture. In addition, accelerometers measure the vibrations generated by the process, and are capabl e of measuring the cutting operation even for the very s mall cutting forces. There is a need for capable v ibration sensors to sense the changes in machining condition s. Piezoelectric vibration sensors measure the mec hanical vibration of the machine structure resulting from t he cutting process, typically up to 10 kHz. It can be used for detecting missing tools, broken tools, out-of-toler ance parts, machine collision and severe process fa ults. It is also possible to monitor any excessive vibration on the spindle. The vibration sensor is easy to be i nstalled on new or existing machines. An ultrasound and vibrat ion sensor is suitable for measuring vibration-indu ced oscillations up to an ultrasonic range (100 –: 80 0 00 Hz) in machine components. A light barrier offers a reliable tool breakage and tool missing monitoring system if tools are too sm all to be monitored by force, true power, or if there is no s uitable place available for an AE sensor to be moun ted. In a reflective single-beam laser system, the tool condi tions are monitored by constantly focusing a laser beam with a spot size of 50 µm on the cutter, and at the same time, directing th e beam reflection towards a receiver. To avoid “hidden” areas, for example, when milling poc kets or grooves, the laser beam is focused just abo ve the tool working length. The light intensity of the re flected beam is measured with the use of the amplif ier for indirectly detecting any tool breakages. The proposed Tool and Spindle Condition monitoring system lies on the combination of Acoustic Emission s and the Acceleration sensor, installed on a 3-axis micro-milling experimental device as shown in Figur e 2. The air-driven spindle has a maximum rotation speed of 100.000 rpm. 487 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES Fig. 2. Overview of Micro Milling experimental setu p It introduces the use of a sensor on the spindle mo unting bracket and a second sensor on the clamping device or the X-Y-Z table. For indirect spindle condition monitoring an accelerometer may be used on the mou nting bracket. This sensor can measure the vibration’s v elocity, due to the spindle’s shaft unbalances. A recommended way of coupling such a sensor on the br acket is the stud mounting method. Fig .3. Close view of the machining area and monito ring sensors Figure 3 shows the details of the micro-milling TCM monitoring setup. The micro-milling machine is equipped with an air driven high speed spindle (1) reaching 100.000 rpm. The acceleration sensor (2) mounted on the spindle bracket is a Kistler Annular Ceramic Shear Triaxial – 8762A50 accelerometer, wi th a sensitivity of 100mV/g and a frequency range from 0 .0005 to 6 kHz. The micro-milling tool (3) may va ry according to the machining needs; from conventional micro tools to synthetic or natural diamond millin g tools. A calibre clamping device (4) ensures the proper pl acement of the work piece. Finally, the Acoustic E mission sensor (5), a Kistler 8152B2 AE one, with a frequen cy range of 100-900 kHz is placed as near to the mi lling point as possible, but without interfering with the process.. The overall connectivity diagrams of th e sensor in the DAQ system are shown in the figures below. 488 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES Fig. 4. Connectivity diagrams of the Acceleration ( left) and Acoustic Emission (right) sensors Data acquisition is accomplished by PXI (PCI eXtens ions for Instrumentation) units, which are a rugged PC- based platform for measurement and automation syste ms. PXI combines PCI electrical-bus features with t he rugged, modular, Eurocard packaging of CompactPCI, and then adds specialized synchronization buses and key software features. PXI is both a high-performa nce and low-cost deployment platform for measuremen t and automation systems that meet the requirements a s mentioned in section 2 of this study. 5.0 Monitoring Strategy Existing strategies of the process monitoring can b e divided into signal based, model based, and class ifying methods depending on the complexity of the manufact uring process [8]. Using signal- or boundary value - oriented methods, the measured signal values should be compared with pre-defined signal values or with a signal range. The basis of model-based monitoring techniques are process models that are either deter mined empirically or from physical relations. In using mo dels for process control, it is important that the model be supplied with useful input variables describing the process that is to be examined. The target of the classifying monitoring systems is to find the link of a feature vector to a certain class of quality features. Th is vector is often determined by feature extraction of the proce ss signals. For the signal based technique, an analogue (electr ical) signal from the sensor is usually (after basi c signal conditioning, e.g. primary filtering) converted int o a digital form. The time series obtained is then processed to extract signal features that are sensitive to th e parameters of interest in the monitored process. The detection of process irregularities is achieved by the implementation of some sensing methodology, cal led a monitoring strategy [20]. In the past, it was suff icient to monitor an upper limit but meanwhile, man y strategies and algorithms have since been developed in order to deal with the many different processes [21]. Most of the techniques that are incorporated into m onitor strategies are based on static limits. A st atic limit can be a threshold for activating alarm signals, an d it remains fixed during the processing of a workp iece. Furthermore, dynamic limits are introduced into mon itoring strategies, which are driven from sensorial signals and tolerance ranges. The monitoring strategy adop ted for this work is the so called dynamic limits s trategy. In this method, two dynamic limits follow the monit or signal continuously for every load level at a li mited adopted speed. In the case of an extremely fast cr ossing of one or two dynamic limits, they are froze n 489 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES (rendered static) and the total breakage, breakage, chipping, workpiece cavity, hard cut interruption, etc. are distinguished from one another via a visual compari son with the monitor signal. Slow but large load c hanges ,due to variations in cutting depth (hardness, over size, out-of-roundness of workpiece), such as those occurring during the initial cuts, in particular, when machin ing cast and forged parts, are tolerated at a ratio up to 1:4. The signal is also automatically used for coping wi th the wide difference in force or signal values pr oduced; for example, by large roughing tools as compared wi th small finishing ones. Signal adaptation automat ically keeps them for analysis at an optimum level. The c ombination of feature conditioning, automatic signa l adaptation and dynamic limits mean that monitoring functions are fully automated over a wide range of force or sensor signals in completely different machining situations, without manual adjustments or a teach- in phase. Tool breakages are practically detected at the inst ant of breakage (typically 5 ms), by means of typic al changes in the sensor signal 6.0 Conclusions A major issue in micro-milling is the unpredictable tool life and premature tool failure. The specific acoustic footprint and very small removal rates, during mach ining, as well as the use of small diameter cutters , makes the detection of tool breakage a very difficult tas k. Thus, it is essential to develop new tool monito ring systems to increase the process productivity, reduce machin ing costs and at the same time, improve the precisi on and quality of machined components. The existing tool condition monitoring systems have a relatively high cost to performance ratio, and they are efficient only in l imited applications and operational conditions. Th ere is a need for reliable, efficient and more economical sy stems, for the better monitoring of the machine and the tool condition, as well as for enhancing the quality of the parts produced. This paper presents the design philosophy, restrictions and considerations of a co mbinational method for developing a process monitor ing system that is capable of monitoring simultaneously the spindle and tool’s condition during the micro- milling operations. References [1] Haber, R. F., Jimenez, J. E., Peres, C. R., Alique, J. R., 2004, An investigation of tool-wear monitor ing in a high- speed machining process, Sensors and actuators A, 1 16:539-545 [2] Stavropoulos, P., K. Salonitis, A. Stournaras, J. P andremenos, J. Paralikas and G. Chryssolouris, "Exp erimental Investigation of Micro-milling Process Quality", Pr oceedings of the 40th CIRP International Seminar on Manufacturing Systems, Liverpool, U.K., (May-June 2 007). [3] Henrioulle, K., Sas, P., Process monitoring of micr o-milling using vibro-acoustic measurements [4] Publications from Artis, Control-integrated tool an d process monitoring, Artis Gesellschaft für angewa ndte Messtechnik mbH, http://www.artis.de [5] Dornfeld, D., S. Min and Y. Takeuchi (2006). Recent Advances in Mechanical Micromachining. Annals of t he CIRP, Vol 55 (2). [6] Gandarias, E., S. Dimov, D.T. Pham, A. Ivanov, K. P opov, R. Lizarralde and P.J. Arrazola (2006). New m ethods for tool failure detection in micro-milling. Procee dings of the I MECH E Part B Journal of Engineering Manufacture, Vol. 220 (2), pp. 137-144 [7] Lange, C., 2002, High productivity due to process m onitoring, Artis Gesellschaft für angewandte Messte chnik mbH, http://www.artis.de [8] Tonshoff, H. K., Jung, M., Mannel, S., Rietz, W., 2 000, Using acoustic emission signals for monitoring of production processes, Ultrasonics, 37:681-686 [9] Chung, K. T. and A. Geddam (2003). A multi-sensor a pproach to the monitoring of end milling operations . Journal of Materials Processing Technology, Vol. 13 9, pp. 15-20. [10] Micro Epsilon (2005). Products catalogue: The Autho rity in Displacement and Position Measurement, Sens ors Systems Software and Solutions 490 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 O N THE DESIGN OF A MONITORING SYSTEM FOR DESKTOP MI CRO-MILLING MACHINES [11] Lee, D. E., I. Hwang, C.M.O. Valente, J.F.G. Olivei ra and D.A. Dornfeld (2006). Precision manufacturin g process monitoring with acoustic emission. International Jo urnal of Machine Tools & Manufacture, Vol. 46, pp.1 76–188. [12] Xiaoli, L. (2002). A brief review: acoustic emissio n method for tool wear monitoring during turning. I nternational Journal of Machine Tools & Manufacture, Vol. 42,pp. 157-165 [13] Henrioulle, K. and P. Sas (2006). Process monitorin g of micro-milling using vibro-acoustic measurement s. National Congress on Theoretical and Applied Mechan ics, NCTAM 2006, May 29th-30th, Mons, Belgium. [14] Dimla, E. and S. Dimla (2000). Sensor signals for t ool-wear monitoring in metal cutting operations—a re view of methods, International Journal of Machine Tools & M anufacture, Vol. 40, pp. 1073–1098. [15] Kang, M.C., J.S. Kim and J.H. Kim (2001). A monitor ing technique using a multi-sensor in high speed ma chining. Journal of Materials Processing Technology, Vol. 11 3, pp. 331-336 [16] Jemielniak, K. (1999). Commercial Tool Condition Mo nitoring Systems. Int J Adv Manuf Technol, Vol. 15, pp. 711–721. [17] Chryssolouris, G. and M. Domroese (1989). An Experi mental Study of Strategies for Integrating Sensor Information in Machining. Annals of CIRP, Vol. 38 ( 1), pp. 425-428. [18] Chryssolouris, G., M. Domroese and L. Zsoldos (1990 ). A Decision-Making Strategy for Machining Control . Annals of CIRP, Vol. 39 (1), pp. 501-504. [19] Chryssolouris, G., M. Domroese and P. Beaulieu (19 9 2). Sensor Synthesis for Control of Manufacturing Processes. Journal of Engineering for Industry, ASME , Vol. 114 , pp. 158-17 4 . [20] Jemielniak, K., 1999, Commercial Tool Condition Mon itoring Systems, Journal of advanced Manufacturing Technology, 15:711-721 [21] Lange, C., 2002, High productivity due to process m onitoring, Artis Gesellschaft für angewandte Messte chnik mbH, http://www.artis.de 491 492 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FOR MICRO-SH EET- FORMING Akhtar Razali, Yi Qin * , Colin Harrison and Andrew Brockett Department of Design, Manufacture and Engineering M anagement, University of Strathclyde, James Weir Building, 75 Montrose Street, Glasgow, G1 1XJ, U.K * qin.yi@strath.ac.uk Abstract A recent review of micro-forming research and technological de velopment suggested that the trend of the development is focused more on the m anufacturing processes, machines and tooling, with efforts on the precision material h andling being insufficient. Most of the developed machines were based on stand-alone concepts that do not support efficient integration to make them fully automat ed and integrated. Material feeding in most cases was not of sufficient precis ion and reliability for high throughput manufacturing applications. Precision feeding is ne cessary to ensure that micro-parts can be produced with sufficient accuracy, espec ially in multi-stage forming, while high-speed feeding is a necessity to meet pr oduction-rate requirements. Therefore, the design of a new high-precision and high-speed feeder for micro- forming is proposed. Several possible approaches are examined wit h a view to establishing feasible concepts. Based on the investigation, seve ral concepts for thin sheet-metal feeding for micro-forming have been generated, the se being argued and assessed with appropriate applied loads and force analysis. Thes e form a basis for designing a new feeder. Keywords: micro-forming, micro-handling, forming pr ess, gripper feed, roll feed, micro-tooling 1.0 Introduction Research into the forming of miniature/micro-produc ts [1] has led to investigation into material-feedi ng methods and devices as a part of the development of a machine system to transfer laboratory-based form ing processes to production [2]. Feeding the materials with higher rates and high accuracy in micro-sheet forming is one of the challenges to be met in micro-forming research and development for engineering applicati ons. Conventional press feeders have to meet three main criteria to be successful. Firstly, the feeder must be flexible in terms of set-up. Secondly, the delivery of material must be of sufficient precision to sat isfy the 493 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING requirements for forming. Thirdly, the feeders must also ensure feeding at the correct time. All of th ese are particularly difficult to meet when forming thin sh eet metals, such as those where the thickness is le ss than 100 microns, where the feeding distance is greater than 10mm and where the feeding rate higher than 500 st roke per minute (SPM). These are the requirements for th e development of a new machine system for micro-she et- forming: Fig. 1 shows a 3D model of the machine dev eloped at the University of Strathclyde. Fig. 1. 