Risk Analysis on Out-of-control Burst of Jacketed Reactor Based on Bayesian Network

4th International Conference on Energy Equipment Science and Engineering

Hu, Qiming^1 ; Tao, Gang^1 ; Zhang, Lijing^1

College of Safety Science and Engineering, Nanjing Tech University, Nanjing, Jiangsu

211800, China^1

关键词: Bayesian reasoning;    Chemical process;    Influence degree;    Maintenance periods;    Posterior probability;    Pressure control valves;    Prior probability;    Process flow diagrams;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/242/3/032006/pdf
DOI  :  10.1088/1755-1315/242/3/032006

来源: IOP

【 摘 要 】

Reactors are widely used in chemical process, but corresponding accidents occur one after another. In this paper, the process flow diagram of a typical jacketed reactor is taken as an example to analyse the causes of jacketed reactor out-of-control burst. The basic events that cause reaction out of control in reactor are found by establishing the fault tree (FTA). On the basis of the fault tree, the risk of jacketed reactor out-of-control burst is analysed by developing Bayesian network using AgenaRisk program. When jacketed reactor runs with one year, two years and three years, the probability of jacketed reactor out-of-control burst is 0.002%, 0.035% and 0.155%, respectively. With the in-service time extending, the probability of jacketed reactor out-of-control burst increases drastically. After Bayesian reasoning, it is found that the posterior probability of pressure control valve and discharge pump is the largest among all basic events, and the posterior probability is more important than the prior probability in comparing the influence degree which causes jacketed reactor out-of-control burst. This paper assumes that pressure control valve and discharge pump are repaired every three months, six months and one year, and the probability of jacketed reactor out-of-control burst with one, two and three years in-service is calculated, respectively. The results show that compared with the risk of no maintenance, the risk after maintenance drops significantly, and the maintenance period is more shorter, the probability of jacketed reactor out-of-control burst is more lower.