Development of Adaptive Sampling Software Based on Relap5

Abstract

The RISMC analysis method is a coupled dynamic safety analysis method that combines deterministic safety analysis and probabilistic safety analysis. It comprehensively simulates the dynamic response process of nuclear power plants under accidents through Monte Carlo sampling and dynamic event tree methods. In order to fully reflect the response process of the nuclear power plant and ensure the accuracy of the calculation results, it is necessary to sample a large number of input space parameters, which will result in unbearable time costs. Adaptive sampling can predict the state of the position input space parameters through a small number of forward sampling calculation results and divide the input space by generating limit surfaces to calculate the failure probability, which can greatly save computing time. Based on researching RISMC analysis tools at home and abroad, the research and development of adaptive sampling and RELAP5 coupling calculation software is realized. The test case of the Qinshan nuclear power plant station blackout accident shows that the adaptive sampling function of calculating failure space probability through a small amount of sampling is realized, and the efficiency of RISMC analysis and calculation is improved.