About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Computational Thermodynamics and Kinetics
|
| Presentation Title |
From Empirical to Machine-Learned Potentials: More than Half a Century of Actinide Oxide Simulation |
| Author(s) |
Michael J D Rushton |
| On-Site Speaker (Planned) |
Michael J D Rushton |
| Abstract Scope |
Thermodynamic and kinetic modelling of actinide oxide nuclear fuels often relies on atomistic simulation, which depends on an accurate description of the forces acting between atoms. Early pair potentials offered computational efficiency but were limited in their ability to reproduce defect energies alongside thermomechanical properties of these materials. As one of the original developers of the Cooper-Rushton-Grimes many-body potential, introduced over a decade ago to address these limitations, this talk will reflect on its development and its successes. The talk considers what this experience reveals about the broader trajectory of interatomic potential development for actinide oxides and looks ahead to the present inflection point in the field, where machine-learned interatomic potentials offer the prospect of density functional theory levels of accuracy at classical potential speed, and what this may bring for the future simulation of nuclear fuel materials. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Nuclear Materials, Modeling and Simulation, Mechanical Properties |