3D model of the micro-sheet-forming machine Two methods of feeding sheet-metals in conventional stamping may be applied to micro-sheet-forming - r oller feeding and gripper feeding [3-4]. The servo roll f eeder uses an electric servomotor while a gripper f eeder mainly uses pneumatic actuation. Although the latte r may exhibit limited flexibility in the varying of its travel distance and feeding speed, this type of the feeder has the potential to compete with the servo roll f eeder in respect of positional accuracy and precision. Great er accuracy is achievable due to the possibility of a gripper feeding mechanism not having a complicated mechanic al transmission and hence, no backlash, wear, tear, etc. which can contribute towards inaccuracies in feedin g. However, an error arising in translation from an gular rotation to linear motion in roll feeding contribut es towards inaccuracies: this does not occur with a linear gripper feeder. With the possible use of a servo sy stem, the performance of a gripper feeder could be improved, regarding the flexibility in travel dista nce within the system length set-up, and in feedbac k and positional accuracy. The research reported in this paper is dedicated to a study of these issues, bas ed on which considerations for a new feeder design are develope d. 2.0 Linear-Displacement Devices for Sheet-Metal Feeding Application s Three categories of devices have been used widely f or high-precision feeding - electromechanical actua tors [5- 9], electrical actuators [10-12] and piezoelectric- actuators [13-14]. Table 1 shows a comparative stud y made on those devices. The proper selection of a high-pr ecision system is essential in order to secure high - precision/accuracy for high yield feeding applicati ons. Linear motors are seen to have superior charac teristics among the others in terms of speed, acceleration, p recision/accuracy, robustness, ease of maintenance, etc. as well as being used widely by researchers as an alt ernative to conventional rotary machinery [10-18]. 494 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING Table 1: Types of linear displacement devices and t heir suitability for micro-press feeding applicatio ns Device Type Accuracy and precision Acceleration rate (g) Force (N) Size Reliability Solenoid Inflexible and uncontrollable over the stroke distance, and hence difficult for precision to be defined. Impressive response time and acceleration over short distance. Fairly high force, which is inversely proportional to stroke distance. Small and compact, suitable for a constrained space application. Reliable in terms of mean time between failures (MTBF). Ball/lead screws, belts, gears, rack and pinions, etc. Flexible with limited stroke distances and accurate up to 6- 7µ m. A greater inaccuracy develops with time, due to wear, tear, etc. Acceleration rate is low due to mechanical transmission being involved. High thrust force is available. Fairly small, depending on the application. Reliable as demonstrated by most of the devices. Linear motors and stages Stroke distance may be limited. Combined with air-bearing, the system could have very high positioning precision. No backlashes to contribute to inaccuracies. Acceleration rate can be high: 5- 10g is typical, and 40g available commercially. High thrust force, ranging from tenths up to thousands of Newtons. Fairly small. Force is proportional to the coil size. Reliable as demonstrated in many cases. Piezoelectric- actuator (linear motor) Very accurate and precise. Has been used in photonics and high precision applications but with very limited travel distances. Response time is better over short distances, reflecting high acceleration rate. Low force ranging between 7-10N, limiting its application to low force feed & positioning tasks only. Depends on the travel distance required. Reliable for low speed positioning. For high speed positioning, heat may tend to build- up as well as wear and tear. Linear motors are seen to have advantages over othe rs in terms of achievable accuracy with impressive acceleration rates. Since there are no backlashes a nd plays, which will contribute towards positioning inaccuracies in the system, greater accuracy is exp ected to be achieved. Therefore, a linear motor is seen to be a good platform when considering feeding design for sheet metal to achieve high precision and high-rat e feeding. In the selection of an appropriate motor for the fe eder for a micro-press among iron-cored and iron-le ss motors, several parameters need to be considered - force density and magnetic attraction, stiffness a nd settling time in terms of dynamic and static characteristics , accuracy, velocity, stability, etc. Both iron-cor ed and iron- less motors have their respective advantages in ter ms of the foregoing factors. The first parameter to be considered is the force density. As the name implie s, due to the presence of iron laminations in the f orcer of an iron-core motor, extra force may be generated by at traction force from the magnetic track on the force r, in addition to the force generated by the electromotiv e force (emf). Magnetic attraction on the iron lami nation in the forcer also contributes to a ‘cogging’ effect on the iron-core motor. ‘Cogging’ affects the smoot hness and repeatability of an iron-cored motor to achieve out standing precision, such as that which an iron-less motor might be able to achieve [12]. An iron-less linear motor was used to study the multi-degrees-of-freedo m error motions of a precision linear aerostatic-bearings s tage [19] and to determine the achievable precision . Such a motor was used because a greater precision is achie vable due to no ‘cogging’-effect existing, compared to an iron-cored linear motor. In another study [20], an iron-less linear motor was used to drive an air-be arings stage and to determine the precision of the motor. Greater precision and repeatability were found with an ironless-linear-motor-driven stage. A high stiffness of the motor in terms of static an d dynamic characteristics is required due to the sh ort settling time and rapid motion involved. A short settling ti me associated with high dynamics or high positional stability requires stiff mechanical characteristics of the motor. The epoxy structure in an iron-less motor may have low inherent stiffness, but the rigidity of th e motor may be enhanced by the copper coils inside the forcer 495 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING which leads to better stiffness. However, the steel structure in an iron-cored motor makes it stiffer than an iron- less motor. Nevertheless, according to previous rep orts [14]-[20], control of the ambient temperature to within 1°C may be needed, which should greatly assist ther mal reduction to protect the structure of the linea r motor from excessive thermal deformation. Greater stiffne ss in both the motor structure and the mounting cou ld eliminate structural deformation that could cause b acklash and play during rapid operation. Therefore, a high- accuracy process is feasible and qualified. Table I I tabulates a comparative study of the iron-less an d iron- cored linear motors. Table II: Comparative study on iron-less and iron-c ored linear motors Description Micro-sheet forming feed application Iron-less Iron-cored Continuous force of 150N Very suitable. Most iron-less motors have a contin uous force ranging from 0 – 450N. This indicates that an iron-less motor is only suitable for low load deman d, for example, in a positioning application with a light- weight feeding mechanism. Very suitable. An iron-cored motor usually has a tw o- times greater continuous force than an iron-less on e, which indicates suitability for applications demand ing greater force. Peak force of 500N Very suitable for feeding. The greatest peak force is rated at up to 1600Nm and normally lasts for a few seconds before the motor starts to overheat and bur n. Very suitable. Higher continuous force leads to hig her peak force, compared to an ironless linear motor. Smooth motion Very suitable. No cogging effect and using an air bearing (gap between forcer and U-channel magnetic tracks) make non-contact smooth, ultra-precision li near motion possible. Iron lamination inside the forcer causes the coggin g effect. Non-smooth motion between the forcer and th e magnetic way experienced. It is not recommended for high smooth-motion applications. Precision down to a sub-micron order Sub-micron accuracy/precision and repeatability is possible and recommended. Sub-micron accuracy/precision and repeatability is feasible in slow speed and can be used as a cheaper alternative option. Speed stability Up to 0.1% error at 1kHz measuremen t - it is very stable in terms of speed. Speed stability is equal to that of an iron-less mo tor and it is recommended as a cheaper alternative. Thermal dissipation Not recommended due to no heat-transfer medium: entirely reliant on air circulation to reduce the m otor temperature. Controlled ambient temperature might reduce thermal build-up. Recommended for high heat build-up applications due to this motor having water and air-cooling mediums. Dimensional constraint Recommended due to compact sizes, e.g. a 200mm x 200mm sized feeder has been developed. Not recommended due to large and bulky structure. A small and compact feeder is not feasible with this type of motor. Acceleration Most of the motors have an acceleratio n rate of up to 40g (more than 115g theoretically). Acceleration rates of up to 10g are feasible. Speed Speed ranges of from 0 – 10m/s. High speed ra nges similar to those of iron-less motors are achievable. Stroke < 50mm Recommended. A moving magnet or movin g forcer can be proposed. Suitable, provided that moving-magnet motion is proposed. Clean room applications Very suitable due to no particle generation (if no moving cable is deployed). Suitable if no moving cable motion is deployed. 3.0 Feasibility Study of a New Gripper Feeder The correct sizing of the linear motor during the d esign stage is crucial, as this will have an impact on the performance of the entire system and will contribut e to the achievement of the designated production r ate. Associated considerations include the designated lo ad and push/pull force. Figure 2 shows the applicab le forces that are taken into account for a linear-mot or sizing analysis. Three types of force contributi ng to the total peak and continuous linear-motor forces are i dentified for this particular application. 496 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING Fig. 2. Types of the related forces, friction betwe en parts and payloads, that contribute to total pea k and continuous force requirements. 