Risk assessment of domino effects under fire and explosion accidents in LNG storage tank farms based on Bayesian network

Abstract

With the increase of liquefied natural gas (LNG) storage capacity, the number of storage tanks in LNG tank farms increases, which makes the possibility of domino effect higher, so it is important to analyze the risk of domino effect in LNG tank farms. However, the traditional domino effect risk assessment only considers a single accident type, which is not consistent with reality. Therefore, a Bayesian network-based risk assessment method for the domino effect of fire and explosion accidents in LNG storage tank farm is proposed in this study. The method utilizes the Bayesian network approach to model the propagation of domino accidents and calculate the failure probability of domino effects under fire and explosion accidents. The most likely sequence of accidents and the critical tank in the LNG storage tank farm can also be defined through the Bayesian network inference. Finally, the reliability of the proposed method is proved by a real case. Compared with only considering a single accident type, mixed accident types lead to a more reasonable result in the domino effect probabilistic analysis. It is believed that the developed approach has great theoretical significance for domino accident prevention and control in LNG storage tank farms.