Non-Intrusive Uncertainty Quantification for U3Si2 and UO2 Fuels with SiC/SiC Cladding using BISON for Digital Twin-Enabling Technology

U.S. Nuclear Regulatory Committee (NRC) and U.S. Department of Energy (DOE) initiated a future-focused research project to assess the regulatory viability of machine learning (ML) and artificial intelligence (AI)-driven Digital Twins (DTs) for nuclear applications. Advanced accident tolerant fuel (ATF) is one of the priority focus areas of the DOE/ NRC. DTs have the potential to transform the nuclear energy sector in the coming years by incorporating risk-informed decision-making into the Accelerated Fuel Qualification (AFQ) process for ATF. A DT framework can offer game-changing yet practical and informed solutions to the complex problem of qualifying advanced ATFs. However, novel ATF technology suffers from a couple of challenges, such as (i) Data unavailability; (ii) Lack of data, missing data; and (iii) Model uncertainty. These challenges must be resolved to gain the trust in DT framework development. In addition, DT-enabling technologies consist of three major areas: (i) modeling and simulation (M&S), covering uncertainty quantification (UQ), sensitivity analysis (SA), data analytics through ML/AI, physics-based models, and data-informed modeling, (ii) Advanced sensors/instrumentation, and (iii) Data management. UQ and SA are important segments of DT-enabling technologies to ensure trustworthiness, which need to be implemented to meet the DT requirement. Considering the regulatory standpoint of the modeling and simulation (M&S) aspect of DT, UQ and SA are paramount to the success of DT framework in terms of multi-criteria and risk-informed decision-making. In this study, the adaptability of polynomial chaos expansion (PCE) based UQ/SA in a non-intrusive method in BISON was investigated to ensure M&S aspects of the AFQ for ATF. This study introduces the ML-based UQ and SA methods while exhibiting actual applications to the finite element-based nuclear fuel performance code