A new research paradigm for lithium-ion battery thermal runaway propagation based on characteristic temperature probabilistic modeling

This study proposes a novel research paradigm of “Statistical data analysis–Probabilistic model construction–Dynamic law extraction” to address the strong randomness and poor repeatability of thermal runaway (TR) propagation in lithium-ion battery (LIB) modules. Within this framework, a probabilistic TR propagation model for 18,650-type LIB modules is developed by integrating the statistical characteristics of key temperatures and energy release with a lumped heat-generation model and a thermal-resistance network. The results reveal that the introduction of probabilistic triggering transforms the TR process from a deterministic mode into diversified propagation patterns, characterized by irregular square-wave expansion along the diagonal. Sequential propagation remains dominant, while jumping propagation emerges under the combined effect of lower triggering temperatures in downstream cells and weaker heat release intensities in upstream cells. Monte Carlo simulations indicate that, for a 10 × 10 cell module, the main distribution ranges of total propagation duration and peak temperature are 842–887 s and 902.8–912.5 K, respectively, with the corresponding probability density functions quantified. The overall TR process exhibits a dynamic evolution pattern of “slow initiation–rapid acceleration–final deceleration”, with each stage displaying distinct statistical characteristics and representative intervals. This paradigm shifts TR research toward probabilistic–statistical thinking and provides a systematic basis for probability-oriented thermal safety design.