| About this Abstract |
| Meeting |
MS&T23: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Physics Informed Reduced Order Model for Directed Energy Deposition Simulations in MALAMUTE |
| Author(s) |
Anant Raj, Hany Abdel-Khalik, Luis Nunez, Yifeng Che, Wen Jiang, Rongjie Song |
| On-Site Speaker (Planned) |
Anant Raj |
| Abstract Scope |
High-fidelity multiphysics codes provide a pathway for process optimization and control in additive manufacturing, relaxing the reliance on expensive experiments. However, the computational cost of modeling the complex coupling between the multiple physics impedes the application of these tools on compute-intensive tasks. In this work, we develop a physics-informed reduced order model (ROM) for the MOOSE-based melt pool modeling code MALAMUTE. Unlike purely data-driven ROMs, which either replace the physics models with parametric surrogate models or project the equations onto a few active degrees of freedom, our approach preserves the dominant physics by leveraging the gradient information from the MOOSE solvers. Our approach is anticipated to reduce the computational cost for repeated execution of melt pool simulations for directed energy deposition, necessary to make the simulations tractable for control applications. |