Dynamic Human Error Assessment in Emergency Using Fuzzy Bayesian CREAM

Abstract

Background: Human error is one of the major causes of accidents in the petrochemical industry. Under critical situation, human error is affected by complex factors. Managing such a situation is important to prevent losses and injury. This study aimed to develop a dynamic model of human error assessment in emergencies in the petrochemical industry. Study design: A cross-sectional study. Methods: Fuzzy Bayesian network was used to improve the capabilities of the method for determining the control mode. Fuzzy-AHP-TOPSIS method was also used to prioritize emergency scenarios and human error assessment was applied for the most important emergency condition. Results: Fire in a chemical storage unit was recognized as the most important emergency condition. Common Performance Conditions (CPCs) were determined based on the opinions of a panel of 30 experts and specialists and 7 CPCs were selected for emergencies; then, based on the results of AHP method the relative weights were determined. Finally, membership functions, inputs, and outputs of fuzzy sets, CPC values in 8 emergency response tasks, and the probability of control modes were determined using Bayesian Cognitive Reliability and Error Analysis Method (CREAM) method. Conclusion: This method could be applied to overcome the weaknesses of traditional methods, provide a repeatable method for human error assessment, and manage human error in an emergency.