Bayesian Methods in Risk Assessment

Abstract

Probabilistic risk analysis has been evolving as a scientific discipline quite rapidly over the last 15 years. The major reason for this growth has been the explicit recognition of the need for a consistent framework for the analysis of the safety of large industrial facilities, especially nuclear power plants, which other approaches such as worst-case analyses, do not provide. It has been realized, however, that the conventional methods of statistics and reliability theory cannot be applied in a straightforward manner because of the rarity of events of interest and consequent lack of meaningful statistical records. The frequent use of judgment and the initial reporting of very low probabilities raised questions concerning the credibility of the whole approach and controversy inevitably resulted, as witnessed by the release of the United States Nuclear Regulatory Commission’s Reactor Safety Study (RSS) (1) and subsequent developments (2–5).KeywordsRisk AssessmentPosterior DistributionNuclear Power PlantPrior DistributionBAYESIAN MethodThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.