Dependency Idioms for Quantitative Human Reliability Analysis

AbstractHuman reliability analysis (HRA) is approaching nearly 60 years of reliance on key aspects of the original HRA method Technique for Human Error Rate Prediction (THERP), including its process for analyzing dependency. Despite advances in computational abilities and HRA-relevant techniques, the conceptualization, modeling, and quantification of dependency have remained largely unchanged since the introduction of THERP. As a result, current HRA methods do not consider dependency in a realistic manner, and there remain foundational gaps related to the definition, lack of causality, and quantification for HRA dependency. In this paper, we review the current conceptualization of dependency and demonstrate that current research in dependency is not addressing all of the technical gaps. To address the outstanding technical gaps in HRA dependency, we propose a set of fundamental dependency structures (HRA dependency idioms) that capture the spectrum of relationships possible between HRA variables. The idioms provide a robust logical structure for HRA dependency that emphasizes causality and is based on a causal Bayesian network modeling architecture. The idioms conceptualize and model HRA dependency in an objective, traceable, and causally informed manner that facilitates data-based quantification of HRA dependency.Keywords: Human reliability analysisdependencyhuman error probabilityhuman failure eventperformance influencing factor AcknowledgmentsThis research is funded by the U.S. Nuclear Regulatory Commission under grant number 31310020M0002. Any opinions, conclusions, or recommendations expressed in this paper belong to the authors alone and do not necessarily represent the views or opinions of the sponsoring agency. The authors would like to thank the A. James and Alice B. Clark Foundation for their support through the Clark Doctoral Fellows program at the University of Maryland.Disclosure StatementNo potential conflict of interest was reported by the author(s).Additional informationFundingThis research is funded by the NRC under grant number 31310020M0002. Any opinions, conclusions, or recommendations expressed in this paper belong to the authors alone and do not necessarily represent the views or opinions of the sponsoring agency.