Digesting human-related incidents in nuclear power plant commissioning – Part II: Results and implications

• By integrating incident data with questionnaire survey data, the sources of associations between PSFs—including direct causality, mediating effects, confounding effects, and collider effects—were explored and discussed. • Associations between PSFs and macro-cognitive function failures were used to infer the relative impacts of individual PSFs across different macro-cognitive functions. This interpretive approach and the inferred results were validated by comparing them with the PSF multiplier design used in current Human Reliability Analysis (HRA) methods. • The significance of knowledge-based errors and inter-team errors in activities outside the control room was highlighted. This study aims to analyze human-related incidents in nuclear power plant commissioning to inform human reliability analysis (HRA) and safety management following the integrated methodology reported in the preceding Part I article. This Part II article presents the key results, yielding several important insights. First, the study underscores the significance of knowledge-based and inter-team errors, both of which have been insufficiently explored in HRA research. In addition, the human error patterns observed during commissioning closely aligned with those in the test and maintenance tasks, indicating the potential for analyzing human errors in ex-control room activities under a unified framework. This study demonstrates how to clarify the associations between performance shaping factors (PSFs), including direct causality, mediating effects, confounding effects, and interaction-induced collider effects, by combining incident data with survey data, with the first three being more prevalent. These different associations have significant implications for HRA researchers and practitioners. Finally, the study showed that the associations between PSFs and human errors in incident data can be used to infer the relative impacts of individual PSFs across different cognitive function failures. A comparison of these inferences with existing HRA methods revealed considerable alignment, suggesting a promising avenue for validating PSF multipliers in current HRA frameworks. In summary, with its high ecological validity, incident data can offer valuable insights for improving HRA methodologies and safety management practices.