Plant-specific human reliability analysis (HRA) tailors human error probabilities (HEPs) to a facility's unique contexts. The diversity of digital human-machine interfaces and operational cultures necessitates methods grounded in empirical data rather than generic databases to ensure true plant-specificity. Advancing beyond prior fragmented data applications, this study establishes a systematic pipeline integrating the HuREX (Human Reliability data Extraction) database with the EMBRACE (EMpirical data-Based crew Reliability Assessment and Cognitive Error analysis) method. By sharing a unified task taxonomy, this framework directly translates plant-specific simulator data into coherent HEP estimates. In an application of this framework to the APR1400 plant, time-dependent failure probabilities were derived using Bayesian inference on 30 simulator performance records. Concurrently, nominal primitive error probabilities (NPEPs) were estimated via logistic regression from 44,585 task records. Due to current data scarcity, performance shaping factor (PSF) multipliers were determined through structured expert elicitation. This data-method integration provides a replicable foundation for generating context-sensitive HEPs in digitalized control rooms, highlighting both empirical strengths and current data limitations. • A plant-specific HRA approach is developed by integrating HuREX with EMBRACE. • HuREX data support the estimation of the time and cognitive failure probabilities. • The proposed approach is applied to estimate HEPs for the APR1400 plant. • Limitations such as reliance on expert judgment are discussed.
Kim et al. (Sun,) studied this question.