Statistical Failure Analysis of TRISO Fuel
Safety analysis of fuels can be critical to ensure the optimal performance and reliability of advanced nuclear reactors. However, given the uncertainties in the fuel properties, fuel safety analysis should be conducted in a probabilistic fashion by accounting for such uncertainties. The outputs of such a statistical failure analysis can be important quantities that may govern the design of fuels such as the probability of failure. Additionally, owing to computational expense of fuel models and low values of fuel failure probabilities (of the order 1E-4 to 1E-7), conducting a full-blown Monte Carlo simulation can be computationally prohibitive. Therefore, this section discusses both standard and advanced Monte Carlo algorithms available in BISON to perform a statistical failure analysis of fuels. The TRISO fuel is considered as an example.
Standard Monte Carlo methods
Standard methods such as crude Monte Carlo and Latin Hypercube Sampling are discussed. The efficacy of these methods to simulate fuel failure under small failure probabilities is low.
Variance reduction Monte Carlo methods
Variance reduction methods reduce the number of fuel model simulations by two to three orders of magnitude when estimating the failure probability. The efficacy of these methods to simulate fuel failure under small failure probabilities is moderate.