An event classification schema for evaluating site risk in a multi-unit nuclear power plant probabilistic risk assessment

Today, Probabilistic Risk Assessments (PRAs) at multi-unit nuclear power plants consider risk from each unit separately and consider dependencies and interactions between the units informally and on an ad hoc basis. The accident at the Fukushima nuclear power station underlined the importance and possibility of multi-unit accidents. These interactions make the operation of multiple units dependent on each other and should be formally accounted for in PRAs. In order to effectively account for these risks in a multi-unit PRA, six main dependence classifications have been identified: initiating events, shared connections, identical components, proximity dependencies, human dependencies, and organizational dependencies. This paper discusses these six classifications and the nature of their resulting dependence between multiple units. As a validation of the classification, this paper will also discuss multi-unit events that have occurred in operating plants. Finally, the paper will present existing methodologies that could be used to more formally quantify unit-to-unit dependencies in the PRAs for each classification.