基于动态贝叶斯网络的石化装置火灾事故动态风险分析

Dynamic risk analysis of petrochemical plant fire accident based on dynamic Bayesian network

为揭示石油炼化装置事故风险动态特性和事故情景演变路径,在对石化装置进行风险因素分析的基础上构建石化装置火灾事故故障树,基于贝叶斯网络非常规突发事故的演变过程,构建情景演变下的动态贝叶斯网络模型,在综合考虑应急措施的基础上,利用MATLAB软件和联合概率公式计算出各种事故场景的状态概率。以丙烯精馏装置火灾事故为例,结果表明:丙烯精馏装置故障主要影响因素为精馏塔体故障和冷凝器、再沸器故障,其中,丙烯泄漏起火被控制概率为73.7%,装置发生流淌火被扑灭概率为60.4%,冷凝器、再沸器抢修成功概率为57%。基于贝叶斯网络的动态风险分析方法能够对石化装置火灾事故的演变进行合理预测,为石化装置火灾事故应急决策提供理论依据。

In order to reveal the dynamic characteristics of accident risk and the evolution path of accident scenarios in petroleum refining equipment,a petrochemical equipment fire accident fault tree is constructed based on the analysis of risk factors of petrochemical equipment,a dynamic Bayesian network model under scenario evolution based on the evolution process of unconventional emergencies of Bayesian networks,and based on comprehensive consideration of emergency measures,MATLAB software and joint probability formulas are used to calculate the state probabilities of various accident scenarios.Taking the fire accident of a propylene rectification device as an example,the results show that the main influencing factors of the propylene rectification device failure are the failure of the rectification tower body,the condenser and the reboiler.Among them,the probability of propylene leakage and fire being controlled is 73.7%,and the device occurs.The probability of extinguishing the flowing fire is 60.4%,and the probability of successful repair of the condenser and reboiler is 57%.The dynamic risk analysis method based on Bayesian network can reasonably predict the evolution of petrochemical plant fire accidents and provide a theoretical basis for the emergency decision-making of petrochemical plant fire accidents.