The long-distance pipeline is the crucial connection for gas producing area and demand location. It may be damaged due to the impact of various factors, resulting in the leakage of gas. Reliable failure assessment can avoid the occurrence of pipeline leakage and reduce losses effectively. Because of the lack of pipeline historical data and the impact of various uncertainties, this paper adopts the fuzzy Bayesian network (BN) and analytical hierarchy process (AHP) method with expert heuristic to analyze the probability of gas pipeline failure. In the proposed model, an AHP is developed to identify the factors affecting pipeline failure by literature, database and expert questionnaire. Their prior probabilities and conditional probabilities are analyzed with expert opinion and fuzzy theory in the absence of historical data. By using the prediction and diagnosis techniques of BN, the failure probability of gas pipeline is achieved and the critical events are identified. The model is applied to an underwater natural gas pipeline to prove the feasibility and rationality of the model. The results indicate that the proposed model can provide guidance to pipeline managers for the safety of gas pipelines and a technical guide for decision makers with pipeline failure.

This content is only available via PDF.
You do not currently have access to this content.