About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Accelerated Discovery and Insertion of Next Generation Structural Materials
|
| Presentation Title |
Accelerating W-Alloy Design with Physics-Guided Machine Learning |
| Author(s) |
Prashant Singh, Sai Pranav Reddy Guduru, Sougata Roy, Mkpe O. Kekung, Ryan Ott, Duane D Johnson, Luke E Gaydos, Hailong Huong, Gaoyuan Ouyang, Andrew Kustas |
| On-Site Speaker (Planned) |
Prashant Singh |
| Abstract Scope |
The stability of tungsten-based structural alloys under extreme conditions is critical to the reliability of plasma-facing components in fusion reactors such as ITER and DEMO. However, tungsten’s inherently high ductile-to-brittle transition temperature (DBTT), irradiation-induced hardening, and neutron driven transmutation significantly constrain its long-term performance. This work presents a physics informed machine learning (ML) framework to accelerate the discovery of W-rich alloys with enhanced mechanical resilience and radiation tolerance. By integrating data from density functional theory (DFT) and thermodynamic modeling (CALPHAD), the ML models learn complex composition–structure property relationships that govern strength–ductility trade-offs and defect evolution under irradiation. This framework enables rapid screening of alloy chemistries optimized for improved room-temperature ductility, reduced DBTT, and greater radiation resistance, while maintaining thermal stability. Predictions are validated using electronic-structure calculations and available experimental data. Embedding physics-based priors into ML ensures interpretable, robust property predictions, advancing multiscale alloy design for extreme fusion energy environments. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
High-Temperature Materials, Modeling and Simulation, Machine Learning |