About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Materials and Chemistry for Molten Salt Systems
|
| Presentation Title |
Improved Predictions of Thermal Transport in Molten Salts based on Machine Learning Interatomic Potentials (MLIPs) |
| Author(s) |
Isaac Walker, Jacob Numbers, Nathan Burlett, Anthony Birri, Troy Munro |
| On-Site Speaker (Planned) |
Isaac Walker |
| Abstract Scope |
Because of the difficulty in experimentally measuring the thermophysical properties of molten salts, there is significant interest in computationally efficient methods to predict properties such as thermal conductivity. Based on recent work to characterize the mean free path of energy carriers based on salt structure, we present an integrated structural coherence and machine learning interatomic potential (MLIP) approach to predict thermal conductivity based on kinetic theory. The chloride-focused semi-universal SuperSalt MLIP is used in classical molecular dynamics simulations to compute the volumetric heat capacity, speed of sound, bulk modulus, and mean free path. The predicted thermal conductivity agrees well with existing experimental data, except in the case of magnesium-bearing salts due to a discrepancy in the predicted speed of sound. This integrated workflow shortens computation time and can provide insight into the structure-property relationships within molten salts. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Nuclear Materials, Modeling and Simulation, |