Comparison of bootstrapped artificial neural networks and quadratic response surfaces for the estimation of the functional failure probability of a thermal–hydraulic passive system

In this work, bootstrapped artificial neural network (ANN) and quadratic response surface (RS) empirical regression models are used as fast-running surrogates of a thermal–hydraulic (T–H) system code to reduce the computational burden associated with estimation of functional failure probability of a T–H passive system. The ANN and quadratic RS models are built on a few data representative of the input/output nonlinear relationships underlying the T–H code. Once built, these models are used for performing, in reasonable computational time, the numerous system response calculations required for failure probability estimation. A bootstrap of the regression models is implemented for quantifying, in terms of confidence intervals, the uncertainties associated with the estimates provided by ANNs and RSs. The alternative empirical models are compared on a case study of an emergency passive decay heat removal system of a gas-cooled fast reactor (GFR).