A Methodology for Reviewing Probabilistic Risk Assessments

Abstract

The starting point for peer review of a Probabilistic Risk Assessment (PRA) is a clear understanding of how the risk estimate was prepared and of what contributions dominate the calculation. The problem facing the reviewers is how to cut through the complex details of a PRA to gain this understanding. This paper presents a structured, analytical procedure that solves this problem.The effectiveness of this solution is demonstrated by an application on the Zion Probabilistic Safety Study. The procedure found the three dominant initiating events and provided a simplified reconstruction of the calculation of the risk estimate. Significant assessments of uncertainty were also identified. If peer review disputes the accuracy of these judgments, then the revised risk estimate could significantly increase.The value of this procedure comes from having a systematic framework for the PRA review. Practical constraints limit the time and qualified people needed for an adequate review. Having the established framework from this procedure as a starting point, reviewers can focus most of their attention on the accuracy and the completeness of the calculation. Time wasted at the start of the review is reduced by first using this procedure to sort through the technical details of the PRA and to reconstruct the risk estimate from dominant contributions.