AN APPROACH TO THE ANALYSIS OF COMMON-CAUSE FAILURE DATA FOR PLANT-SPECIFIC APPLICATION

The analysis of common cause failures of redundant systems, i.e. failu re of multiple components due to a shared cause within their mission t ime, involves the estimation of probabilities of events for which data based on operating experience are hard to obtain and, when available, do not often provide enough information for unambiguous interpretatio n and use in the analysis. Because of rarity of common cause failure e vents on a plant- and system-specific basis, plant-specific probabilis tic risk assessments have to rely heavily on the industry (generic) ex perience to develop a statistically significant data base for estimati on of common cause failure probabilities. The overall approach is desc ribed in detail in NUREG/CR 4780 (1989). The process of interpreting g eneric experience event data and translating them for plant-specific a pplications involves considerable judgment for which only limited expl icit guidance was provided in the aforementioned document. The methods and suggestions given in this paper are intended to fill the gap and to provide the analyst with tools that would enable him or her to adop t a self-consistent, systematic and well-documented approach to failur e event interpretation and to provide a quantitative assessment for pl ant-specific studies. Formulae and parameter estimates are developed t o address the most commonly encountered cases in common cause failure data analysis, i.e. events involving degraded states of components, an d failures separated in time. Several examples are provided to illustr ate the application.