Study On Accident Analysis And Risk Assessment For Oil Depot Based On Safety Barrier

A large amount of flammable and explosive substances are stored in the oil depot.When a material or energy release occurs in the tank,the fire or explosion accident may occur.Furthermore,the domino effect may be triggered due to the propagation for physical effects of fire or explosion accidents.These accidents can lead to irreparable loss of corporate reputation,assets and human life.At present,there is no widely accepted accident model for oil depot,and the systematic method for oil depot accident analysis and risk assessment is still needed.Through in-depth analysis of a large number of historical accidents in oil depots,the various risk factors that affect the safety of oil depots and the causality of accidents are clearly identified.The model and method for accident analysis and risk assessment are developed,which can provide scientific basis and technical support for safety management.A total of 1144 historical accidents in domestic oil depots are collected.Through single-factor analysis of the year of occurrence,type of accident,accident unit,operation status,ignition source,causes and consequences of the accident,the frequency distribution of the characteristic factors of each accident is clarified.Through interactive analysis of accident causes,types of accidents,and accident consequences,the cause-effect interaction of accidents and the relative risks of different contributory factors to different accident types or consequences are obtained.Data support and theoretical guidance for accident modeling and analysis and risk assessment are provided by accident case analysis.The safety barrier-based accident model for oil depot is established based on improved SHIPP model and the process characteristics and accident feature of oil depot.According to the causality contained in the collected accident cases,a comprehensive Bayesian network reflecting the dependency of the safety barriers is constructed.Parameter learning is performed through the EM algorithm to realize the quantitative analysis of the degree of interaction between safety barriers and the dynamic prediction of accident types and consequences in different scenarios.The generic fault tree models of the accident process and each safety barrier are established through the accident initiation and development process characterized by the proposed accident model for oil depot and in-depth analysis of accident cases and operation and management requirements in oil depot.The generic fault tree-based accident analysis method is proposed by applying these fault tree models to a specific accident case.A total of 336 abnormal events occurred in a crude oil depot are investigated.The scenario and sequence of each abnormal event are comprehensively analyzed based on the proposed accident model for oil depot.The cause-effect relationship model is constructed by fault tree and event tree analysis,and the fault tree and event tree are further mapped into the Bayesian network to analyze uncertainty and conditional dependencies.For new evidence information,the probability updating was implemented through the Bayesian network update mechanism.Furthermore,the experience learning was performed by learning from field abnormal event data based on Bayesian theory to minimize the uncertainty of the prior probability.The dynamic risk prediction was performed by the update mechanism and experience learning of Bayesian network.The bow-tie model of the major leak of external floating roof tank was established by comprehensive consideration of the barrier functions to avoid,prevent,control and limit the consequences of the critical event and the logical relationship between causes and events based on the proposed accident model.The dynamic risk assessment method based on safety barriers is proposed by mapping the bow-tie model into the Bayesian network model.