Multi-scale investigation of highly anisotropic zinc alloys using crystal plasticity and inverse analysis

Abstract

Zinc and its alloys are important industrial materials due to their high corrosion resistance, low cost and good ductility. However, the characterization of these materials remains a difficult task due to their highly anisotropic behavior, the latter being due to the influence of microstructural effects, i.e. loading orientation-dependent activation of different families of slip systems and subsequent texture evolution, rendering the development of a reliable material model considerably difficult. A micro-mechanical approach based on polycrystal plasticity would better describe the physical mechanisms underlying the macroscopic behavior. This improved model should ostensibly improve the comprehension of the mechanical behavior, compared to the macroscopic approach using solely phenomenological anisotropy models along with a prohibitively large number of experiments required to identify the material parameters. In this paper, a multi-scale Visco-Plastic Self-Consistent (VPSC) approach is used. It is based on a micro-scale model calibrated by microstructural and deformation mechanism information based on Electron Back-Scattered Diffraction (EBSD) to describe the macroscopic anisotropic mechanical response during sheet metal deformation. The critical resolved shear stress (CRSS) as well as the micro-scale crystal parameters are obtained by an inverse analysis comparing the simulated and experimental results in terms of obtained tensile curves along three different directions. In order to obtain a global solution for the identification, we then use the Covariance Matrix Adaptation-Evolution Strategy (CMA-ES) genetic algorithm to the inverse problem. We validate our approach by comparing the simulated and experimental textures and activated slip systems. Finally, the identified mechanical parameters are used to investigate the anisotropy of the alloy and predict its formability by determining the corresponding R-values and Hill yield coefficients.