Advanced methodology for uncertainty propagation in computer experiments with large number of inputs

International audienceComplex computer codes, as the ones used in thermal-hydraulic accident scenario simulations, are often too time expensive to be directly used to perform uncertainty propagation. A solution to cope with this problem consists in replacing the cputime expensive computer model by a cpu inexpensive mathematical function, calledmetamodel. Among the metamodels classically used in computer experiments, the Gaussian process model has shown strong capabilities to solve practical problems.However, in case of high dimensional experiments (with typically several tens of inputs), the Gaussian process metamodel building process remains difficult. To face this limitation, we propose a methodology which combines several advanced statistical tools initial space-filling design, screening to identify the non-influentialinputs, Gaussian process metamodel building with the group of influential inputs as explanatory variables. The residual effect of the group of non-influential inputs is captured by another Gaussian process metamodel. From this joint metamodel, uncertainty propagation (here 95%-quantile estimation) can be performed.The efficiency of the methodology is illustrated on a thermal-hydraulic calculation case simulating accidental scenario in a Pressurized water Reactor. More precisely, a Loss Of Coolant Accident (LOCA) is considered, which takes into account a double-ended guillotine break with a specific size piping rupture