An Integrated Risk Evaluation Model based on Bayesian Networks for the Process Safety of Petrochemical Industry
Risk evaluation is an important part of safety management for petrochemical plants, and there still exist limitations in traditional safety assessment methods, especially for the sophisticated process.After taking into full account the characteristics of petrochemical process system, an integrated risk evaluation model based on the both-ward inference of Bayesian network (BN) was established in this work.The new evaluation model combines the advantages of three traditional safety assessment methods, namely, hazard and operability analysis (HAZOP), layer of protection analysis (LOPA), and bow-tie analysis.In this model, firstly, establish the fault tree of process system and map it into a corresponding BN, then predict the events that can easily trigger accidents on the basis of posterior probabilities of basic events in the fault tree.The posterior probability of each basic event is obtained from the inference of BN by GeNIe software,which is based on the prior probabilities of basic events.The events with high-risk are regarded as Bayesian fault nodes.Conduct an overall evaluation on these Bayesian fault nodes by HAZOP-LOPA, and determine the risk grade of the accident caused by the Bayesian fault node according to a semi-quantitative risk matrix.The risk grades of Bayesian fault nodes are relatively high, so some independent protection layer (IPL) measures should be taken.After that, compare the failure frequency of the used IPL with that of the permissible IPL.If the residual risk after adopting IPL measures is still beyond acceptable levels, in this risk evaluation model, it is suggested to make use of bow-tie analysis to identify key measures that can ensure the IPL measures effectively for a long time.Finally, take the explosion of heating furnace in a decayed coking unit as an example, the application of this integrated risk evaluation model in actual production is shown.
bayesian networks risk evaluation hazard and operability analysis layer of protection analysis bow-tie analysis
YAO Xiwen XU Kaili LIANG Yu
College of Resources and Civil Engineering, Northeastern University, Shenyang 110819, Liaoning, Chin College of Information Science and Engineering, Northeastern University, Shenyang 110819, Liaoning,
国际会议
The 2016 International Symposium on Safety Science and Technology(2016安全科学与技术国际会议)
昆明
英文
36-44
2016-10-17（万方平台首次上网日期，不代表论文的发表时间）