Human reliability analysis in inert gas operations with fuzzy CREAM-based Bayesian networks

Human error remains a leading cause of maritime accidents, especially in safety-critical operations like inert gas (IG) handling. This study presents a structured framework for assessing human reliability in IG operations by integrating the Cognitive Reliability and Error Analysis Method (CREAM), Fuzzy Set Theory (FST), and Bayesian Networks (BNs). The proposed model overcomes limitations of conventional Human Reliability Analysis (HRA) by addressing uncertainty and interdependencies in performance-shaping factors. Expert evaluations of Common Performance Conditions (CPCs) were processed using fuzzy membership functions, and probabilistic relationships were modeled via a Bayesian framework constructed in GeNIe. Task-specific Human Error Probabilities (HEPs) were calculated using defuzzified control mode distributions. The findings revealed that tasks involving manual intervention and time-sensitive decisions, such as boiler uptake valve operation (HEP: 0.00234) and IG plant startup (HEP: 0.00153), have the highest error potential. Routine tasks, like pressure checks, exhibit low HEP values (e.g., 0.00040). The model was validated against Basic-CREAM for consistency. The novelty of this study lies in the integration of Fuzzy CREAM with Bayesian inference to create a probabilistic human reliability model, an approach not widely applied in maritime safety research. This hybrid method enhances HEP estimation and supports risk-informed decision-making in IG operations.