Predicting Plastic Anisotropy Using Crystal Plasticity and Bayesian Neural Network Surrogate Models

In this work we integrate Variational Bayesian Inference techniques into feed forward neural networks to establish a data-driven and low-computational cost surrogate model capable of accurately predicting plastic anisotropy from initial crystallographic texture. Furthermore, the developed model is capable of quantifying the uncertainty on the predicted plastic anisotropy values. The model was trained on 54,480 crystal plasticity simulations results that characterized (and parameterized) the plastic anisotropic behavior of single crystal and polycrystalline textures, which were robustly represented using generalized spherical harmonics (GSH), using Hill’s anisotropic yield model. The trained Bayesian neural network linked the GSH-based representation of the different textures to their corresponding Hill’s anisotropic coefficients. The trained model was critically validated with 20,000 new textures. Finally, the predictions obtained with the trained Bayesian neural network model showed excellent agreement with results obtained from experiments and high-fidelity crystal plasticity finite element simulations.<br/>This work was supported by the Laboratory Directed Research and Development program at Sandia National Laboratories. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy National Nuclear Security Administration under contract DE-NA0003525. The views expressed in the article do not necessarily represent the views of the U.S. Department of Energy or the United States Government. Sand no. SAND2021-13152 A