About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Artificial Intelligence Applications in Integrated Computational Materials Engineering (AI-ICME)
|
| Presentation Title |
Robust Physics-Informed Neural Networks for Modeling Oxide Film Growth in Corrosion Science |
| Author(s) |
Conrard Giresse Tetsassi Feugmo |
| On-Site Speaker (Planned) |
Conrard Giresse Tetsassi Feugmo |
| Abstract Scope |
Physics-informed neural networks (PINNs) are an emerging AI-driven approach in materials engineering, enabling direct incorporation of physical laws into neural network models for solving complex multiphysics problems. We apply PINNs to simulate the growth of oxide films at metal interfaces using the point defect model, confronting realistic challenges such as stiff, coupled equations and moving boundaries representative of corrosion processes. Benchmarking against finite element methods, we assess PINN performance, systematically tackle failure modes, and demonstrate that precise engineering—employing loss weighting, equation scaling, and curriculum training—yields reliable predictions. Our work offers actionable best practices for deploying PINNs in materials science, emphasizing their promise for accurate interfacial modeling as well as their integration within Integrated Computational Materials Engineering frameworks. These findings support the next generation of interpretable, data-driven design tools for high-impact materials discovery. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Machine Learning, ICME |