Thermal runaway propagation path and fire risk assessment in electric vehicles based on full-vehicle experiments

Thermal runaway (TR) in electric vehicle (EV) battery systems poses a critical safety hazard, often escalating rapidly into full-vehicle fires under real-world conditions. To evaluate and manage this risk, this study conducts a full-vehicle TR experiment. By analyzing the evolution characteristics of temperature, voltage and pressure of multiple components, combined with TR phenomena, the TR propagation path throughout the full-vehicle is identified. This propagation pattern is further validated through post-fire analysis of thermal residues. A data-driven framework combining Pearson correlation and Granger causality analysis is developed, and the results reveal that temperature rise precedes voltage collapse, while pressure buildup lags behind, forming a sequential failure chain. Guided by these insights, a quantitative risk matrix is constructed, integrating both the values and rates of change of key parameters to classify fire risk into three levels. A stage-based safety management strategy that includes early-stage thermal anomaly detection, mid-stage containment and late-stage emergency response is proposed to dynamically adapt to evolving risk conditions. This work bridges the gap between cell-level fire risk evaluation and vehicle-level fire risk assessment, providing experimental evidence and data-driven analytical methods for the development of intelligent battery management systems capable of real-time risk detection and mitigation under realistic TR conditions.