Higher SOC Increases Risk of Battery Thermal Runaway, Study Finds
As the global electric vehicle (EV) market continues its rapid expansion, concerns over battery safety are intensifying. With over 14.6 million new energy vehicles sold worldwide in 2023—64.8% of them in China—the demand for safer, more reliable lithium-ion batteries has never been greater. Among the most critical safety issues facing the industry is thermal runaway, a self-sustaining chain reaction that can lead to fires and explosions within battery packs. A new study conducted by researchers from Wuhan Technical College of Communications and Dongfeng Motor Corporation Technical Center sheds light on how the state of charge (SOC) of lithium-ion batteries significantly influences the likelihood and severity of thermal runaway propagation in battery modules.
Published in the Journal of Wuhan Engineering Polytechnic, the research led by Cheng Lu, Liu Liang, Ye Guojun, and Tang Qiong presents a comprehensive simulation-based investigation into the thermal behavior of lithium battery packs under various SOC conditions. Using COMSOL Multiphysics software, the team developed a lumped-parameter model to simulate the electrochemical and thermal dynamics of a commercial prismatic lithium-ion battery module. Their findings reveal a clear correlation between higher SOC levels and increased risk of thermal runaway spreading across cells within a pack—a discovery that could have far-reaching implications for EV design, battery management systems (BMS), and emergency response protocols.
The study comes at a time when public confidence in EV safety is being tested by high-profile incidents involving battery fires. While manufacturers have made significant strides in improving crashworthiness and fire containment, the underlying mechanisms of thermal runaway remain complex and not fully understood. Most existing research relies heavily on physical testing, which, while valuable, is costly, time-consuming, and often limited in scope. The approach taken by Cheng and her colleagues offers a more scalable and flexible alternative: a computational model that accurately replicates real-world thermal behaviors using experimentally derived input parameters.
At the heart of their methodology is the use of a lumped model, a simplified representation of the battery system that aggregates complex internal processes into manageable variables. Unlike detailed electrochemical models that require precise data on ion diffusion, electrode kinetics, and temperature gradients—information typically unavailable to automakers purchasing cells from third-party suppliers—the lumped model uses measurable external characteristics such as open-circuit voltage, internal resistance, and heat generation rates. This makes it particularly suitable for large-scale simulations of full battery packs, where computational efficiency is essential.
The simulated battery pack consists of six modules connected in series, each containing five individual cells arranged in a 10-row by 3-column configuration. Aluminum plates separate the modules, serving both structural and thermal functions. The cells themselves are based on a nickel-cobalt-manganese oxide (NCM111) cathode and graphite anode chemistry, with a total capacity of approximately 675 Ah. By assigning realistic material properties—including thermal conductivity, specific heat capacity, and electrical resistivity—the researchers were able to create a physically accurate representation of the system’s thermal response under stress.
To initiate the simulation, the team assumed that Cell 1 had already entered thermal runaway due to an external trigger such as mechanical impact or localized overheating. This cell was treated as a heat source, releasing energy rapidly and raising the temperature of adjacent components. The simulation then tracked how this initial failure propagated—or failed to propagate—through the rest of the pack over a 20-minute period, under six different SOC conditions: 20%, 30%, 50%, 70%, 90%, and 100%.
The results were striking. At lower SOC levels—specifically 20% and 30%—the thermal event remained localized. Although Cell 1 experienced extreme heating, the surrounding cells did not reach the critical threshold required to trigger their own thermal runaway. Temperatures in neighboring cells peaked at 81°C and 93°C respectively, well below the typical ignition point of over 200°C. This suggests that batteries operating at partial charge states may be inherently more stable and less prone to catastrophic failure cascades.
However, once the SOC reached 50%, the dynamics changed dramatically. In these scenarios, the excess chemical energy stored in the electrodes acted as a kind of “fuel” for the spreading reaction. Heat from the failing cell was sufficient to push adjacent cells past their thermal tipping points, initiating secondary thermal runaway events. At 50% SOC, the failure spread to three or four neighboring cells; at 70% SOC, the propagation was faster and slightly more extensive. But it was at 90% and 100% SOC that the most alarming results emerged. In these fully charged conditions, the thermal wave moved rapidly through the pack, with Cell 2 reaching peak temperatures of 680°C and 694°C respectively within just 200 seconds. The simulation even showed evidence of lateral spread to side cells, indicating a potential for full module or pack-level combustion.
These findings align with prior experimental work showing that higher SOC correlates with greater heat release during thermal runaway. However, what sets this study apart is its ability to model the propagation process in a spatially resolved manner across a multi-cell system. The time-to-peak-temperature metric provides crucial insight: at 100% SOC, Cell 2 reached dangerous temperatures in under three and a half minutes, leaving little time for intervention. In contrast, at 50% SOC, the same cell took about five minutes to reach peak temperature—still rapid, but potentially allowing for mitigation strategies such as active cooling or electrical isolation.