3.1 Forces due to the Payload As depicted in Fig. 3(a), for the case studied, abo ut 0.3kg payload of the clamping mechanism and weig ht were required to be moved. As designed, the 0.3kg p ayload is: to advance 19mm in 0.120s; to dwell for 0.030s; to retract for 0.120s; and to dwell for 0.0 30s; the cycle being repeated thereafter for 3.33H z operation. In this case, analysis and calculation of the requi red forces in order to determine the best linear mo tor and amplifiers are major requirements. The first factor considered was the motion characte ristics - the peak speed needed to accelerate the m ass from the origin to the end point, the time duration whic h the travel takes, and the dwell period when the m oves end. In general, for this type of motion, which is from point to point, the basic profile is trapezoidal mo vement. With this movement the time is divided equally into three parts. The first part is acceleration, the s econd part is constant velocity, and the third part is deceler ation. Such motion characteristics should ensure a balance between the speed and acceleration to give the best motor combination. Based on trapezoidal motion, th e time taken to accelerate is calculated as: s s 040.0 3 120.0 = Then the peak speed required to make the movement w as calculated and in this case, because the movemen t is symmetrical and divided into three parts, the equat ions below were used. The load cannot accelerate instantaneously from 0 to 0.475m/s and, as previous established; it will take 0.040s to reach this spe ed. Therefore accelerate rate then was calculated as sh own in Fig. 3(b): (a) (b) Fig. 3. (a) The load exerted on the forcer, and th e direction and distance of movement;, and (b) the trapezoidal profile representing the acceleration of the forcer in one cycle. (1) 497 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING gsm s sm t uv a sm s m t s v 8.1/81.17 04.0 0/7125.0 /7125.0 04.02 019.03 2 3 2 ≈= − = − = = × × == The peak payload rating force, fp, considers: the frictional force, ff, (with the assumption of re-circulated ball bearings being used to carry the load in the system , having a coefficient of friction of about 0.002 u p to 0.003); the force for acceleration, fa, calculated using Newton’s laws of motion; and the gravitational force for an inclined plane, fg; as well as the external force, fe,, caused by the cable management. Therefore, the p ayload rating force may be expressed as: efgap fffff +++= fa represents the forces for the load, including the forcer mass, and is used to calculate the final coi l- temperature rise, peak and the continuous current a nd minimum bus voltage. In the mechanical-transmiss ion linear stage, the frictional value on the lead scre w system should also be taken into account as it us ually affects the system positional accuracy. Nf f Nmgf mgf Nmaf p e f g a 352.50009.00343.5 0 009.0003.081.93.0 081.91.0)0sin()sin( 343.581.173.0 =+++=∴ = =××== =××== =×== µ θ By adding a safety factor of 25%, the calculated fo rce was increased to compensate for the degradation of the motor efficiency, the new force to move the payload was calculated to be 6.7N. Due to the minimum pull ing force required being almost 7N, a piezoelectric act uator is not the best option for this application. As for a mechanical-transmission linear stage system, carryi ng a low/light load is not recommended due to the u neven distribution on load that might also contribute to positional inaccuracies, as has been reported in [9 ]. 3.2 Friction Forces Friction forces due to the contact between the shee t metal and the guide plates inside the machine (50 µ m thick strip, 50mm width and contact length 500mm) is anal yzed as follows. The mass of material in contact w as calculated as: gramkgm mkg mV Vm 1010775.91082.71025.1 /1082.7 1025.100005.05.005.0 336 33 36 ≈×=×××=∴ ×= ×=××= = −− − ρ ρ The coefficient of friction between the strip (carb on steel) and the tool-steel surfaces is taken as 0 .15 hence: (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) 498 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING Nf maf f f 02.081.901.015.0 15.0 =××= = = µ µ By adding a safety factor of 25%, the new frictiona l force value is found to be 0.03N. This value will be added to the calculation of the total peak force for the sizing of the linear motor. 3.3 Pulling Force for the Coil Reel Another force which contributes to the total peak a nd continuous forces is the force for pulling the m aterial from its reel, as illustrated in Fig. 4. This can b e estimated as follows:. Fig. 4. Free-body diagram of the coil and reel prod ucing a total of torque labeled as τcomposite. knowing that: 0 04.0 0625.0 0190.0 = = = = u st mr ms ; and assuming the following: 0, 2 3 , 2 == − = = = == u t s v t uv a mrI r a Ifr α ατ From the trapezoidal motion: 2 2 /285 0625.0 8125.17 /8125.17 04.0 07125.0 /7125.0 04.02 0190.03 srad sma smv == = − = = × × = α Supposing that the outer diameter of the coil (carb on steel) and the reel (made of Perspex) are 125mm and 110mm respectively, and by by assuming that both th e coil and the reel are solid, moment of inertia of the composite material can be calculated, assuming the following:: 33 /1190,/7820, mkgmkg v m reelcoil === ρρρ Mcoil + Mreel Torque to rotate coil and reel from rest, τcomposite Force to move the coil and reel from rest, Fpeak Speed, acceleration and angular acceleration at which the coil and reel to be moved; v, a and α Strip Coil and reel rcoil rreel (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) 499 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING mthickness mwidth mlength coil coil coil 00005.0 05.0 50 = = = kgm mv coil coil 9775.07820000125.0 000125.000005.005.050 3 =×= =××= mh mr reel reel 012.0 055.0 = = 342 1014.1012.0055.0 mv reel −×=××= pi ; kgm reel 1357.011901014.1 4 =××= − )( 2 22 inneroutercoil rr mI += 2322 1041.3)0555.00625.0( 2 9775.0 kgmI coil −×=+= 2 2 reel reel mr I = 24 2 1005.2 2 055.01357.0 kgmI reel −×= × = reelcoilcomposite III += 2343 1062.31005.21041.3 kgmI composite −−− ×=×+×= ; Nmsradkgm composite 032.1/2851062.3 223 =××= −τ Nf composite 5.160625.0 032.1 == Therefore, the total force required for rotating th e combination of the coil and the reel is found to be at 16.5N. By adding a safety factor of 25% (to overcome inter nal frictional forces, etc), the total of the new f orce is calculated to be at 20.6N. 3.4 The Uncoil Braking Force A study on the effect of the coil braking force tow ards strip tension during the stamping process was conducted previously [19]. However, effort was focu sed more on the quality in punching a hole quality instead of on feed accuracy. The relationship between the s trip tension and the feed accuracy still needs to b e studied in detail. As proposed previously [20], the strip t ension should be kept at the same level as that of the torque needed when coiling the metal strip in the first pl ace. An adjustable braking force was proposed to ma ke the uncoil process flexible in terms of adjusting the t orque to various values [19]. Supposing that the co iling process requires a similar level of torque as the u ncoiling of the metal strip, given by the analysis in 3.3, the maximum force to uncoil is 20.6N. By taking frms of the rated force, therefore, the designated vari able braking force for this application is as follows: Nf brake 64.1015.0 04.06.20 2 = × = Since 10.64N force acts during the whole cycle, thi s is therefore the value needed to be included in t he peak force in the sizing of the linear motor. 3.5 Influences of Other Interfacial Forces Micro-forces which may be neglected in the macro-wo rld may no longer be neglected in the micro-world. The ratio of micro-forces to the weight of micro-parts is at about the same value, whereas in most common situations, the micro-forces are proportionally is greater than the weight of micro-parts, when the si zes of the (25) (26) (27 & 28) (29) (30) (31) (32 & 33) (34) (35) 500 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING latter are, 100µm 3 and smaller. Hence, such parts will stick and sta y on the handling mechanism, overriding the effect of gravity and thus resulting in problem s in the manipulation process. Three types of micro -forces which have a significant influence on micro-parts a re known and have of late been studied extensively; adhesion forces, van der Waal’s forces and electros tatic forces. Usually, the adhesive force between p article surfaces is due to the presence of van der Waal’s f orces and electrostatic forces [23-24]. However, th ese forces do not have a significant effect on material handli ng for this application, since the handled material is larger than 100µm 3 . 3.6 Predicted Peak and Continuous Forces Based on the analysis presented above, the peak for ce can be calculated as follows: NffffF brakecompositefppeak 3864.1063.2003.07.6 =+++=+++= The rms force is the average force, frms from the motor and helps to determine the final te mperature that the coil will reach. Based on the above trapezoidal-pro file case, the calculation is as follows: cycle peak rms t tF F × = 2 st st cycle 15.0 04.0 = = NF rms 6.19 15.0 04.038 2 = × =∴ The new summations of the peak and continuous force s are 38N and 19.6N respectively. Both of these for ces were considered when selecting a suitable linear mo tor. An Iron-less linear motor is chosen due to the demanded peak and continuous forces being relativel y small. In addition, by using an iron-less linear motor, a compact system is feasible. 3.7 Coil Temperature, Peak and Continuous Current, and Minimum Bus Voltage 3.7.1 Final Coil Temperature Analysis The final coil temperature represents the temperatu re at which the linear motor may be operated withou t adversely affecting its materials of construction. Therefore this final temperature can be as a guidel ine when deciding which servo controller is best suited to t he designated linear motor. In order to determine t he rise of coil temperature, the following equation was used. With the assumption of ambient temperature is 25°C, the rise can be calculated as below with the data suppl ied from one of the iron-less-linear-motor catalogu es [25]. Back Electro-motive force (BEMF) = 24.5 V/m/s; Forc e constant = 21.3 N/amp; Motor constant = 26.3 N/ √W; Coil resistant = 0.7 Ω; Thermal Resistant, R T = 0.64 W/°C; from the analysis, fp = 38.0N; frms = 19.6N. With the assumed ambient temperature, coil temperat ure rise may be calculated as below: (36) (37) 501 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING C M f RT C rms T °=   =    = 36.0 3.26 6.1964.0 22 ; Therefore, the final coil temperature = 25 + 0.36 = 25.36°C. 3.7.2 Sizing the Amplifier This analysis was conducted to determine the most a ppropriate size and type of servo controller to be used so that the linear motor can provide its best performa nce without suffering from current and voltage drai n-out. Based on the given value of the motor’s force const ant and the calculated peak and continuous forces, the peak and continuous current and minimum bus voltage are calculated as follows: Peak current= fp/force-constant = 38/21.3 = 1.78A Continuous current= frms/force-constant = 19.6/21.3 = 0.92A Drive Voltage min. = (peak current x coil resistant) + (velocity x ba ck EMF) = (1.78 x 0.7) + (0.7125 x 24.50) = 18.7V (41) Therefore, the servomotor controller must be capabl e of supplying a peak current and a continuous curr ent minimum of 1.78A and 0.92A respectively. 3.7.3 Thermal Expansion due to Temperature Increase A thermal effect due to temperature changes in micr o-manufacturing may not be negligible [26-27]. Apparently, a small increment in temperature may co ntribute significantly to the performance of the ma chine elements. The heat generated by the motor coil, if it is not well controlled, can cause thermal deflec tion of mechanical parts. Aluminum alloys are often found t o be used as linear-stage material due to their goo d heat conductivity, in dissipating heat efficiently from the coils of the linear motor [25], [28-30]. The an alysis below is conducted to understand how much deflection may be experienced by the parts of a gripper that is lo cated on the top of the stage of the feeder. Supposing th at aluminum alloys 6061-T6 have a conductivity of 24.3µ m/m°C [31], the thermal expansion due to the h eat generated by the motor during operation is [32] : mmTExp thermal µµ 7.83.24)2536.25(3.24 =×−=×∆= Supposing a 20mm travel distance, the total thermal expansion expected is: mm m mExp mmthermal µ µ 174.002.0 1 7.8 20 =×=− Therefore, the thermal expansion due to the heat ge nerated by the coil of the motor for the gripper is expected to be a maximum of 0.174µ m for 100% transferred hea t. This small deformation due to 100% transferred h eat can be negligible as the error is relatively low an d not contributing significantly towards positional inaccuracy. 