The safety case for continuing operation of nuclear reactors requires reliable assessment of the likelihood of the coolant temperatures exiting the fuel channels exceeding certain critical levels. Temperature measurements are typically made at a fixed sample of fuel channels and used for reactor control. No sample measurements will exceed the predetermined control limit, whereas it is likely that some of the unobserved temperatures will exceed this limit. The challenge is to use the control measurements reliably to assess the risk of the critical temperature exceedance over all channels, while also accounting for the uncertainties in the risk estimation.A novel non-stationary extreme value mixture modelling technique is developed to provide rigorous extrapolation of the risk past the observed range of the sample data. The proposed technique builds upon previous deterministic and statistical approaches, while providing more accurate risk predictions with fuller account for uncertainties in the estimation. A spatial random effects model is used to capture the non-stationary spatial structure in the temperature variation across the reactor. Bayesian inference for the extreme value mixture model parameters is undertaken, which permits assessment of all sources of uncertainty including potential inclusion of expert prior information.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.