This Paper addresses a two-step computational approach to building a robust modeling environment for the titanium–aluminum alloy, Ti-7Al, a candidate aerospace material owing to superior mechanical performance under high stresses. To be used in aerospace applications, the large deformation behavior of the alloy should be investigated with a high-fidelity crystal plasticity model. However, there is no universal agreement on the crystal plasticity parameters, and previous efforts are only based on deterministic techniques. Therefore, our goal is to build a crystal plasticity model for Ti-7Al, which is validated for the global (component-scale) and local (grain-level) features by considering the experimental uncertainty. In the first step, the lower and upper bounds of the crystal plasticity parameters are determined with an inverse problem that is solved to match the computations with the experimental stress-strain data. The second step validates the local features by solving an optimization problem that minimizes the difference between the simulated and experimental microstructural textures. The optimization is performed using an Artificial Neural Network (ANN-)based surrogate model that is trained within the lower and upper limits of the parameters obtained in the first step. The outcomes of the two-step approach demonstrate significant improvement over the previous deterministic solutions.
Machine Learning Reinforced Crystal Plasticity Modeling Under Experimental Uncertainty
AIAA Journal ; 58 , 8 ; 3569-3576
2020-08-01
Article (Journal)
Electronic Resource
English
Epistemic Uncertainty Modeling of Johnson-Cook Plasticity Model Using Evidence Theory
British Library Conference Proceedings | 2010
|