The ICSCREAM methodology: identification of penalizing configurations in computer experiments using screening and metamodel applications in thermal-hydraulics

voir aussi article : https://hal.archives-ouvertes.fr/hal-02535146Nuclear Science and Engineering, Volume 196, 2022 - Issue 3International audienceIn the framework of risk assessment in nuclear accident analysis, best-estimate computer codes, associated to a probabilistic modeling of the uncertain input variables, are used to estimate safety margins. A first step is often to identify the critical configurations (or penalizing, in the sense of a prescribed safety margin) of several input parameters (called “scenario inputs”), under the uncertainty on the other input parameters. However, the large CPUtime cost of most of the codes involves to develop highly efficient strategies. To achieve it with a very large number of inputs and from a small-size sample of simulations, a specific and original methodology, called ICSCREAM (Identification of penalizing Configurations using SCREening And Metamodel), has been proposed. The screening of influential inputs is based on an advanced global sensitivity analysis indices, namely the Hilbert-Schmidt Independence Criterion. Then, a Gaussian process metamodel is sequentially built and used to estimate, within a Bayesian framework, the conditional probabilities of exceeding a high-level threshold, according to the scenario inputs. The efficiency of this methodology is illustrated on a high-dimensional (hundred inputs) use case simulating an accident of primary coolant loss in a pressurized water reactor. The study focuses on the peak cladding temperature (PCT) and critical configurations are defined by exceeding the 90%-quantile of PCT