Simulation-Informed Probabilistic Methodology for Common Cause Failure Analysis

Common Cause Failures (CCFs) are critical risk contributors in complex technological systems as they challenge multiple redundant systems simultaneously. To improve the CCF analysis in Probabilistic Risk Assessment (PRA), this research develops the Simulation-Informed Probabilistic Methodology (S-IPM) for CCFs. This new methodology utilizes simulation models of physical failure mechanisms to capture underlying causalities and to generate simulation-based data for the CCF probability estimation. To operationalize the S-IPM in PRA, a computational algorithm is developed that generates simulation-based estimates of CCF parameters and, using the Bayesian approach, integrates them with the data-driven CCF parameters (if relevant data available) from the existing PRA. This computational algorithm is equipped with the Probabilistic Validation that quantifies the degree of confidence in the simulation-based parameter estimates by characterizing and propagating epistemic uncertainty in multiple levels of analysis. The S-IPM can (i) provide more realistic CCF probability estimates by considering CCF data generated from simulations; (ii) reflect as-built, as-operated plant conditions, considering the updates in design, operational, and maintenance policies; and (iii) contribute to more effective prevention and mitigation of CCFs by providing “cause-specific” quantitative risk insights. The paper shows a case study that applies S-IPM to the CCFs of emergency service water pumps of NPPs.