Statistical Methodology for a Quantified Validation of Sodium Fast Reactor Simulation Tools

Abstract

This paper presents a statistical methodology for a quantified validation of the OCARINa simulation tool, which models the unprotected transient overpower (UTOP) accidents. This validation on CABRI experiments is based on a best-estimate plus uncertainties (BEPU) approach. To achieve this, a general methodology based on recent statistical techniques is developed. In particular, a method for the quantification of multivariate data is applied for the visualization of simulator outputs and their comparison with experiments. Still for validation purposes, a probabilistic indicator is proposed to quantify the degree of agreement between the simulator OCARINa and the experiments, taking into account both experimental uncertainties and those on OCARINa inputs. Going beyond a qualitative validation, this work is of great interest for the verification, validation and uncertainty quantification or evaluation model development and assessment process approaches, which leads to the qualification of scientific calculation tools. Finally, for an in-depth analysis of the influence of uncertain parameters, a sensitivity analysis based on recent dependence measures is also performed. The usefulness of the statistical methodology is demonstrated on CABRI-E7 and CABRI-E12 tests. For each case, the BEPU propagation study is carried out performing 1000 Monte Carlo simulations with the OCARINa tool, with nine uncertain input parameters. The validation indicators provide a quantitative conclusion on the validation of the OCARINa tool on both transients and highlight future efforts to strengthen the demonstration of validation of safety tools. The sensitivity analysis improves the understanding of the OCARINa tool and the underlying UTOP scenario.