Development of a computational optimisation system to determine constitutive equation constants for metal alloys

Abstract

Simulations of metal forming processes typically use constitutive equations to describe material properties over a wide range of temperatures and strain rates. Classically, determining a set of values for the large number of constants present in more complex constitutive models presented great difficulty to researchers. In this paper, an algorithmic method of determining the constants of such constitutive equations is presented. The Genetic Algorithm implementation utilising MATLAB is described, and example fits to experimental data are presented. A study is performed to improve the performance of the method by combining the Genetic Algorithm and Particle Swarm optimisation methods into a hybrid algorithm and to exploit the inherent advantages of both methods of solution. Comparative performance between the GA-only and hybrid algorithm are presented, and it is found that such a Hybrid Solver is significantly more computationally efficient and produces more accurate parameter estimations.