Models, methods and algorithms for supply chain planning

Abstract

An outline of supply chains and differences in the problem types is given. The motivation for a generic framework is discussed and explored. A conceptual model is presented along with it application to real world situations; and from this a database model is developed.

A MIP and CP implementations are presented; along with alternative formulation which can be use to solve the problems. A local search solution algorithm is presented and shown to have significant benefits.

Problem instances are presented which are used to validate the generic models, including a large manufacture and distribution problem. This larger problem instance is not only used to explore the implementation of the models presented, but also to explore the practically of the use of alternative formulation and solving techniques within the generic framework and the effectiveness of such methods including the neighbourhood search solving method.

A stochastic dimension to the generic framework is explored, and solution techniques for this extension are explored, demonstrating the use of solution analysis to allow problem simplification and better solutions to be found. Finally the local search algorithm is applied to the larger models that arise from inclusion of scenarios, and the methods is demonstrated to be powerful for finding solutions for these large model that were insoluble using the MIP on the same hardware.