Intelligent processing and development of high-performance automotive aluminum alloys: Application of physics-based and data-driven modeling

Abstract

Drawing from recent developments in data analytics, machine learning, and computational capacity, this chapter aims to review different modeling approaches to support the development of a new generation of aluminum extrusion alloys with enhanced performance for the automotive industry. The effectiveness of physics-based models for predicting mechanical properties in these alloys requires an in-depth understanding of how critical microstructural features, such as crystallographic texture and precipitates, evolve throughout the different steps of the processing route. The chapter addresses the application of finite element methods to model the extrusion process focusing on predicting the development of crystallographic texture. Moreover, developments in modeling of clustering and precipitation hardening are discussed. Multistage aging and the effect of plastic deformation on precipitation kinetics are also considered within the framework of mean-field models to describe the interaction between aging and deformation. We examine different data-driven approaches for predicting mechanical properties as an alternative to traditional alloy and process design methods. Finally, a potential extension of such models is proposed based on a combination of conventional physics-based modeling of precipitation hardening with data-driven methods. Such machine learning modeling would thus offer an opportunity to close the gap regarding developing a conceptual understanding of the physical systems and quantifying the complex (and often unknown) interrelationships between parameters to enhance the capabilities of physics-based models.