Reliability Quantification of Passive Containment Cooling System Through Response Surface Methodology

Abstract For the new generation of nuclear power plants that widely adopt passive system design, the reliability analysis of passive safety systems shows their important impact increasingly. In addition to hardware failure, functional failure of passive systems is a special and important type of failure modes, which is physical process failure caused by the uncertainty of design functions. This study took the probabilistic safety analysis study of the Passive Containment Cooling System as an example, taking loss of coolant accident as passive system reliability analysis scenario. The functional criteria was based on the pressure within the containment during the long-term cooling period after the PCCS was put into use. The study determined the key operating parameters and design parameters that affect PCCS based on the above scenario and criteria. The study used rank correlation coefficient to accomplish the sensitivity analysis of key parameters on system functions, found out the key influencing factors, and simplified the model. After reducing the number of parameters involved in the study and focusing on key parameters, the artificial neural network (ANN) method was adopted. Based on the best estimation procedure calculation results of design points, the ANN obtained an accurate limit surface with respect to the functional criterion. By quantifying the uncertainty of key parameters, a large number of samples were obtained to complete the response surface calculation, and the failure probability of the system under corresponding accident conditions was obtained. Compared with traditional probability evaluation methods, this research provided a new method to quantify the safety of passive systems, which has higher calculation efficiency and ensured calculation accuracy.