Dynamic risk analysis of oil depot storage tank failure using a fuzzy Bayesian network model

Abstract

This paper presented a fuzzy Bayesian network (FBN) model for failure risk analysis of oil storage tank leakage. The Bow-tie model for storage tank failure in oil depots was established and used to determine key risk factors. Fuzzy set theory and expert judgments were used to calculate the prior probability of basic events, after which the model was mapped to a Bayesian network. The probability of storage tank failure was calculated as 7.80 E-02. As new evidence was obtained, probability updating was tested using forward inference. Posterior probability of a basic event, seal gasket aging, showed the greatest risk increase, from 1.05 E-02–1.10 E-01. Using these newly acquired key factors, probability adaptation was used to conduct a dynamic analysis of storage tank failures and the consequence states. The dynamic failure probability was found to increase by 28.67 %. The model underwent sensitivity analysis and literature data analysis. This fuzzy Bayesian network model is beneficial to dynamic risk analysis in oil depot management. The findings can be used to reduce the likelihood of oil tank failure and decrease the severity of accidents. Suggestions were made, for site engineers and management, to reduce the risk of hazardous events.