Multi-Island Genetic-Algorithm-Based Approach to Uniquely Calibrate Polycrystal Plasticity Models for Magnesium Alloys

Abstract

The single-crystal yielding and hardening behavior of polycrystals is important for understanding their mechanical behavior. Many parameters (> 10) are usually required to achieve this for magnesium alloys based on physics-based models. However, the efficient and precise determination of these parameters is a very challenging task. An efficient and practical method is proposed herein to determine the parameter set by dividing the parameters into those for yielding and hardening, thereby significantly reducing the time cost of automatic parameter calibration. This method is then applied to calibrate the parameter set used in the viscoplastic self-consistent (VPSC) model to describe the mechanical behavior of the rare-earth magnesium alloy ZEK100 from multiple mechanical test data. The obtained best-fit parameters can be considered realistic and nearly unique, and can successfully reproduce the mechanical behavior and textural evolution. This procedure can be generally applied to calibrate near-unique parameters in other materials and constitutive models.