Predicting the In-Plane Mechanical Anisotropy of 7085 Aluminum Alloys Through Crystal Plasticity Simulations and Machine Learning

Abstract

The crystallographic texture of rolled 7xxx series aluminum alloys may lead to mechanical properties anisotropy which undermines the formidability and performance of the material. The key to the problem lies in lacking an in-depth understanding of the quantitative relationship between texture and the in-plane anisotropy of the r-value. In this work, by using the integrated computational method combining crystal plasticity simulations with machine learning and experiment, we established an analytical model and trained a machine learning model to predict the anisotropy of rolled sheets. We first performed a full-field crystal plasticity spectral simulations using fast Fourier transformation (CPFFT) to predict the anisotropy of 7085 aluminum alloy at different annealing times. The values of anisotropy predicted by CPFFT simulations agree very well with experimental measurements, and are more accurate than those predicted by the linear mixing model. Then, based on the CPFFT simulation data, the quantitative relationship between texture and the in-plane anisotropy of the r-value was established by using the multiple linear regression and gradient boosting regressor model. By comparing with experimental results, it has been proven that our models can well describe the quantitative relationship between planar anisotropy and texture of 7085 aluminum alloy, providing effective guidance for improving the performance of aluminum alloy sheets.

Keywords: Crystallographic texture, In-plane anisotropy, Crystal plasticity simulation, Machine Learning, Structure-property relationship