Human reliability plays a pivotal role in ensuring safety and operational resilience across high-risk systems such as nuclear power, petrochemical plants, transportation networks, and industrial maintenance. Conventional Human Reliability Analysis (HRA) approaches are largely static and assume static Performance Shaping Factors (PSFs), limiting their ability to reflect temporal variability and dynamic human behavior. This study proposes a Hybrid Static–Dynamic Human Reliability Analysis (HSD-HRA) framework that integrates causal weighting of sub-PSFs using the DEMATEL method with a Proportional Hazard Model (PHM) featuring both fixed and random-effect covariates. The framework captures complex interdependencies among PSFs while modeling operator-level behavioral variability, providing a more realistic representation of human error dynamics. A case study in a cement manufacturing maintenance system demonstrates the framework’s applicability: the static PHM estimated a Human Error Probability (HEP) of 0.0856, whereas the mixed-effects PHM produced an HEP of 0.1112, indicating a 2.6% degradation in reliability attributable to behavioral uncertainty. The HSD-HRA framework offers a rigorous, data-driven basis for predictive human reliability assessment, supporting risk-informed maintenance planning, improved human–machine interface design, and enhanced decision-making in safety-critical environments.

Keywords