A probabilistic analysis model of oil pipeline accidents based on an integrated Event-Evolution-Bayesian (EEB) model

Abstract

Abstract Over the past several decades, the high frequency of oil pipeline accidents has drawn substantial attention around the world. Many oil pipeline accident analysis models have been established based on the event tree method, the Bayesian network method and Computational Fluid Dynamics (CFD) simulation models. Considering the disadvantages of current models for comprehensively representing the incident evolution process and quantitative analysis for consequences, this paper proposes a probabilistic analysis model for oil pipeline accidents that integrates three methods the event tree (E), the incident evolution diagram (E) and the Bayesian network (B). Therefore, the model is called the “EEB model”. The EEB model can identify the initial event and secondary events, illustrate the accident evolution path, identify the key influencing factors, analyze their effects, and calculate the probabilities of different consequences of oil pipeline network accidents. Compared with other models, the EEB model considers more factors, such as key environmental conditions and the emergency response. Probabilistic analysis of different consequences, including casualties, economic losses, environmental pollution and the influence on social order, can be obtained. For a general scenario of an oil pipeline network accident, the probabilities for different consequences are 71.3% for “less than 5 persons affected”, 68.2% for “less than 10 million RMB lost”, 50.4% for “less than 1 km2 of water pollution” and 59.5% for “influence on social order of less than 100 persons”. The risk for the accident can be estimated by assuming the probability of the initial event as P. The model also denotes the emergency targets to be achieved and the response missions to be executed. Based on this information, a response plan can be developed for decision making. Since the incident evolution process is complex, the effects of the influencing factors should be analyzed. The EEB model highlights the significant influences of the water area (e.g., the probability of “10–50 km2 of water pollution” decreases from 38.7% for “near and large” water bodies to 17.4% for “far and small” water bodies) and the emergency response (e.g., the probability of “50–100 million RMB economic loss” increases from 11.5% for an “effective” response to 29.3% for a “poor” response). The probabilistic analysis obtained by the EEB is more comprehensive than those of other models, and the results can be used for risk analysis, decision making and effect analysis of oil pipeline networks.