Common cause failure PSA methodology update and its application to Darlington PSA models

Canadian CANDU Owners Group (COG) members previously used a simplified approach to quantify residual common cause failure (CCF) events through a parametric model. Generic CCF data was used to derive CCF multipliers for quantification of the residual CCF events, and the residual CCFs were manually introduced directly in the system fault trees. To bring Canadian practice in line with best international practice, the CCF methodology has been updated under a COG project. The paper describes all aspects of the CCF methodology update: the selected Alpha Factor parametric model, collection of CCF data for CANDU-specific components, the ability to interface with international CCF data sources including the International Common Cause Data Exchange (ICDE) data, and development of a software tool for calculation of CCF alpha factors. The effectiveness of the new CCF methodology is demonstrated by its application to the Darlington at-power and outage internal events Level 1 and 2 PSA models. The CCF events importance to Level 1 and Level 2 risk metrics is summarized to illustrate their impact on the risk results. (author)