An empirical method for inverse uncertainty quantification in nuclear thermal–hydraulic codes

The quantification of the uncertainty in the output of the thermal–hydraulic and severe accident codes used for the safety analysis of Nuclear Power Plants (NPPs) typically proceeds by propagating the uncertainty of the input variables and parameters through the codes. The input uncertainty is usually characterized by expert-based probability density functions. Inverse Uncertainty Quantification (IUQ) can be used to inform the characterization of such probability density functions based on experimental data. In this paper, we propose an empirical IUQ method that exploits data from experimental facilities and apply it to a case study of the ATRIUM project of OECD/NEA/CSNI/WGAMA. Specifically, the TRACE best-estimate thermal–hydraulic system code is used to replicate the data of the Sozzi-Sutherland, Super Moby Dick and Marviken experimental facilities with regards to an accident scenario with guillotine break of the pressurizer surge line in a generic three loop PWR-900; the input uncertainties are first characterized and then propagated to the output.