About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithm Development in Materials Science and Engineering
|
Presentation Title |
Monte Carlo Based Uncertainty Quantification of Crystal Plasticity Simulations Using ExaConstit |
Author(s) |
Venkata Sai Harshit Gaddam, Robert Carson, Jim Belak, Michael D Sangid |
On-Site Speaker (Planned) |
Venkata Sai Harshit Gaddam |
Abstract Scope |
Uncertainty quantification (UQ) of crystal plasticity (CP) simulations is important to identify risk and reliability of predicted results. To date, due to computational cost involved in running CP simulations, UQ analysis has traditionally been conducted on surrogate models of CP simulations. Surrogate models can introduce some level of uncertainty which can change the results of the UQ analysis. Utilizing exascale HPC systems and ExaConstit, an open-source CP code developed by LLNL, we can efficiently complete hundreds of thousands of CP simulations. With this framework, we can then perform Monte Carlo (MC) sampling of the CP input parameters to identify uncertainties and sensitivities of the initial parameterization of the model. We will also discuss several techniques used to reduce the computational time associated with the MC simulations, without compromising accuracy. |
Proceedings Inclusion? |
Planned: |
Keywords |
Modeling and Simulation, Other, Other |