Machine Learning Model to Predict Failure Modes of TRISO Particles under High-Temperature Gas-Cooled Reactor Accident Conditions

Tristructured istropic particles, which contain a central kernel of uranium oxycarbide and are surrounded by a multilayered ceramic shell, are the fuel form proposed for high temperature gas-cooled reactors (HTGRs) and very high temperature gas-cooled reactors (VHTRs). These next generation nuclear reactors are proposed for operation at temperatures in excess of 600°C. Though He is the primary fuel coolant, the fuel form could see appreciable amounts of moisture, oxygen, carbon monoxide and carbon dioxide during certain accident scenarios. The exposure to a highly corrosive, mixed gas atmosphere can impact the integrity of the fuel form. The research presented incorporates machine learning approaches using the established thermodynamics of SiC-oxidant reactions and experimental data acquired to enable the prediction of catastrophic reactions, failure atmospheres, and SiC reactions with gas mixtures relevant to certain HTGR and VHTR accident scenarios.