To improve the accuracy of a sheet metal forming simulation, the constitutive model is calibrated using results from multiaxial material testing. However, multiaxial material testing is ...time-consuming and requires specialized equipment. This study proposes two different deep neural network (DNN) approaches, a two- and three-dimensional convolutional neural network (DNN-2D and DNN-3D), to efficiently estimate biaxial stress-strain curves of aluminum alloy sheets from a digital image representing the sample's crystallographic texture. DNN-2D is designed to estimate biaxial stress-strain curves from a digital image of {111} pole figure, while DNN-3D estimates the curves from a 3D image of the texture. The two DNNs were trained using synthetic texture datasets and the corresponding biaxial stress-strain curves obtained from crystal plasticity-based numerical biaxial tensile tests. The accuracy of the two trained DNNs was examined by comparing the results from that of the numerical biaxial tensile tests. It was observed that both the DNNs could estimate biaxial stress-strain curves with high accuracy. Though DNN-3D provides with a better estimation than DNN-2D, it displays lower computational efficiency. Thus, the two DNNs and their training procedures offer a new and efficient method to provide virtual data for material modeling.
Display omitted
•DNNs were used to estimate biaxial stress-strain curves of aluminum alloy sheets.•Pole figure images and 3D orientation maps were explored as input data.•DNNs were as accurate as numerical biaxial tensile tests, but much faster.•A new approach to virtual data generation for material modeling was demonstrated.
•Develop a DNN surrogate model of crystal plasticity-based numerical material test.•Propose Bayesian texture optimization using DNN and Bayesian optimization.•Crystallographic textures were optimized ...to reduce in-plane anisotropy of r-value.•Bayesian texture optimization provide solution space for finding desirable textures.
The formability of an aluminum alloy sheet can be improved by optimizing its crystallographic texture. Computational methods for texture optimization that combine crystal plasticity simulations with mathematical optimization algorithms are computationally inefficient. The crux of the problem is that conventional texture optimization strategies rely on multiple time-consuming crystal plasticity simulations. In this paper, we propose a new computational method for mitigating computational effort in numerical crystallographic texture optimization. The key point of the proposed method is that it achieves a significant speed-up factor of approximately three-fold. First, we propose a deep neural network-based approach for the computationally efficient estimation of mechanical properties based on the crystallographic texture. Second, we adopted Bayesian optimization to deal with a small number of trials robustly and efficiently. It is shown that the proposed computational method, christened Bayesian texture optimization, enables the determination of optimal volume fractions of preferred texture components to obtain a plastically isotropic aluminum alloy sheet. Moreover, unlike conventional methods, Bayesian texture optimization provides a framework that enables a profound understanding of the solution space that may consist of other desirable textures and associated uncertainties. Bayesian texture optimization paves the way for useful engineering tools that can improve the mechanical properties and formability of aluminum alloy sheets.
Display omitted
Bayesian optimization (BO) has attracted attention in various research fields as a powerful probabilistic approach for solving optimization problems. To further facilitate the use of BO, we developed ...a graphical user interface-based Python application called BOXVIA. BOXVIA enables the use of BO without the construction of a computing environment and/or the need for programming skills. Moreover, BOXVIA helps users interpret the results of the BO process effectively through certain useful functionalities available for visualizing the mean function, standard deviation, and acquisition functions.
PDF
