Insights from the Estimation of RPS/ESFAS Component Demand Failure Probabilities based on Performance Monitoring Data after a Risk Informed Surveillance Test Interval Extension

摘要

In the risk evaluations to support a RI STI extension at a reference plant, it was assumed that the demand failure probabilities of Reactor Protection System/Engineered Safety Feature Actuation System (RPS/ESFAS) components would increase by a factor of N as their STIs are increased by a factor of N. Such an assumption is believed to produce an upper bound failure probability because not all causes of a components failure are time-dependent. The purpose of this study was to see what actually happened to the demand failure probabilities of the affected RPS/ESFAS components after the RI STI extensions. The reference plant specific experience data collected from a four-year-long performance monitoring for the affected components after the RI STI extension and demand failure probabilities before the STI extensions, which were based on the experience data from the reference plant and other similar plants, were obtained for this study. A null Hypothesis, that the demand failure probability of a selected RPS/ESFAS component after the STI extension remains the same as the value before the STI extension, was tested. The null hypothesis was not rejected for all of the selected RPS/ESFAS components. Based on the insight from the hypothesis tests, the demand failure probabilities before the STI extension were chosen as the prior distributions for the Bayesian data updates for the selected RPS/ESFAS components. The Bayesian updates showed that the demand failure probabilities of the affected RPS/ESFAS components remained almost the same for all components but one whose failure probability increased by 7%. In conclusion, for the selected RPS/ESFAS components, the demand failure probabilities did not seem to increase by a factor of N when their STIs had been increased by a factor N. Rather, the probabilities after the STI extensions seem to remain close to the probabilities before the STI extension. A continuous monitoring and Bayesian update would provide better insights for the demand failure probabilities after the RI STI extensions. It should be noted that the insights and conclusions from this study may not be directly applicable to other type of components and also they are not applicable to the case where an STI is extended to such an extent that aging effects take effect during the extended STI.