4.0 Conclusions and Further Considerations (38) (39) (40) (42) (43) 502 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING Based on the studies described above, a linear actu ation method that uses a linear motor is proposed a s a strategy for developing a new feeder for micro-shee t-forming. Direct drive from a linear motor ensures that no mechanical transmission will be required, which lat ter could contribute to play and backlash that affe ct the accuracy of the feeding in micro-forming. For the c ase analysed (the strip materials and micro-stampin g process specified), theoretically, 38.0N and 19.6N peak and continuous forces respectively are require d to serve material feeding for this particular applicat ion. At least 2g of acceleration rate is needed to accelerate the payload, i.e. the gripper and strip moving for up t o 200 parts per minute (ppm) in a high-precision op eration. The level of precision to be achieved is significan tly greater than that available in the use of a ser vo roll feeder, which, currently, offers only 50µ m accuracy . A solenoid has been chosen to serve the clamping ap plication for the gripper design due to its impress ive response time, holding force, ease of integration, etc. The logic output from the servomotor controlle r is used to control the solenoid, hence giving peace of mind on the integration side by eliminating the necessi ty of using another type of controller. A new feeder for micro-sheet-forming is now being constructed. Acknowledgements The support from the European Commission for conduc ting research in “Integration of Manufacturing Syst ems for Mass-manufacture of Miniature/micro-products (M ASMICRO)” (www.masmicro.net) (NMP2-CT-2004- 500095) is acknowledged. Beneficial discussion and the exchanging of experience with the EU researcher s and engineers in micro-manufacturing are particular ly acknowledged. References [1]. Qin, Y. (2006) Micro-forming and miniature ma nufacturing systems – Development needs and perspec tives, Keynote Paper of the 11th Int. Conf. of Metal Formi ng, J. of Mater. Proc. Technol., 177 (1-3), 8-18. [2] Qin, Y., Ma, Y. Harrison, C., Brockett, A., Z ao, J., Zhou, M. and Razali, A. (2008), Development of a new machine system for the forming of micro-sheet-produ cts, Proc. of the ESAFORM2008 Conf., DOI: 10.1007/s12289-008-0098-9, April, 2008, pp. 1-4. [3] www.bruderer.com [4] www.pa.com [5] SATO, K. & MAEDA, G. J. (2008) Practical control me thod for ultra-precision positioning using a ballsc rew mechanism. Precision Engineering, in-press. [6] ERKORKMAZ, K. & KAMALZADEH, A. (2007) Compensation of axial vibrations in ball screw drives. Annals of the CIRP, 56. [7] CHUNG, S.-C. & KIM, M.-S. (2006) Integrated design methodology of ball-screw driven servomechanisms with discrete controllers. Part I: Modelling and pe rformance analysis. Mechatronics, 16 , 491-502. [8] CHERN, G.-L. & RENN, J. C. (2004) Development of a novel micro-punching machine using proportional solenoid. Journal of Material Processing Technology, 25 , 89-93. [9] MEI, X., TSUTSUMI, M., TAO, T. & SUN, N. (2003) Stu dy on the load distribution of ball screws with err ors. Mechanism and Machine Theory, 38 , 1257-1269. [10] HSUE, A. W.-J., YAN, M.-T. & KE, S.-H. (2007) Compa rison on linear synchronous motors and conventional rotary motors driven wire-EDM processes. Journals of materials Processing Technology, 192-193 , 478-485. [11] SHINNO, H., YOSHIOKA, H. & TANIGUCHI, K. (2007) A n ewly developed linear motor-driven aerostatic X- Y planar motion table system for nano-machining. Annals of the CIRP, 56. [12] YAMAZAKI, K., SRIYOTHA, P., NAKAMOTO, K. & SUGAI, M . (2006) Development of 5-axis linear motor driven super-precision machine. Annals of the CIRP, 55. [13] MRACEK, M. & HEMSEL, T. (2006) Synergetic driving c oncepts for bundled miniature ultrasonic linear motors. Ultrasonics, 44 , e597-e602. 503 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 INVESTIGATION OF FEEDING DEVICES AND DEVELOPMENT OF DESIGN CONSIDERATIONS FOR A NEW FEEDER FO R MICRO-SH EET-FORM ING [14] MRAD, R. B. & TENZER, P. E. (2004) On amplification in inchworm(tm) precision positioners. Mechatronics, 14 , 515-531. [15] KIM, K., CHOI, Y.-M., GWEON, D.-G. & LEE, M. G. (20 08) A novel laser micro/nano-machining system for FPD process. Journal of Materials Processing Technology, 201(1-3), 497-501. [16] CHEN, S.-L. & HSIEH, T.-H. (2007) Repetitive contro l design and implementation for linear motor machin e tool. International Journal of Machine Tools & Manufactur e, 47 , 1807-1816. [17] HILLERY, M. T. & GORDON, S. (2005) Development of a high-speed CNC cutting machine using linear motors. Journal of Materials Processing Technology, 166 , 321-329. [18] CHO, D.-W., KIM, J.-J. & JEONG, Y. H. (2004) Therma l behavior of a machine tool equipped with linear motors. International Journal of Machine Tools & Manufactur e, 44 , 749-758. [19] GAO, W., ARAI, Y., SHIBUYA, A., KIYONO, S. & PARK, C. H. (2006) Measurement of multi-degree-of- freedom errors motions of a precision linear air-be aring stage. Precision Engineering, 30 , 96-103. [20] PARK, C. H., OH, Y. J., SHAMOTO, E. & LEE, D. W. (2 006) Compensation for five DOF motion errors of hydrostatics feed table by utilizing actively contr olled capillaries. Precision Engineering, 30 , 299-305. [21] CHERN, G.-L., WU, Y.-J. E. & LIU, S.-F. (2006) Deve lopment of a micro-punching machine and study on th e influence of vibration machining in micro-EDM. [22] RUSSELL, J. (2003) Handling appliance steel - Tips for processing surface-sensitive materials. Stamping Journals(R). [23] TOMAS, J. (2007) Adhesion of ultrafine particles—A m icromechanical approach. Chemical Engineering Science, 62 , 1997-2010. [24] ROUGEOT, P., REGNIER, S. & CHAILLET, N. (2005) Forc es analysis for micro-manipulation. Computational Intelligence in Robotics and Automati on, IEEE. [25] www.parkermotion.com [26] BRUSSEL, H. V., PEIRS, J., REYNAERTS, D., DELCHAMBR E, A., REINHART, G., ROTH, N., WECK, M. & ZUSSMAN, E. (2000) Assembly of microsystems. Annals of the CIRP, 49 , 451-472. [27] ARONSON, R. B. (2004) Micromanufacturing is Growing . Manufacturing Engineering. [28] www.kollmorgen.com [29] www.yaskawa.com [30] www.baldor.com [31] www.sas.org [32] KALPAKJIAN, S. & SCHMID, S. R. (2006) Manufacturing Engineering and Technology, Prentice Hall. [33] www.pi.ws [34] www.ledex.com 504 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE COMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE Panikowska K., Tiwari A., Alcock J. R. School of Applied Sciences, Cranfield University, M K43 0AL, UK k.e.panikowska@cranfield.ac.uk , a.tiwari@cranfield.ac.uk , j.r.alcock@cranfield.ac.uk Abstract This paper summarizes the main findings of a survey of comple xity literature for engineering design and reviews the use of this word in mic ro-devices literature. The general view on the definition of the word complexity is captured and comple xity types are identified. The paper underlines the subjectivity and context-dependence of meaning of complexity, as it is currently used. The paper provi des identification of the common characteristics of complexity definitions and the reason s why people attempt to develop or influence definitions of complexity. The paper c oncludes that a sufficient definition of complexity for micro-devices has not been provide d and highlights how this issue is currently viewed in literature. Keywords: complexity definition, type, micro-device s. 1.0 Introduction Miniaturization has absorbed the attention of resea rchers from many decades. Increasing demand for new and smaller solutions with incorporation of multi-funct ionality has lead to the increasing “complexity” of these devices. The word “complexity” has been used to des cribe the large number of designed and manufactured micro- and nano- devices, the multidisciplinarity o f the designs, the high technology equipment used f or their production and assembly, as well as the lack of kno wledge about micro-scale physics, chemistry and bio logy and hence of device function. Owing to this issue, the complexity literature has been investigated wit h the aim of identifying a sufficient definition of this word for the micro-devices domain. To fully understand how complexity is viewed five main topics were investig ated: 1) universal definitions of complexity, 2) ty pes of complexity, 3) reasons to define complexity, 4) sou rces of complexity and factors influencing it, and 5) complexity in micro-devices. All of them are presen ted in following sections of the paper. 505 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE 2.0 Complexity Definition Complexity is established as important field of stu dy [1]. However, the word “complexity” is not only hard to define [2], [3] but in many areas, a precise defini tion is still not available. Factors that influence this difficulty are the context-dependence and subjectivity of comp lexity [4]-[6]. Researchers have made attempts to generate a universal definition of complexity, whic h have resulted in several journal publications, co nferences, books and doctoral dissertations. Resulting from th is body of work, “Complexity Theory” has been estab lished as a separate domain of study with diverse applicat ions. Despite this effort, the definition of comple xity provided by researchers still varies in different f ields (and sometimes even across the same field) sh owing a discrepancy in terms of meaning, usage and quantifi cation. In an attempt to define complexity, many researcher s have started by identifying what it does not mean . They have indicated differences between complexity and c omplicatedness [6], [7], randomness [8] and other i ssues which influence complexity and can be confused with it, such as size, lack of knowledge, variety, and order/disorder [4]. Other authors have tried to est ablish its meaning by highlighting common character istics, such as those given by Corning [5] who describes co mplex phenomenon as those that consist of many part s, with have high number of relationships/interactions , and in which the parts produce combined effects t hat are not easily predicted and may often be novel. Other features are pointed out by Simon [9] who stated th at: complexity critically depends on system description , which can be simplified by correct representation , that a complex system is characterized by redundancy and t hat its hierarchy can be often described in economi cal terms (aggregation of redundant components and cons ideration of them as integrated units). Complexity has been defined in many areas of study such as chaos theory, fuzzy logic, networks, philos ophy, psychology, and statistics. [9]. Amongst these defi nitions are: algorithmic information context (AIC) 1 or “Kolmogorov’s Complexity”, length of the message, o r “Crude Complexity”, introduced by Gell-Mann, logical depth of a string in programming, created b y Bennet, average amount of information stored at a ny time in order to make an optimal forecast, “Forecasting Complexity”, established by Grassberger and many mo re. Each of these definitions is context specific. The majority of them suffer from a defect in construct ion, as they contain within the explanatory definition the word “complex”. A trend is observable in the literature for the presentation of such circular definitions of comple xity. These are then followed by the core part of t he work, which is a focus on the measurement of this phenome non and, having gained this quantitative tool, on methodologies to decrease complexity. The area which provides more suitable definition fo r products, systems and any other materialistic cre ations is engineering design. Although, definitions particula r to engineering design are focused mainly on the information which the system, device, and product c ontain, several diverse definitions are available. El-Haik & Yang [10] present complexity as “a quality of an ob ject with many interwoven elements, aspects, detail s, or attributes that makes the whole object difficult to understand in a collective sense”. Although, this definition is valid for every object it is not specific enough to allow for quantification as well as not present th e whole meaning of complexity. Another, frequently cited, d efinition of complexity was introduced by Suh in connection with axiomatic design. He defined comple xity very broadly with the aim of providing an abso lute 1 Simultaneously discovered by three independent sci entists: Kolmogorov, Chaitin and Solomonoff 506 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE measure for it, this quantitative approach being vi sible in first words of definition. According to Su h, complexity is ’a measure of uncertainty in understa nding what it is we want to know or in achieving a functional requirement (FR)’ [6]. Both these defini tions are focused around understanding a design fro m the points of view of difficulty and uncertainty. Hence , these definitions may cause problems where the de sign is “fully understood” or could be represented in simpl e manner, but would still be considered as “complex ” by an observer. In the case of full understanding of desi gn, the complexity would be measured as zero, which would indicate that there is no complexity in the device, despite the clear appearance of complexity to the observer. Is it possible to design a device which is characte rized by a complete lack of complexity? Some resear chers claims that the answer to this question is ‘no.’ El -Haik & Yang [10] presented the idea of “irreducibl e complexity” which they considered a universal quali ty in all objects. However, they underlined that th is level of complexity may significantly vary. This view was supported by Colwell [11] who based his opinion of the minimum amount of complexity required on systems pe rformance – the impossibility of separate parts of the system performing the functions required from the d evice, or performing them inadequately, if they are not connected. He supported his view by citing Einstein ’s statement of the simplicity limitations in order to achieve required performance of a design outcome. 