Pipeline Failure Assessment Based on Fuzzy Bayesian Network and AHP
Qiaoyan Yu, Lei Hou, Yanhao Li, Chong Chai, Kai Yang, Jiaquan Liu
Abstract
A long-distance pipeline is a crucial connection for oil and gas (O&G) producing areas and demand locations. Pipeline failure may be caused by various factors, which can escalate into a disastrous event, causing huge loss of life and property. Reliable failure assessment is urgently needed to avoid the occurrence of pipeline failure and reduce losses effectively. Owing to a lack of historical pipeline data and the impact of various uncertainties, this study presents a model that systematically integrates Bayesian network (BN), fuzzy theory, and analytic hierarchy process (AHP) to analyze the probability of pipeline failure. In the proposed model, factors contributing to O&G pipeline failure are obtained from the literature, database searches, and expert questionnaires. AHP is used to establish a hierarchical network represented by a causality diagram. The hierarchical network is mapped to a BN structure, and the conditional probability of nodes in the BN is obtained by AHP and expert judgment. The prior probabilities of basic factors are analyzed by expert opinion with fuzzy theory in the absence of historical data. The model can be used for the quantitative analysis of pipeline failure. Moreover, diagnostic analysis can also be performed by updating the probabilities in the BN model with new information. The feasibility and rationality of the model are validated in a gas pipeline, which indicates that the proposed model can provide effective decision-making for pipeline managers to prevent and manage pipeline failures.