Reliability evaluation for ventilation system of gas tunnel based on Bayesian network

Abstract The ventilation system is the only channel for gas exchange inside and outside the gas tunnel. Therefore, it is very important to evaluate its reliability for the safe operation of the gas tunnel. In this paper, the ventilation system of gas tunnel in Songjiang District of Shanghai City is taken as the study object, a fault tree model of the ventilation system is established, and a corresponding evaluation model of Bayesian network is formed according to the mapping rule. The fuzzy probability obtained from the fuzzy set theory and the multi-layer expert judgment is taken as the prior probability, the reliability evaluation of the ventilation system was carried out. Finally, the hierarchical management measures of the ventilation system are formulated. The results are as follows: (1) The main risk sources of the ventilation system include the third party damage, system aging, design defect and misoperation. (2) The aging of exhaust fan and no emergency and alarm function seriously reduce the reliability of the ventilation system. (3) The events with high sensitivity and promoted index to the failure of the ventilation system are completely consistent and concentrated on the system aging and design defect. (4) The aging of lines, exhaust fan and alarm instruments are the most likely evolution paths of the failure of the ventilation system. (5) At present, the reliability level for the ventilation system of gas tunnel in Songjiang District of Shanghai City is level IV, and it can run safely.