The risk assessment of a crude oil pipeline using fuzzy and bayesian based bow-tie analysis

Abstract

Cross country pipelines are one of the major modes of transportation of petroleum products. Due to external and internal factors, the failure probability of pipeline is increasing over the past decades. The failure of pipelines can cause fires, which can result in environmental damage, loss of materials and further resulting into casualties. Effective risk assessment is vital to prevent and mitigate such potential incidents. In this work, an efficient tool that makes use of Bayesian approach along with Bow-tie analysis is used to obtain the failure frequencies of all the identified causes that may lead to failure of the selected pipeline located in Kerala, India. Bayesian networks can perform effective dynamic risk analysis by considering the conditional dependencies between various basic events leading to pipeline failure. The fuzzy logic and expert elicitation method are incorporated to determine the prior failure frequencies of all the identified causes of pipeline failure. The effectiveness of Bayesian network in performing forward analysis to determine the probabilities of pipeline failure consequences is demonstrated. This study also identifies the weak links associated with the occurrence of particular consequences, so that adequate measures can be taken to rectify them.