About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Modeling, AI Applications, and Method Development in Nuclear Materials
|
| Presentation Title |
Machine Learning-Enhanced Process Modeling Framework for Solid-Phase Processing of Nuclear Cladding Materials |
| Author(s) |
Lei Li, Shadab Anwar Shaikh, Lukasz Kuna, Xi Tan, Aishani Sharma, Jens Darsell, Xiao Li, Eda Aydogan, Stuart Maloy, Ayoub Soulami |
| On-Site Speaker (Planned) |
Lei Li |
| Abstract Scope |
Solid-phase processing (SPP) enables energy-efficient, single step fabrication of advanced nuclear cladding by avoiding melting. Severe plastic deformation and frictional heating generate high strains and elevated sub-solidus temperatures that drive dynamic recrystallization and promote microstructural homogeneity. Yet, linking SPP process parameters to process thermomechanical conditions and final component performance still relies on expensive, time-consuming experiments, leaving much of the process space unexplored. This study presents a machine learning–enhanced process modeling framework that efficiently connects process parameters to process conditions. A smoothed particle hydrodynamics model is used to predict transient thermomechanical fields during processing. To accelerate field prediction and enable rapid exploration of the design space, an autoencoder-based Transformer surrogate model is integrated into the workflow. The framework is demonstrated for friction extrusion of oxide dispersion–strengthened (ODS) steels, providing a mechanistic basis to guide processing toward improved strength and irradiation resistance and to optimize friction-based SPP of ODS alloys. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Modeling and Simulation, Nuclear Materials, Iron and Steel |