Equally important was the observation of voltage behavior during the thermal event. While all test cases showed a gradual decline in Cell 2’s voltage over time, the rate of decline was most pronounced at lower SOC levels. This counterintuitive result challenges the assumption that voltage drop is always a reliable early warning sign of impending failure. Instead, the data suggest that in highly charged batteries, voltage may remain relatively stable even as internal temperatures rise dangerously. This has direct implications for battery management systems, which often rely on voltage monitoring as a primary diagnostic tool. If a cell at 90% SOC shows minimal voltage fluctuation while rapidly heating, traditional BMS algorithms might fail to detect the threat until it’s too late.
The researchers emphasize that their model does not replace physical testing but complements it by enabling rapid scenario analysis under controlled conditions. For example, engineers can now simulate the effects of different cooling strategies, insulation materials, or module layouts without building multiple physical prototypes. They can also explore “what-if” scenarios, such as the impact of delayed fire suppression or the effectiveness of thermal barriers between cells.
One of the key advantages of the lumped model is its adaptability. Because it relies on parameters that can be obtained through standard laboratory tests—such as differential scanning calorimetry (DSC) for heat generation rates or impedance spectroscopy for internal resistance—it can be applied to various cell chemistries and form factors. This flexibility makes it a powerful tool for both academic researchers and industry professionals working on next-generation battery systems.
The study also contributes to a growing body of knowledge about the stages of thermal runaway propagation. Building on earlier work by Jiang Fachao and colleagues, who identified three phases—initial module-level spread, inter-module propagation, and flashover—the current research provides granular detail on the first phase. It demonstrates that the transition from single-cell failure to multi-cell ignition is not linear but exponential, with SOC acting as a multiplier for both speed and intensity.
From a practical standpoint, these findings support several operational recommendations. First, for applications where safety is paramount—such as public transit, emergency vehicles, or fleet operations—it may be prudent to limit maximum charging to 80% or 90% rather than 100%. While this slightly reduces driving range, it significantly lowers the risk of cascading failures. Second, battery management systems should incorporate temperature-based triggers alongside voltage and current monitoring, especially during charging cycles when SOC is highest. Third, passive safety features such as phase-change materials, ceramic coatings, or enhanced airflow channels should be optimized for high-SOC scenarios.
Moreover, the research underscores the importance of standardized testing protocols that account for SOC variability. Current safety standards often evaluate batteries under fixed charge conditions, but real-world usage involves constant fluctuations. A vehicle involved in a collision after a long highway drive—when the battery is likely near full charge—faces a fundamentally different risk profile than one that has been partially discharged. Regulatory bodies and testing laboratories may need to update their evaluation frameworks to reflect this reality.
The computational approach also opens new avenues for predictive maintenance and failure forecasting. By integrating real-time telemetry data from vehicles into digital twin models, fleet operators could potentially identify early signs of cell degradation or thermal instability before they escalate. Such systems could alert drivers to seek service or automatically adjust charging behavior to reduce stress on vulnerable cells.
While the study focuses on NCM111 chemistry, the principles are likely applicable to other nickel-rich cathode materials such as NCM622 or NCM811, which are increasingly common in high-performance EVs. These chemistries offer higher energy density but are generally considered less thermally stable. Future work could extend the model to compare different cathode formulations, assess the impact of aging on thermal propagation, or evaluate the performance of solid-state batteries under similar stress conditions.
In conclusion, the research conducted by Cheng Lu, Liu Liang, Ye Guojun, and Tang Qiong represents a significant step forward in understanding and mitigating the risks associated with lithium-ion battery thermal runaway. By combining accessible modeling techniques with rigorous simulation, they have provided actionable insights into one of the most pressing challenges in modern transportation. Their work highlights the critical role of SOC management in battery safety and offers a robust framework for evaluating thermal propagation in multi-cell systems.
As the automotive industry moves toward greater electrification, studies like this will become increasingly vital. They not only advance scientific understanding but also inform engineering decisions that directly affect consumer safety. With millions of electric vehicles on the road and many more coming, ensuring the reliability and resilience of their power sources is not just a technical challenge—it’s a societal imperative.
Cheng Lu, Liu Liang, Ye Guojun, Tang Qiong. Higher SOC Increases Risk of Battery Thermal Runaway, Study Finds. Journal of Wuhan Engineering Polytechnic, 2024. DOI: 10.13542/j.cnki.1671-3524.2024.01.001