Gas Pipeline Leakage Risk Assessment Based on Dynamic Bayesian Network

Abstract

To solve the problems of numerous influencing factors, high uncertainty, and leakage risk of gas production pipelines in high sulfur gas fields, a dynamic analysis of the gas production pipeline’s leakage risk using a dynamic Bayesian network is proposed. By means of the Bow-tie model analysis, the primary risk sources of gas pipeline leakage and different accidents are summarized. The temporal dimension is introduced to construct a dynamic Bayesian network model, utilizing the Leaky noisy-or-gate model to rectify and compute the conditional probability, thereby facilitating dynamic risk assessments of gas pipeline leakage. Taking the first section of the pipeline of a municipal gas collection station as an example, with the help of Genie software, the influence degree of each basic event on the pipeline gas leakage is revealed. The change curve of the gas leakage probability over time is drawn, and the occurrence probability of potential accident consequences is computed. The results indicate that the status of flanges, valves, and pipelines are the key factors in determining the occurrence of gas leakage accidents, and six risk sources, including medium corrosion in gas leakage accidents, are determined, which have practical conspicuousness for strengthening the leakage protection of gas pipeline and providing proper support for the formulation of relevant safety measures.

Keywords: Dynamic analysis, Bow-tie model, Primary risk source, Leaky-noisy-or-gate model, Dynamic risk assessment