Digesting human-related incidents in nuclear power plant commissioning – Part I: An integrated methodology
本研究提出了一套综合方法论，通过多学科专家团队对核电站调试中的133起人因事件进行编码和分析，探讨了绩效塑造因子与人为错误之间的关联，旨在提升人因可靠性分析和安全管理水平。
• A multidisciplinary team of experts was assembled to code human-related incidents in nuclear power plant commissioning activities, ensuring the reliability and validity of the coding results through integrative expertise. • The expert team coded the incidents through a structured three-stage process, allowing interactive discussions among experts. • For the correlated Performance Shaping Factor (PSF) pairs identified in incident data, a questionnaire survey on normal task executions was conducted to further investigate these associations. • The associations between PSFs and human errors are interpreted as the nonhomogeneous impacts of individual PSF across different cognitive functions. This study aims to take advantage of incident data to enhance human reliability analysis (HRA) and safety management, with the research presented in two parts. Part I of this article reports the methodology for incident coding and analysis. Generally, human-related incident analysis aims to answer WHAT can go wrong (the failed task), HOW it can go wrong (the human error mode), and WHY it went wrong (the contextual factors). To address these questions, an HRA framework focusing on generic task types (GTTs), cognitive failure modes (CFMs) and performance shaping factors (PSFs), was applied to code 133 incidents in nuclear power plant commissioning. A multidisciplinary expert team comprising HRA, domain, and integrative experts conducted the coding through a structured three-stage process, ensuring reliability and validity. Statistical association tests were conducted to explore the relationships between different PSFs and between PSFs and CFMs along with our interpretations. Regarding PSF inter-associations, a supplementary questionnaire survey was administered to capture the strengths of these associations in general tasks. The combined incident and survey data helped to clarify the origins of the association, including direct causality, mediating effects, confounding effects, and interaction-induced collider effects. Additionally, PSF–CFM associations were used to infer the relative impact of PSFs across different cognitive functions, offering a new approach for validating PSF multipliers in HRA.