| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Atomistic Simulations Linked to Experiments to Understand Mechanical Behavior: A MPMD Symposium in Honor of Professor Diana Farkas
|
| Presentation Title |
Multiscale Models for Materials at Extreme Conditions Using Physics-Informed Machine Learning |
| Author(s) |
Alejandro Strachan |
| On-Site Speaker (Planned) |
Alejandro Strachan |
| Abstract Scope |
Large-scale molecular dynamics simulations of fcc alloys subjected to high strain rate deformation revealed a change in the underlying mechanism of plastic deformation with load path. Uniaxial tension along [100] is dominated by dislocation slip and we observe a transition to twinning and grain refinement based on the interaction of twins as the deformation path is continuously morphed into biaxial tension on the (001) plane. Not surprisingly, current models used in continuum simulations cannot capture the stress-strain behavior across these regimes. We tackled this challenge using physics-informed machine learning. A custom architecture is trained to predict stress in terms of the current state of the system and a strain increment. The resulting ML models can capture the response of materials at extreme conditions for a wide range of deformations, beyond what is possible with the current state of the art. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Modeling and Simulation, Mechanical Properties |