3.0 Sub-types of Complexity The inconsistency in definitions of complexity caus es differences in identification of their sub-types in the literature. Suh [6] identified four time-related su b-types of complexity: time-independent real comple xity – ‘a measure of uncertainty when the probability of achi eving functional requirements is less then 1.0 beca use the system range is not identical to the design range’, time-independent imaginary complexity – caused by lack of knowledge, time-dependent combinatorial complexity – caused by unpredictability of future events and t ime- dependent periodic complexity – existing in finite time period with predictable number of combinations of events. Adami [3] divided complexity into physical and structural. His domain of study was biological organisms; however he adapted the AIC definition of complexity, which was created for programming. Zamenopoulos and Alexiou [12] recognized sub-types of complexity as: functional and behavioural, where as Tomiyama et al [13] noted both complexity by design and intrinsic complexity of multi-disciplinarity. These sub-types of complexity were created based on particular characteristics identified by researche rs and each author has provided their own sub-types referr ing to particular domain of research. However, some overlap of these sub-types of complexity, in terms of their meaning, can be identified This overlap is tabulated in Table 1. Since the development of a universal me asure of complexity is “hard to imagine” [3], the c reation of complexity subdivisions makes it possible, in th e majority of cases, to group features which can be measured in order to provide a quantitative indicat ion of complexity level. 507 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE Table 1: Sub-Types of Complexity Type of Complexity indicated Researcher(s) / Literature Source He yli ghe n [2] Ada mi [3] Edm on ds [4] Tho ms on et al. [5] Ear l, Eck er t & Cla rks on [7] McG uir e [8] Col we ll [11] Zam en opo ulo s & A lex iou [12] Tom iya ma et al. [13] Fun es [14] Suh [15] Kim [16] Bos e, Alb on es i & Mar cu les cu [17] Gel l-M an n [18] Own definition x x x x x Irreducible complexity x Information complexity x Kolmogorov x x x x x x System complexity x Observer complexity x Löfgren’s Interpretation and Descriptive Complexity x Kauffman’s number of conflicting constraints x Physical x Structural x x x Functional x x Structural hierarchical x Functional hierarchical x Behavioural x Crude complexity x Logical depth x x Forecasting complexity x Computational Complexity x x x Gell-Mann's Effective Complexity x Complexity by design x Intrinsic complexity of multi- disciplinarity x Suh complexity x x Time-independent real x x Time-independent imaginative x Time-dependent combinatorial x Time-dependent periodic x 4.0 Why a Definition of Complexity is Required Many authors have put considerable effort into defi ning complexity, but what was their purpose? What a ctions did they undertake once their definition of complex ity was established? A number of authors have state d that the reason for their work is that complexity is har mful. However, others have pointed out that only s pecific types of complexity are damaging, whereas other typ es are useful and even required. 508 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE Suh [6] claimed that a ’vast sum of human and finan cial resources are wasted due to our inability to d eal with engineering complexities.’ Thomson et al. [5] point ed out that higher complexity than originally antic ipated for the project, participates in cost and schedule overruns. Both authors accepted the unavoidability of complexity but blamed incorrect or inadequate manag ement of complexity for badly influencing design. T hey criticized the general lack of knowledge about comp lexity, which lead to its misunderstanding. Their v iews have some commonality with the idea of “irreducible complexity”, however they do not provide informati on about what level of complexity is acceptable. As a reason to properly define complexity in a spec ific context, Suh [6] provided a view of the opport unity of its reduction and an increase in the system’s relia bility and robustness. In his complexity theory the re are 3 harmful types of complexity: time-independent real and imaginary complexity, cause over-runs of projec ts in terms of time and cost, and time-dependent combinat orial complexity, leads system to a chaotic state a nd results in a system’s failure. Suh underlined first ly, the necessity of reducing time-independent imag inary complexity, which could be achieved by writing down the design equation (showing relationship between the functional requirements and design parameters for p articular product)[15], and, secondly, the need to change time-dependent combinatorial complexity to periodic complexity, what can provide long-term stability o f the system. Colwell [11] highlighted that the reduction of comp lexity is compromised by minimization of functional ity and/or other tradeoffs. This value-adding complexit y view is, in his opinion, only reasonable to a cer tain extent, beyond which the cost of increasing complex ity is not necessary. He stated that each attempt t o create complexity in design should be justified, and when this justification cannot reasonably be provided co mplexity should be reduced. Negative impacts of this additio nal amount of complexity, in his opinion, included: longer development schedules; design errata, follow-on des ign issues and cost and time overruns. 5.0 Sources of Complexity Since, complexity is such an important aspect in an y design, sources of it should be characterized. Identification of the reasons for a particular leve l of complexity, as well as those features which in fluence it, can help with its measurement and then, potentially , changes in its level, if required. Rodríguez-Toro , Jared and Swift [9] claimed that proper management of com plexity sources can help in the reduction of ‘desig n effort’ which results in a shortening of developmen t time and in cutting project costs. According to Suh [6], complexity is caused by poor design, which can be result of, for example, a non- systematic approach to design, or a lack of knowled ge (understanding) about the system under considera tion. Earl, Eckert & Clarkson [7] stated that complexity has its origins in a combination of order and uncer tainty, where the ordered background of existing designs, p rocesses and requirements is combined with an uncer tain change process and unpredictable outcome. However, both of these approaches are very broad, and hence can be very freely interpreted. 509 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE Thomson et al. [5] introduced more detailed identif ication of the factors which influence complexity i n design, which can be considered as sources of complexity. T hey established, the concept of a “Design Complexit y Map”, which represents those attributes of a design affecting complexity. They identified six groups o f factors: knowledge and sources, artefacts, design activity, external and internal aspects (e.g. technology, lif e phase systems), decision making and actors. Each of these groups contains at least two subgroups and each su bgroup has number of positions underneath. Although, this map has been designated to represent complexity of the team environment during the design process, it is a lso valid for the design outcome itself. When apply ing this framework to a product, issues presented have to be divided into those that have direct impact on the complexity of design outcome, such as part artefact s, and show potential to be measured, and those wit h indirect impact such as actors participating in the design process. This framework shows potential to influence the complexity of the design outcome in the concept ual phase by both indicating which elements have to be taken into account and by providing an opportunity to measure complexity. 6.0 Complexity in Micro Devices With regard to the high number of definitions provi ded for complexity and their sub-types, the assumpt ion of the possibility of a special meaning of “complexity ” for micro-scaled devices seems reasonable. Severa l attempts to define the complexity of micro-devices are available in the literature. However, it is not able that within the domain of micro-devices, devices are oft en stated to be either simple or complex without a definition of “complexity” or an explanation of whe re is the border between simple and “complex” lies. Within this domain, there are three main methods by which definition of complexity is derived: by crea tion of a definition by the researcher, by adaptation of so meone else’s approach or by the identification of characteristics. Zhou [19] represents an example of the first method . He defines complex micro-devices as ‘devices comp osed of parts made from different materials fabricated b y various technologies,’ and claims that this compl exity is continuously increasing due to new demands on the m arket. This definition, created for micro-assembly, is very broad and does not provide sufficient meaning of the word “complexity” for whole micro-devices domain. The second approach, to adapt approach to complexit y and its measurement from the macro scale, was undertaken by Kim [16], [20]. He applied the “axiom atic” approach to multi-scale systems design with a focus on micro and nano-scale. His work showed the possib ility of a reduction its quantification. However, t his is one of few attempts identified were a definition cr eated for macro-scale was adapted in micro-scale do main. Kim states that usage of “functional periodicity” w ill allow the decrease of overall complexity by transformation of a system with time-dependent comb inatorial complexity to a system with time-dependen t periodic complexity, which was identified as less h armful. He also claims that by consideration of unc ertainty associated with functions axiomatic design approach can help in understanding complexity in micro- and nano- assembly. Although he noted that ‘information conte nt well-characterizes the real complexity of tiny p roduct manufacturing,’ Kim neither states that the definit ion of complexity provided by Suh [6] is suitable f or micro- devices nor created his own definition for this dom ain. 510 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE Finally, Albers and Marz [21] are an example of las t method. They noted that every micro device is a m ulti- technology product. They stated that the design of these small devices, if they are aimed to be optima l and innovative, has to be realized as an integration of technology, process and product development, mater ial sciences and simulation, embracing all these discip lines. They described the process of micro-technolo gy design and manufacturing as very complex due to the unavailability of proper tools and the high degree of uncertainty of the functionality of products after manufacturing processes. This uncertainty, accordin g to certain definitions of complexity confirms the high complexity of these devices, however it does not q uantify its level nor solve the problem of identifying the sources of complexity. Although, these attempts at definition of complexit y for the micro-scale have been identified, the amo unt of available literature regarding this topic is small. However, several authors have described the necess ity to decrease the level of complexity in micro-devices, especially regarding the negative influence of comp lexity on micro-architecture in terms of testability and m anufacturing cost [17]. At the macro-scale, this ha rmful impact of complexity, beyond “irreducible complexit y”, as well as the concept that complexity increase s rapidly as the system scale order grows [16], have convinced many researchers to attempt to measure an d influence it. However, any impact, if achieved, has been measured relatively to the prior state, and n ew methods created have not been applied universally o wing to the subjectivity of the judgments incorpora ted in their definition. 