Fire risk assessment in lithium-ion battery warehouse based on the Bayesian network

Abstract

Because of the instability and susceptibility to thermal runaway of lithium-ion batteries (LIBs), their storage has always been at high risk of fire. Numerous studies have analyzed the risk of fire in lithium-ion battery (LIB) warehouses. Still, most of them have focused on the factors influencing the fire following the thermal runaway of a LIB occurs, and there has been a lack of research on the causes of the LIB thermal runaway being occurred. To address this shortcoming, this paper proposed a comprehensive framework considering multiple influencing factors for fire risk assessment in LIB warehouses that combines Bayesian networks (BNs) and expert evaluation. The proposed framework adopts the Bayesian network (BN) model that considers three main types of parameters: fire causes, fire spread influencing factors, and fire consequences. The main objective was to analyze the evolution and consequences of LIB warehouse fires dynamically and provide effective suggestions for fire risk reduction. The results showed that the proposed framework was a reliable avenue for fire risk assessment of LIB warehouses. In addition, it was demonstrated that human factors commonly cause the LIB warehouse open fires. Thus, to prevent open fires from continuing spreading, the LIB warehouses should avoid storing batteries with high state of charge values; furthermore, if it is necessary to store them, automatic sprinklers and other firefighting facilities should be installed. In addition, timely firefighter response and the installation of mechanical ventilation systems are essential to substantially curtail casualties.