Statistical Analysis of Common Cause Failure Data to Support Safety and Reliability Analysis of Nuclear Plant Systems

This report describes the findings of a project “Statistical Analysis of Common Cause Failure Data to Support Safety and Reliability Analysis of Nuclear Plant Systems for the CNSC” under the contract No. 87055-12-0221. Analysis of Common Cause Failures (CCF) is an important element of the Probabilistic Safety Assessment (PSA) of systems important to safety in a nuclear power plant. Based on the conceptualization of the CCF event, many probabilistic models have been developed in the literature. This Report provides a comprehensive review of CCF modeling techniques, which shows that the a modern method, called “General Multiple Failure Rate Model”, is the most suitable method for probabilistic modeling of CCF events. Therefore, the GFMR is described in detail in the report and adopted for the case studies. To estimate the parameters of the GMFR model, the Empirical Bayes (EB) method is adopted. The report describes the data mapping methods and the EB method for combining data from different component groups and plants in the statistical estimation. This project presents detailed case studies to illustrate the data mapping and EB method. The case studies are based on CCF data for motor operated valves (MOVs). These case studies serve as templates to analyze CCF data from other safety systems. The report provides analysis methods to the CNSC staff to analyze CCF rates and evaluate the adequacy of input data used in the PSA of Canadian plants. This project demonstrates the development of the capacity to analyze CCF data in line with best international practices.