7.0 Conclusions The literature presented above shows the increasing interest of scientists in “complexity.” However, i t also underlines the inconsistency in definitions of this word, its context dependence and subjectivity acro ss different domains as well as inside an area of rese arch. A large number of definitions have been outli ned, most of them created ad hoc to undertake projects, and characterized by a focu s on quantification of a particular issue. The development of complexity definitions, h owever vague and/or narrow, in the majority is aime d at decreasing the level of complexity owing to the con sideration of complexity by majority of researchers as having a destructive effect. The literature shows that some investigation of com plexity has been undertaken in micro-devices domain . However, there is no sufficient definition of compl exity identified particular to this domain. This le ads to the suggestion that further studies should be undertake n to define and influence complexity for micro-devi ces. Acknowledgements The authors would like to thank the Innovative Manu facturing Research Centre (IMRC) at Cranfield for t heir funding and support. References [1] Lewin, R. (1992) “ Complexity. Life at the edge of chaos ”, The University of Chicago Press, London, UK. 511 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 C OMPLEXITY IN ENGINEERING DESIGN AND MICRO DEVICES: A REVIEW OF LITERATURE [2] Heylighen, F. (1996) “The Growth of Structural and Functional Complexity during Evolution”, to be publ ished in: F. Heylighen & D. Aerts (eds.) " The Evolution of Complexity " (Kluwer Academic Publishers). [3] Adami, C. (2002) “What is complexity?”, BioEssays, vol.24, pp.1085-1094. [4] Edmonds, B. (1995) “What is Complexity? - The philo sophy of complexity per se with application to some examples in evolution”, http://cfpm.org/bruce/evolcomp/evolc omp_1.html (accessed 19th March 2008) [5] Thomson, A., Kumar, B., Chase, S. & Duffy, A. (2005 ) “Measuring Complexity in a Design Environment”, Proceedings of the ECCS 200 5 Satellite Workshop: Em bracing Complexity in Design – Paris, November 17 , Paper 100, pp. 67-75, France. [6] Suh, N.P. (2003) “A Theory of Complexity and Applic ations”, Rough Draft #2, http://wisdom.usc.edu/inspiring/resource/bioenginee ring/Namsuh.pdf (accessed 28th March 2008) [7] Earl, C., Eckert, C. & Clarkson, J. (2005) “Design Change and Complexity”, In: 2nd Workshop on Complex ity in Design and Engineering, 10-12 Mar 2005, University of Glasgow, UK. (Unpublished) [8] McGuire, K. “Theory of Complexity”, http://kevin.mcguireclan.net/papers/TheoryofComplex ityWeb/TheoryofComplexity.htm (accessed 28th March 2008) [9] Rodríguez-Toro, C., Jared, G. & Swift, K. (2004) “P roduct-development complexity metrics: A framework for proactive-DFA implementation”, International Design Conference - DESIGN 200 4 , Dubr ovnik, May 18 - 21 , Croatia, pp.484-490. [10] El-Haik, B. & Yang, K. (1998) “The components of co mplexity in engineering design”, IIE Transactions, vol.31, no.10, pp.925-934. [11] Colwell, B. (2005) “Complexity in Design”, Computer , vol.38, no.10, pp. 10-12. [12] Zamenopoulos, T. & Alexiou, K. (2005) “Linking desi gn and complexity:a review”, UCL, Proceedings of the ECCS 200 5 Satellite Workshop: Embracing Complexity in De sign - Paris 17 November 200 5 , Paper 100, pp. 91-102. [13] Tomiyama, T., D'Amelio, V., Urbanic, J. & ElMaraghy , W. (2007) “Complexity of Multi-Disciplinary Desig n”, Annals of the CIRP , vol.56, no.1, pp.185-188. [14] Funes, P. (1996) “Complexity measures for complex s ystems and complex objects”, http://www.cs.brandeis.edu/~pablo/complex.maker.htm l (accessed 31 st March 2008) [15] Suh, N.P. (2007) “Ergonomics, axiomatic design and complexity theory”, Theoretical Issues in Ergonomics Science, vol.8, no.2, March-April, pp.101-121. [16] Kim, S.-G. (2006) “Complexity of Nanomanufacturing” , Proceedings of ICAD2006 4th International Confere nce on Axiomatic Design, June 13-16, Firenze [17] Bose, P., Albonesi, D.H. & Marculescu, D. (2003) “G uest Editors' Introduction: Power and Complexity Aw are Design”, Published by the IEEE Computer Society , pp. 8-11. [18] Gell-Mann, M. (1995) “What is complexity?”, Reprint ed with permission from John Wiley and Sons, Inc.: Complexity , vol1, no.1. [19] Zhou, Q., Albut, A., del Corral, C., Esteban, P.J., Kallio, P., Chang, B., Koivo, H.N., "A Microassemb ly Station with Controlled Environment", Microrobotics and Microass embly III, Bradley J. Nelson, Jean-Marc Breguet, Ed itors, Proceedings of SPIE Vol. 4568, Boston, USA, October 2001, pp. 252 - 260, 2001 [20] Kim, S.-G. (2004) “Axiomatic Design of Multi-Scale Systems”, Proceedings of ICAD2004 3rd International Conference on Axiomatic Design, June 21-24, Seoul, Korea. [21] Albers, A. & Marz, J. (2004) “Restrictions of produ ction engineering on micro-specific product develop ment”, Microsystem Technologies, vol.10, no.3, pp.205–210. 512 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES Tahsin Tecelli Öpöz 1 , Bülent Ekmekci 2 , Abdulkadir Erden 1 1. Department of Mechatronics Engineering, Atılım University, Ankara, Turkey 2. Department of Mechanical Engineering, Zonguldak Karaelmas University, Zonguldak, Turkey Abstract Micro Electric Discharge Machining (micro-EDM) is one of the most common micromachining techniques for manufacturing of micro holes and mini cavities. Micro-structures such as micro-holes, micro-channels, micro-gear and other complex shapes can be easily machined by using micro-EDM irrespective of material hardness. There are a lot of electrical and technological parameters which are effective in the machining characteristics and machined material surface integrity in micro-EDM. In this study, effects of energy parameters on the machining performance are investigated based on experimental results. Series of experiment were performed by keeping all parameters constant except pulse energy. Variations in micro-hole geometry, material removal rate, micro-hole depth and over-cut in micro-hole diameter were investigated. Experimental results have revealed that using high pulse energy setting and 400 µm diameter tool electrode during machining result in deeper micro-holes when compared to micro-holes machined under lower energy settings. However, machining with 100 µm diameter tool electrode, the result is reversed i.e. by using lower pulse energy, deeper holes were obtained. In addition, a defect formation is also observed inside the machined hole when tool electrode suppression rate and pulse energy is increased. Keywords: Micro-EDM, micro-hole, micromachining. 1.0 Introduction Miniaturization of parts and components play an imp ortant role in the development of today’s and futur e technology in various fields. With the increasing d emand for micro parts and structures in many indust ries, and also with rapid developments in micro-electro mecha nical systems (MEMS), micro manufacturing technique s, especially micro electrical discharge machining, fo r producing these parts become increasingly importa nt. Micro structures including micro holes, micro slots , micro shafts, and micro gears are mostly used mic ro 513 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES products needed in industry. Especially micro holes are needed in optical devices, medical instruments and automobile engine parts [1]-[2]. Basically, electric discharge machining (EDM) is a thermal material removal process. The process is ca rried out in a dielectric liquid with a small gap between the workpiece and electrode. Electric discharge oc curs when the dielectric is broken down by the application of voltage pulse. Some of the released energy during discharge is transferred to the electrodes and results in hea ting of highly localized regions of the electrodes. When temperature of heated region exceeds melting temper ature of the electrodes, some of the melted and all of the evaporated material is then quenched and flushed aw ay by dielectric liquid in the form small globular particles (debris material) and the remaining melt recast on the finished surface. Debris material is washed awa y from the sparking area by the continuously flushing diel ectric fluid. Flowing pressure of the dielectric fl uid should be adjusted to an appropriate value since high pres surized fluid result in vanishing the influence of electrical sparks, on the other hand, low pressure flow result in rising debris concentration in sparking area an d cause secondary discharge, arc, and short circuit. The application of EDM is not limited by the hardne ss or strength of material to be machined. EDM can be used to machine any conductive material and there i s no direct contact between the electrode and the workpiece during machining that make possible to ma chine complex geometries using thin electrodes. Thermal properties such as melting point, boiling p oint, and electrical conductivity of workpiece mate rials influence the machining characteristics. The materi al removal rate of EDM process is primarily determi ned by the electrical conductivity and melting temperature of the workpiece material. A workpiece with higher electrical conductivity and lower melting temperatu re can be machined more efficiently. 2.0 Micro Electric Discharge Machining Micro electric discharge machining (micro-EDM) is a derived form of EDM, which is generally used to manufacture micro and miniature parts and component s by using the conventional electric discharge machining principles. Similar to EDM, material is r emoved by a series of rapidly recurring electric sp ark discharges between the tool and the workpiece elect rodes in micro-EDM. Actually main differences of mi cro- EDM from conventional EDM are the type of pulse gen erator, the resolution of the X-, Y- and Z- axes movement, and the size of the tool used. In micro-E DM; pulse generator produces small pulses within pu lse duration of a few micro seconds to nano seconds. Th us, micro-EDM utilizes low discharge energies (~ 59 1010 −− − joules) to remove small volumes (~ 50005.0 − µm 3 ) of material [3]. The most important factor which makes micro-EDM very important in micr omachining is its machining ability on any type of conductive and semi-conductive materials with high surface accuracy irrespective of material hardness. It is preferred especially for the machining of difficult -to-cut material due to its high efficiency and pre cision. Small volumetric material removal rate of micro-EDM provides substantial opportunities for manufacturi ng of micro-dies and micro-structure such as micro holes, micro slot, and micro gears etc. The use of micro- EDM has many advantages in micro-parts, the main advant age is that it can machine complex shapes into any conductive material with very low forces. The force s are very small because the tool and the workpiece do not come into contact during the machining process. Thi s property provides advantages to both the tool and the 514 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES workpiece. For example, a very thin tool can be use d because it will not be bent by the machining forc e hence, eliminating mechanical stress, chatter and vibratio n problems during machining. Other advantages of mi cro- EDM include low set-up cost, high aspect ratio, enh anced precision and large design freedom. Therefore , relying on the above advantages, micro-EDM is very effective to machine any kind of holes such as smal l diameter holes down to 10 µ m and blind holes with a spect ratio of 20. In micro-EDM, tool electrode in different diametric sizes can be prepared by using Wire-Electric Discharge Grinding (WEDG) [4]. It is a prerequisite machining for drilling micro-holes and milling micro-cavities with micro-EDM technology. WEDG is used to prepare smaller size tool down to Ø 10 µ m by using electrical discharge machining princ iple with reverse polarity. High aspect ratio micro-hole EDM was studied by Mas uzawa et al. [5], Takahata et al. [6], and Lim et a l. [7]. Improvement of micro-hole quality could be obtained by lower discharge energy [8]. Masuzawa [1] pointe d out that the key point for lower discharge energy w as the minimization of the stray capacitance betwee n the electrode and workpiece. Effects of polarity, elect rode shape, and rotational speed of electrode in mi cro-hole EDM drilling of carbide were investigated by Yan et al. [9] and the experimental results showed that p ositive polarity must be used in micro-hole EDM drilling to reduce tool wear and maintain micro-hole accuracy. The effects of two electrode materials, copper and tung sten carbide, on micro-hole EDM drilling were studi ed by Her et al. [10] and it was reported that the copper electrode could provide better surface roughness, lower electrode wear, but lower MRR than the tungsten car bide electrode. In this study, effects of energy pa rameter are investigated experimentally. A series of micro- hole machining were performed by keeping all parame ters constant except pulse energy. Variations in micro-h ole geometry, material removal rate, and micro-hole depth and over cut in micro-hole diameter were analyzed. 3.0 Materials and Data Acquisition Plastic mold steel (70x10x2 mm) is used as a wokpie ce material. Tungsten carbide (WC) electrodes with a standard diameter of 400 µ m and 100 µ m are used as a tool electrode. Dielectric liquid used for flushi ng is hydro carbide composed of mineral and synthetic oil s. An Agilent 54621D mixed signal oscilloscope with a value of 60MHz and 200 MSa/sec is used to display t he shape of voltage and current pulse forms. Low induc tance 0.2 ohm 1%R resistor is serially connected to discharge circuit to capture current pulse forms. 4.0 Experimental Results A series of experiments was performed to estimate t he undisclosed parameters defined in the micro-EDM machine. Those parameters only entered as a setting without a unit, which express the resultant pulse energy and current forms used during machining. Although d ischarge energy depends on pulse voltage and curren t, it is given in the machine settings as a type of pulse shapes that actually limits discharge energy. The current defined in the machine settings also carry the appr oximate meaning like in the energy. Machining condi tions are summarized in Table 1. 