Dynamic risk assessment method for urban hydrogen refueling stations: A novel dynamic Bayesian network incorporating multiple equipment states and accident cascade effects

Hydrogen refueling stations are increasingly being built in densely populated urban areas and operating under high temperature and pressure conditions. Therefore, the safety of hydrogen refueling stations has received great attention. In present work, a new dynamic quantitative risk assessment method is established for hydrogen leakage at refueling stations based on Bayesian networks (BN). Firstly, risk factors are identified using a fault tree, considering multiple equipment operating states. A Dynamic Bayesian Network (DBN) is established, and event tree analysis is combined to determine accident consequence probabilities. Furthermore, a Bayesian Domino Model is constructed to assess the impact of accident consequence propagation on the probability and risk of hydrogen leakage accidents. Through a typical case study of an off-site hydrogen refueling station, the practicality of the proposed method is validated, the risk factors and safety barriers, the dynamic evolution of hydrogen leakage probabilities within a three-year period are obtained. The probability of hydrogen leakage in hydrogen station is 2.69 × 10−2 in the first month and 2.45 × 10−2 in the 36th month. Simultaneously, the maximum personal risk value for the first month is 5.07 × 10−4. When the Domino effect is considered, the maximum accident probability is 1.96 × 10−3 is obtained in the first month, and the equipment unit with the greatest impact of the accident is A2. Measures were also proposed to reduce hydrogen leakage risks during the operation of off-site hydrogen refueling stations, with an emphasis on prioritizing safety measures for hydrogen storage bundles and risk mitigation strategies.