Conclusions on human reliability analysis (HRA) methods from the international HRA empirical study

The International Human Reliability Analysis (HRA) Empirical Study collected data in the Halden Reactor Project's HAMMLAB (HAlden huMan-Machine LABoratory) nuclear power plant simulator facility. HRA analysis teams performed predictive analyses of operating crew performance in several accident scenarios and the results of these analyses were compared to reference data derived from the actual performance of real crews in the scenarios. The comparisons examined both the qualitative and quantitative method predictions. The results provided both general and specific insights about HRA methods and their application. Most of the insights were derived from assessing: 1) whether the methods have the capacity to identify performance drivers and operational details of the performance of the required actions, and 2) whether they have the ability, if applied correctly, to use this information in the evaluation of the human error probabilities (HEPs) in such a way that they reflect the difficulty associated with the performance of the associated actions. This paper discusses some of the final lessons learned from the study, addressing both specific aspects of the methods involved and more general conclusions about HRA methodology. It addresses aspects of HRA methods identified as needing improvement.