Hojun Lim1,Taejoon Park2,David Montes de Oca Zapiain1,Farhang Pourboghrat2
Sandia National Laboratories1,The Ohio State University2
Hojun Lim1,Taejoon Park2,David Montes de Oca Zapiain1,Farhang Pourboghrat2
Sandia National Laboratories1,The Ohio State University2
Typical plastic anisotropy characterizations of metal alloys require repetitive mechanical tests in various loading directions and stress states. To efficiently characterize and predict plastic anisotropy without extensive mechanical tests, crystal plasticity finite element method (CPFEM) simulations using initial microstructural data from EBSD and XRD measurements are performed and compared with experiments. It is shown that CPFEM model incorporating the texture and grain morphology of various aluminum alloys captures anisotropic mechanical behavior and r-values reasonably well. In addition, ~70,000 crystal plasticity data were generated to train a neural network model that instantly and accurately predicts normalized yield stresses and r-values from the initial texture. The model is then used to provide anisotropy parameters of various yield models. This capability provides fast and accurate prediction of material’s anisotropic response without involving extensive experimental characterization or expensive computational calculations.