515 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES Table 1: Machining conditions for varying energy pa rameter Parameter type Parameter value Machining time (min.) 20 Frequency (kHz) 100 Width (µs) 4 Open Voltage (V) 80 Gap Voltage (V) 75 Gain 10 Temp. of dielectric liquid (ºC) ~20 Temp. of medium (room) (ºC) ~25 Energy type is classified into six different ranges by the machining tool manufacturer. Description of energy levels is given in Table 2. Table 2: Description of Energy parameters (Sarix op erating manual version 1.20) Energy Parameter Family Description From 13 to 15 Very short pulses From 100 to 114 Short pulses From 200 to 215 Long braked pulses From 250 to 265 Long, delayed, beaked pulses From 300 to 315 Long, delayed pulses From 350 to 365 Long pulses The voltage and current in micro-hole machining is monitored and recorded. The monitoring technique ca n assist in the selection and optimization of micro-h ole EDM process parameters. Shape of the pulse form s with different energy parameters are given in Figure 1. Selection of lower value of energy parameter resul ts in the lower value of discharge time and discharge current . Peak current value approximately rises to 20 A fo r long pulses and gradually decreases for lower energy set tings. It was observed that peak current decreases nearly to 5 A for very short pulses. Discharge time (pulse on time) is also longer about 2 µ s in long pulse shap e when comparing to short pulses of less than 200 ns. When the short pulses are considered, shapes of the pul ses are not recorded properly due to insufficient sampling rate of the oscilloscope. An advanced data acquisit ion device should be used to record and to keep the ver y short pulse shape in more reliable range. Three micro-holes were machined for each energy par ameter to observe the repeatability of the process. Machined micro-hole diameter is always larger than the tool electrode diameter; the difference is vary ing depending on the machining conditions. Expansion in the micro-hole diameter is defined as over-cut, caused by side spark erosion. Variation in micro-hole diam eters, electrode removal, standard deviation (STD) and over-cut are tabulated in Table 3. It was observed that, micro-hole expansion (over-cut) is slightly i ncreasing with respect to energy parameter which is varying f rom very short pulses to long pulses. This differen ce in over-cut is due to the difference in spark intensit y. Higher the energy value parameter leads to more intense spark and which erode more particles from the side surface of the micro-hole. 516 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES Fig. 1. Analyzed voltage and current pulse forms Magnified edge for each hole (Figure 2) is given to analyze the geometrical shape and edge surface rou ghness. Dimensional measurement for hole diameter and machi ned length of hole may sometimes not consistent wit h the diameter of other holes machined by the similar settings. Micro-hole diameter may be sometimes mea sured larger than the actual one. This inconsistency in t he measurement is caused by the measurement devices and software used in the microscopic system, focusing p roblems arouse while magnified pictures are taken a nd image processing error are observed while measuring dimensions. Cross-sectional views of the machined holes (Figure 3) are taken to analyze the geometrical shape and parallelism of the micro-hole wall side. A good sha pe can be obtained by using the entire machining conditions. One point in the cross-sectional view m ay draw attention about inconsistent cross-sectiona l dimension along the micro-hole radial axis that is induced by difficulty during sectioning process. Be cause of the very small size grinding and polishing process, it is difficult to control whether to reach the ce nter of the machined hole, sometimes due to over grinding or in sufficient grinding process; the approximate centra l cross- section is taken and analyzed. Actually, disregardi ng the dimensional measurement, the error during se ctioning process is not substantially affecting the overall geometrical shape of the machined holes. 517 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES Table 3: Micro-hole diameter variation with energy type Energy Target depth Removed electrode length Hole depth Hole diameter Mean diameter STD for hole diameter Average over-cut 1 8 0 31 150 447 165 31 130 427 14 1 5 5 25 130 430 435 10.91 17.58 843 161 682 438 858 168 690 426 105 8 5 0 157 693 446 437 10.07 18.3 897 237 660 435 880 230 650 433 114 8 7 0 224 650 429 433 2.65 16.1 272 0 115 0 157 0 457 272 0 114 0 158 0 458 205 2 8 2 0 118 0 164 0 441 452 9.34 26 291 0 165 0 126 0 461 306 0 178 0 128 0 443 250 3 1 5 0 179 0 134 0 437 448 12.33 23.5 340 0 166 0 174 0 451 348 0 172 0 176 0 451 305 3 4 6 0 171 0 175 0 446 449 2.78 25 340 0 146 0 194 0 457 338 0 143 0 195 0 464 350 3 2 5 0 140 0 185 0 451 457 6.71 28.5 • Dimensions are in µm a) E=14 b) E=105 c) E=205 d) E=250 e) E=305 f) E=350 Fig. 2. Micro-hole wall edge photographs with X1000 magnification with respect to energy level a) E=14 b) E=105 c) E=205 d) E=250 e) E=305 f) E=35 0 Fig. 3. Cross-sectional photographs of micro-holes machined with different energy level 518 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES The other most significant remark realized by means of this experimental work is the influence of the energy parameter while using smaller size electrode such a s Ø100 µ m. When Ø100 µ m electrode is used to machin e a micro-hole, very short pulses should be selected, o therwise, it become impossible to machine a correct drilled hole. Figure 4 explains the above statement explici tly, nearly aspect ratio of 17 is obtained by using very short pulses, whereas by using short pulses only a little cavity which is far from becoming micro-hole is ob tained. Fig. 4. Cross-sectional view of holes machined by Ø 100 µm electrode with a machining time of 60 min. and en ergy parameter for a) 14, b)105 and c) 114 5.0 Conclusion It is well known that machining accuracy of micro-E DM is limited by tool wear. The tool wear is charac terized by corner and end wear which mean tool material rem oval in radial and axial directions, respectively. Thus, blind-micro-holes can get different forms and it is difficult to be measured without additional machin ing. Tool wear is found to be high especially at the corners of the tool electrode. This result can be easily at tributed to the high discharge intensity on the corners. Gradua l forward movement of the tool electrode results in a formation of a narrow edge clearance between tool e lectrode and the workpiece. Thus, dielectric liqui d circulation became an important aspect since debris produced during machining alters dielectric liquid strength and therefore discharge conditions during machining . By means of the distorted flushing conditions, de bris particles can not be flushed away appropriately fro m the sparking area. High concentration of debris p articles in the sparking gap increases the chance of occurri ng secondary discharges, arcs and short circuits du e to decreased dielectric strength which are undesirable discharge phenomena for obtaining precious EDMed feature. From the pulse energy perspective, very short pulse s used during machining result in small debris part icles which provide proper flushing conditions from sharp tip edges. Thus, when smaller size electrode is us ed during machining, proper machining conditions can b e established. On the other hand, higher pulse ener gy used during machining result in high energy intensi ty on the tip of electrode and the energy on the el ectrode body immediately tends to discharge to the workpiec e and this phenomena occur consecutively many times without waiting for suitable conditions and which r esult in short circuit away from eroding workpiece. 519 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 EFFECT OF PULSE ENERGY IN MICRO ELECTRIC DISCHARGE MACNINING OF MICRO HOLES Acknowledgement This study is based on the research project entitle d “Development and implementation of micro machinin g (especially micro-EDM) methods to produce (design a nd manufacture) of mini/micro machines/robots”; funded by the State Planning Organization of Turkey (DPT). References [1] Masuzawa, T., 2000, State of the art of micromachin ing, Ann. CIRP 49 473-488 [2] Kim, D. J., Yi, S. M., Lee, Y. S., Chu, C. N., 2006 , Straight hole micro EDM with a cylindrical tool a variable capacitance method accompanied by ultrasonic vibrat ion, J. Micromech. Microeng. 16 1092-1097. [3] Moylat, S. P., Chandrasekar, S., Benavides, G. L., 2005, High-Speed Micro Electro Discharge Machining, Sandia Report, Sandia National Laboratories, SAND2005-5023 , printed September. [4] Masuzawa, T., Fujimato M. Kobayashi, K., 1985, Wire electro-discharge grinding for micromachining, Ann . CIRP 34 431-434. [5] Masuzawa, T., Kuo., C. L., Fujino, M., 1990, Drilli ng of deep microholes by EDM using additional capac ity, Bulletin of the Japan Society of Precision Engineer ing, 24(4), pp.275-276. [6] Takahata, K., Shibaike, N., Guckel, H., 2000, High- aspect-ratio WC-Co microstructure produced by the c ombination of LIGA and micro-EDM, Microsystem Technologies, 6( 5), pp175-178. [7] Lim, H. S., Wong, Y. S., Rahman, M., Edwin Lee, M.K ., 2003, A study on the machining of high-aspect ra tio micro- structures using micro-EDM, Journal of Materials Pr ocessing Technology, 140 (1-3) pp.318-325. [8] Allen, D. M., and Lechebeb, A., 1996, Micro electro -discharge machining of ink jet nozzles: optimum se lection of material and machining parameters, Journal of Mater ials Processing Technology, 58(1), pp.53-56. [9] Yan, B. H., Huang. F. Y., Chow, H.M., Tsai, J.Y., 1 999, Micro-hole machining of carbide by electric di scharge machining, Journal of Materials Processing Technolo gy, 87(1-3), pp.139-145. [10] Her, M. G. and Weng, F. T., 2001, Micro-hole machin ing of copper using the electro-discharge machining process with a tungsten carbide electrode compared with a c opper electrode, International Journal of Advanced Manufacturing Technology, 17(10), pp. 715-719. 520 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS Xichun Luo 1 , Wenlong Chang 1 , Jining Sun 1 , Ning Ding 1,2 , Chris Mack 1 1. School of Engineering & Physical Sciences, Heriot-Wa tt University, Edinburgh EH14 4AS, UK 2. College of Mechanical Engineering, ChangChun Univer sity, ChangChun, 130 0 2 2 , P.R. China Abstract Edge chipping is a major problem in glass drilling. The fabric ation of a new MEMS device requires micro-drilling a number of holes (800 µm diam eter) on Pyrex (7740) glass wafer. The edge qualities of these holes are crucial to the function of this new MEMS device. However, current available ultrasonic drilling tec hnique cannot meet the requirements on edge quality and positioning accuracy for d rilling these holes on the glass wafer. This paper presents the initial results in the developm ent of a near chipping-free micro-drilling process for glass wafer. The drilling tests are performe d on two machine tools using multilayered diamond drill, together with water as coolant. The cutting forces are measured by Kistler dynamometer and the drilled holes are inspected under a microscope. Finite Element simulation is carried out to opt imize machining parameters to minimize edge chipping. The theoretical and exp erimental studies show that the contact pressure on diamond grit has significant effects on crac k generation in the glass. Low contact pressure will result in short crack generation length. This will result in near chipping-free drilled surface. The high c oncentration of multilayered diamond drilling, use of water as coolant, high dynamic loop stiffn ess machine tool and optimized machining conditions are all contribute to low c ontact pressure on diamond grit and therefore a near chipping-free drilled holes on the glass wafer. Keywords: Micro-milling, MEMS, edge chipping, glass drilling. 1.0 Introduction The fabrication of a new MEMS device requires micro -drilling a number of holes (from 300 µ m to 3 mm diameter) on Pyrex (7740) glass wafer. The edge qua lities of these holes are crucial to the function o f this new 521 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS MEMS device. Two glass wafers will be bonded togeth er before the dicing operation. Therefore, it also requires 50 µ m positioning accuracy when drilling t hose holes in order to meet the alignment requireme nt of the MEMS devices. However, glass is a very difficul t material to be drilled as edge chipping is a majo r problem. Ultrasnoic drilling is an enable technique to minimize edge chipping but it is very difficult to achieve the required positioning accuracy. A cost-effective drilling process is needed to obtain near chipping -free edge quality and 50 µ m positioning accuracy in this glas s drilling process. This paper shows some preliminary results in develo ping this near chipping-free glass drilling process . Finite Element simulation is carried out to optimize machi ning parameters to minimize edge chipping. The dril ling tests are performed on two machine tools using mult ilayered diamond drill, together with water as cool ant. The cutting forces are measured by Kistler dynamometer and the drilled holes are inspected under a microsc ope. It summarized with a solution for obtaining near chipp ing-free edge quality in glass drilling. 