Generation of common cause failure rate distributions for Monte Carlo analysis using the binomial failure rate model

Northeast Utilities has adopted modified versions of the WAMCUT and SPASM Codes which were developed under EPRI sponsorship for quantifying system unavailability and uncertainties in PRA studies. In order to include and statistically quantify the effects of potential common cause failure events in the system fault trees, a number of possible techniques were considered. The Beta factor approach however is difficult to adapt to the needs for performing an uncertainty analysis. After consideration the Binomial Common Cause Failure Rate model was chosen. The critical consideration in this choice was the fact that Atwood of EG and G has developed a substantial data base of point estimates (mean values) and upper (95%) bounds based on operating experience. In order to utilize this data base for evaluation of the impacts of common cause failure using Monte Carlo methods, such as SPASM, it is necessary to: develop an uncertainty distribution (specify distribution type) and characterize distribution in terms of mean and variance. By performing these steps, common cause failure and its associated uncertainty can be systematically treated along with random failures in the fault tree logic