A model-based approach to operational event groups ranking

Operational experience (OE) provides benefits for all major industrial activities. The problem is the limited access to industry-wide OE. A proper understanding is still developing. The major issue is to identify the most important and interesting groups of events from the accumulated experience in order to understand the root causes, to orient industry response to the collected events over the long term and to prioritise further investigations. Usually, collected data are grouped by defined criteria and described with known parameters (i.e. frequencies, safety importance and trends). This paper discusses several alternative methods of ranking event groups. It evaluates twenty years of experience with nuclear power plants in France and Germany. It is found that alternative methods lead to different rankings. Accordingly, the best approach is to use a reduced set of continuous indexes of base parameters with different weighting. The most consistent and comprehensive weighting can be determined with supporting assessment using the Analytical Hierarchy Process (AHP). Different ranking methods have sufficient agreement on the main groups of events, but their order differs. The benefit of AHP is important because it allows step-by-step pairwise index comparison and, if needed, the combination of more than one expert opinion. Further analysis of the ranking methods could examine the influence of uncertainty and the robustness of parameter weights. This could further increase the credibility of the ranking results.