Insights into PRA methodologies

Probabilistic Risk Assessments (PRAs) for six nuclear power plants were examined to gain insight into how the choice of analytical methods can affect the results of PRAs. The PRA sreflectope considered was limited to internally initiated accidents sequences through core melt. For twenty methodological topic areas, a baseline or minimal methodology was specified. The choice of methods for each topic in the six PRAs was characterized in terms of the incremental level of effort above the baseline. A higher level of effort generally reflects a higher level of detail or a higher degree of sophistication in the analytical approach to a particular topic area. The impact on results was measured in terms of how additional effort beyond the baseline level changed the relative importance and ordering of dominant accident sequences compared to what would have been observed had methods corresponding to the baseline level of effort been employed. This measure of impact is a more useful indicator of how methods affect perceptions of plant vulnerabilities than changes in core melt frequency would be. However, the change in core melt frequency was used as a secondary measure of impact for nine topics where availability of information permitted. Results are presented primarily in the form of effort-impact matrices for each of the twenty topic areas. A suggested effort-impact profile for future PRAs is presented