2.0 Theoretical Basis for Precision Machining of Glass There have been tremendous efforts to precision mac hining brittle materials including glass by utilizi ng a so- called “brittle-to-ductile” transition phenomenon [ 1-7]. Lawn made a great contribution in finding “br ittle-to- ductile” transition in brittle materials. His inden tation test using a sharp point indenter on glass s howed the progression of the plastic and fracture region. He found that the surface plastic deformation takes pl ace at a certain loading. At some critical load and penetrat ion depth a median crack developed which continues to grow with additional applied load/penetration. On t he unloading process the mismatch stress causes lat eral cracks to grow. These lateral crakes can propagate towards the surface which causes large levels of su rface chipping to occur. Lawn has deduced functions to es timate the critical load of formation of median cra cks and the corresponding critical crack length [1]. Hagan [2] has carried out similar indentation work and de duced the critical load to nucleate a micro-crack just beneat h the elastic/plastic boundary and associated criti cal crack length. The equations are presented in the form of:     = 3 4 1 * H KCP c (1) 2 2 *     = H KCC c (2) where Kc and H are fracture toughness and hardness of glass. C 1 and C 2 are constants. For glass (Pyrex 7740), critical load P * for a median crack and corresponding critical crac k lengths are 0.023 N and 1.63 µm respectively. Obviously the value P * is a single point load, and therefore is less than the cutting force when multiple diamond grits are involved in drilling pro cesses. 522 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS Lawn also introduced an equation for the calculatio n of critical penetration depth for initiating a fr acture, it can be described as: 2         = H K H Ed cc ψ (3) where ψ is a dimensionless constant dependant on indenter geometry, E is the Young’s modulus. dc is often called critical depth of cut in literature. From re sults of ductile modes grinding tests, Bifano has d educted that ψ equals to 0.15 in his model [3]. He also concluded that maximum chip thickness should be less than th e critical penetration depth in order to achieve duct ile regime grinding and minimum subsurface damage. This glass drilling process will using multilayered diamond drills. The drill process is very similar to diamond grinding process using a cup wheel [4-7] although t he feed direction is vertical to the machined surfa ce in drilling process. So initially this research will b e focus on process development to find out the opti mized feed rate and grinding speed to achieve good edge qualit y with a material removal rate as high as possible. 3.0 Finite Element Study for Optimization of Machining Parameters 3.1 FEM Model Finite element is used to model the material remova l process by single diamond grit in order to find o ut the optimized machining parameters. The material behavi our of glass is modeled by Drucker-Prager damage model. Both plastic deformation and brittle fractur e are considered in this model. The constitutive fu nction for Drucker-Prager damage model is expressed as: ⋅ ⋅ Θ⋅Γ⋅= )()(),(),,,( 11 TJGTJ pp εεεεσ (4) Where ),( 1JF pε is the function for strain hardening and hydrostati c pressure. ⋅ Γ )(ε is strain rate sensitivity and )(TΘ is thermal softing function. The damage function is described as: ∑ ∆= i p f p i i D ε ε , (5) 523 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS where D is the dimensionless cummulative damage, ∆εip is the instantaneous increment of strain and εfip is the instantaneous strain to failure. This model is obta ined from a curve fit to strain to failure verus te mperature diagraph of glass [8]. The Drucker-Prager damage mo del will be compiled and loaded by the simulation software to calculate the glass material state thro ughout simulation. Orthogonal model is used to model the drilling proc ess by single diamond grit. Two sets of simulation experiments are devised. Simulaiton experiment 1 is designed to investigate the effect of dilling spin dle rotational speed on normal cutting forces. Rotation al speed of drilling spindle varying from 2000 rpm to 3500 rpm are used. They are equivalent rotational speeds of 5026 mm/min to 8796 mm/min for outer edge of a 0.8 mm diameter drills that is going to be used in the drilling test. In the simulation tests penetration depth is kept at 10 µ m. In Simulation 2, penetration depths of 12.5 µ m, 18 µ m, 25 µ m and 30 µ m are used. In both simula tion experiments, -45° rake angle diamond grit is ued. I ts clerance angle is assumed to be at 10°. A commercial finite element packge AdvantEdge is us ed for the numerical simulation of this orthogonal cutting process. Triangle element and adaptive remenshing t echnique are used to correct the problem of element distortion due to high deformations in simulation c omputation. 3.2 Simulation Results 3.2.1 Surface Generation Fig. 1 is a snapshot of simulated cutting process b y a single diamond grit which locates at the outer edge of the drill. Both plastic deformation and brittle fractur e can be observed in the formed chip. It indicates that the material removal is a combination of brittle fractu re and plastic deformation. This observation confor ms to the result of Scattergood’s experimental observation th rough “interrupted cut method” in single point diam ond turning. His test approves there exits a “brittle-d uctile” transition plane, under which the plastic d eformation takes place and ductile removal is achieved [9]. A lthough there are many diamond grits involved in ma terial Fig. 1. Snapshot of FE simulation 524 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS removal in the drilling process the simuilation stu dy can still be used to find out optimized machinin g parameters so as to minimize length of brittle frac ture that extend down to the “brittle-ductile” tran sition plane. 3.2.2 Effects of Machining Parameters The variations of normal cutting forces with cuttin g speed are shown in Fig.2. It can be seen that the normal cutting force are all bigger than the critical pres sure predicted by Eq. (1). But when a spindle rotat ional speed of 2500 rpm is used the normal cutting forces is th e smallest one within the test range. Therefore, th e FEM simulation results indicate that 2500 rpm is a good spindle rotational speed to be used for a 0.8 mm d iameter 0 0.0 05 0. 01 0.0 15 0. 02 0.0 25 0. 03 0.0 35 15 00 2 00 0 25 0 0 30 00 3 50 0 40 00 Spindle speed (rpm) No rm al cu tti n g fo rc e (N ) Fig. 2. Variation of normal cutting forces with spi ndle speed 0 0. 1 0. 2 0. 3 0. 4 0. 5 0. 6 0 5 10 15 20 25 30 3 5 Penatration depth (um) No rm al cu tti ng fo rc e (N ) Fig. 3. Variation of normal cutting forces with pe netration depth Fig. 4. Drilling test setup Fig. 5. (a) 3. 0 mm drill (b) 0.8 mm drill Spindle Diamond drill Dynamometer Glass wafter 525 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS drill. Fig. 3 shows the variation of normal cutting force dose not possess linear relationship with pe netration depth. When a penetration depth of 12.5 µ m is used it will obtain the smallest normal cutting forces i n the test range. Therefore, diamond drill with grit size of 2 5µ m is chosen in the following drilling test. 4.0 Glass Drilling Test 4.1 Experimental Condition Two set of drilling tests are devised. Test 1 is ca rried out on a conventional CNC machining centre Ta kang VMC-1202. The test setup is shown in Fig. 4. A piec e of glass is attached on the fixture by wax. Kistl er Dynamometer 9257BA is used to measure drilling forc es. 0.8 mm and 3.0 mm diameter drills are used in t est 1. Spindle speeds of 2000 rpm, 2500 rpm, 3000 rpm and 3500 rpm are used when feed rate is kept at 0.25 mm/min. When spindle speed is kept at 2500 rpm, fee d rates of 0.125, 0.18, 0.25 and 0.3 mm/min are use d. A state-of-the art micromachining centre Kern Evo is used in Test 2 to study the effects of spindle stif fness on edge quality in glass drilling. Another 3.0 mm diam eter drill which has the same specification as the one used in test 1 is adopted. In test 2 spindle speed of 25 00 rpm and feed rate of 0.25 mm/min are used. Water is used as coolant. The images of the two kinds of diamond drills are shown in Fig. 5. They are electroplated drills with three layers of diamonds. Diamond grit size is 25 µ m. The drilled holes are measured by OLYMPUS T041 Microscope. 4.2 Experimental Results and Discussions Fig. 6 and Fig. 7 shows the variations of normal cu tting forces against spindle rotational speed and f eed rate, which indicates the spindle speed of 2500 rpm and f eed rate of 0.25 mm/min are optimized machining parameters to drill glass. This has been further co nfirmed by the images of holes drilled under differ ent machining conditions which are shown in Fig. 8. Fig. 9 shows the image of a glass wafer with six ho les on it. The measured alignment accuracy by using Renishaw’s CMM is 10 µ m, which can meet the require ment. Fig. 10 shows that the hole drilled on Kern 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 1 50 0 20 00 25 00 3 00 0 3 50 0 40 0 0 Spindle speed (rpm) No rm a l c u tti ng fo rc e (N ) 0 0. 5 1 1. 5 2 2. 5 0.05 0.1 0.15 0.2 0. 25 0. 3 0.35 Feed rate (mm/min) No rm a l c u tti ng fo rc e (N ) Fig. 6 Variation of normal cutting forces with spin dle speed. Fig. 6 Variation of normal cutting forc es with feed rate. 526 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS (a) spindle speed = 2500 rpm, feed rate = 0.25 mm/m in (b) a) spindle speed = 3500 rpm, feed rate = 0.25 mm/min (c) spindle speed = 2500 rpm, feed rate = 0.125 mm/ min (d) a) spindle speed = 2500 rpm, feed rate = 0.3 mm/min Fig. 8 Comparison of edge chipping under different machining conditions . machine is obviously has less edge chipping than th at machined on Takang machine. It is because both a xial and radial stiffness of the HSK spindle on the Kern machine is bigger than spindle of Takang machine. The effects of spindle stiffness on edge chipping will be investigated in the future experimental study. 5.0 Conclusion Fig. 9. One completed glass wafer (a) on Kern mac hine (b) on Takang Fig. 10. Holes drilled on two machines 527 The 6 th International Conference on Manufacturing Research (ICMR08) Brunel University, UK, 9-11 th September 2008 MICRO-DRILLING HOLES ON GLASS WAFER FOR FABRICATION OF MEMS This paper presents the development of a near chipp ing-free micro-drilling process for glass wafer. Fi nite Element theoretical study has been carried out to g et an early indication of optimized machining param eters for obtaining good edge quality. The finding by FE simulation has been further proved by drilling test s. The experimental results show that near chipping-free h oles can be obtained by using high concentration of multilayered diamond drilling with 2500 rpm spindle speed and 0.25 mm/min feed rate. Drilling tests on two different types of machine tools show stiffness of spindle also has an important effect on edge qualit y. Acknowledgement The authors wish to acknowledge the assistance and support of Mr. John Hedge, Dr. Tan Jin, Dr. Enhao L i and Prof. Anthony Walton in the drilling test at Cranfi eld University and supply of glass wafer from Scott ish Electronics Centre. References [1] Lawn, B.R. Wilshaw, T.R.; Fracture of Brittle Solid s, Cambridge university press; 1975. [2] Hagan, J. T., Shear deformation under pyramidal ind entations in soda-lime glass, Journal of Materials Science, Vol. 15, pp.1417-1424, 1980. [3] Bifano, T.G., Dow, T.A., Scattergood, R.O., Ductile regime grinding – A new technology for machining b rittle materials; Journal of Engineering for Industry, Vol . 113, No.2, pp.184-189; 1991. [4] Malkin, S., Hwang, T.W., Grinding mechanism for cer amics, Annals of the CIRP, Vol. 45, pp. 569-578, 19 96. [5] Li, Y., Funkenbusch, P.D., Gracewski, S.M., Ruckma n, J., Tool wear and profile development in contour grinding of optical components, International Journal of Machin e Tools & Manufacture, Vol.44, pp.427-48, 2004 [6] Puttick, K.E., Jeynes, C., Whitmore, L., Rud man, M.R., Yamasaka, M., Shore, P., Gee, A.E., Surf ace damage in nanoground silicon, Proceeding of IMECH - Tribology in metal cutting and grinding, pp .49-51, 1992. [7] Yoshihara, N., Kuriyagawa, T., Ono, H., Syoji , K., Nano-topography on axisymmetric aspherical gr ound surfaces, International Journal of Manufacturing Tech nology and Management, Vol. 9, pp. 51-63, 2006. [8] Third Wave Systems, AdvantEdge User’s Manual, Version 5.1, (2008) 38. [9] W. S. Blackley and R. O. Scattergood, Chip to pography for ductile-regime machining of germanium, Transaction of the ASME: Journal of Engineering for Indust ry, Vol. 116, pp. 263-266, 1994. 528