Probabilistic modeling of electric vehicle fires under conditions of initial data uncertainty

The rapid global electrification of motor vehicles is raising a host of fire safety issues. Primarily, concerns arise regarding the behavior of lithium-ion batteries, which are subject to thermal runaway, a self-sustaining, uncontrolled exothermic process. Existing information on the frequency and causes of electric vehicle fires is fragmented due to the lack of standardized methods for investigating, classifying, and recording these incidents internationally. This article examines the feasibility of using a Monte Carlo simulation to estimate the likelihood of electric vehicle fires. Given the scarcity of statistical information on electric vehicle fires, standard models are insufficient to accurately assess the likelihood of fire hazards in the presence of uncertainty in the initial data. A stochastic model is proposed that accounts for two causes of thermal runaway in lithium-ion batteries: mechanical deformation and manufacturing defects (taking into account the battery's charge level and temperature). Probability distributions of the parameters are specified for each cause. The computational component includes 10000 independent trials, guaranteeing the stability of the resulting probability characteristics. It is confirmed that the Monte Carlo method enables quantitative assessment of the fire probability even with a limited empirical base.