Finite mixture models for sensitivity analysis of thermal hydraulic codes for passive safety systems analysis

• Uncertainties of TH codes affect the system failure probability quantification.
• We present Finite Mixture Models (FMMs) for sensitivity analysis of TH codes.
• FMMs approximate the pdf of the output of a TH code with a limited number of simulations.
• The approach is tested on a Passive Containment Cooling System of an AP1000 reactor.
• The novel approach overcomes the results of a standard variance decomposition method.

For safety analysis of Nuclear Power Plants (NPPs), Best Estimate (BE) Thermal Hydraulic (TH) codes are used to predict system response in normal and accidental conditions. The assessment of the uncertainties of TH codes is a critical issue for system failure probability quantification. In this paper, we consider passive safety systems of advanced NPPs and present a novel approach of Sensitivity Analysis (SA). The approach is based on Finite Mixture Models (FMMs) to approximate the probability density function (i.e., the uncertainty) of the output of the passive safety system TH code with a limited number of simulations. We propose a novel Sensitivity Analysis (SA) method for keeping the computational cost low: an Expectation Maximization (EM) algorithm is used to calculate the saliency of the TH code input variables for identifying those that most affect the system functional failure. The novel approach is compared with a standard variance decomposition method on a case study considering a Passive Containment Cooling System (PCCS) of an Advanced Pressurized reactor AP1000.