Study On Risk Evaluation Method Of Injection Point In Salt Cave Gas Storage Based On Bayesian Network

Gas storage is used as an important means of peaking, the security situation of which can not be ignored. Risk assessment on gas storage has been still in infancy research, and effective method has not yet formed. So, it’s of great significance to carry out the research on risk assessment method for gas storage injection and production station.Due to many kinds of facilities which affected each other in the station,and it’s tough to assess the station with traditional method,therefor risk assessment method was proposed to evaluate the operation situation of station based on Bayesian networks, so as to guarantee the station operation safely. According to the characters of station, five subsystems were divided. There were in-and-out valves subsystem, gas adjustment and production subsystem, gas injection subsystem, venting subsystem, and draining subsystem. Fault tree and index system,which had been traversed into fault Bayesian networks, were built up according to failure characters of each subsystem. Failure possibility of each subsystem was calculated within probability reasoning. With influence of human factor, environment factor and management factor expressed by function failure, the method of calculating function failure coefficient was proposed. Then, consequence of station was analyzed with event tree, and consequence comprehensive severity, which was made up of casualties radius, direct economic losses and intangible losses, showed the consequences of accident. At last, risk matrix was proposed to evaluate the risk level of station. By assessing the salt cave gas storage of west-to-east pipeline company, the practicality of risk assessment method was affirmed.Assessment indexes were separated into shared indexes and individual indexes,which not also embodied the common characteristic of the subsystem, reducing redundancy evaluation,but still reflected the personality difference between devices. Combining the easy maneuverability of index system and logic stringency of Bayesian networks, the process of evaluation is relatively easily understood by technical stuff, and the result of evaluation is closer to the actual situation.