With the rapid growth of energy storage systems, lithium-ion battery energy storage stations have become the predominant form due to their high energy density, long cycle life, and flexibility. However, the fire safety hazards associated with these stations cannot be ignored. Current fire warning methods often suffer from issues such as low detection accuracy, gas cross-interference, and sensor poisoning. In this research, we delve into the causes, development processes, critical conditions, and fundamental principles of thermal runaway in lithium-ion batteries. By comprehensively analyzing various monitoring technologies, we propose a multi-parameter early warning method based on temperature, characteristic gases, and solid particles. We also establish an experimental platform to validate this approach. Our findings indicate that the multi-parameter monitoring method offers excellent timeliness, reliable warnings, and accurate detection.
The increasing deployment of lithium-ion battery energy storage stations highlights the urgent need for effective safety measures. Thermal runaway in lithium-ion batteries is a complex phenomenon that can lead to fires and explosions, causing significant economic losses and hindering the development of the energy storage industry. Therefore, developing advanced warning technologies is crucial. In this article, we present our research on thermal runaway warning technology from a first-person perspective, focusing on the integration of multiple parameters for early detection.
Lithium-ion battery thermal runaway typically originates from a series of exothermic reactions, including SEI film decomposition, reactions between the anode and electrolyte, separator shrinkage and melting, electrolyte decomposition, and cathode decomposition. These reactions often overlap and alternate, leading to a cascade effect. The heat generation in a lithium-ion battery can be described by the following equation:
$$ Q = I^2 R + \sum \Delta H_i r_i $$
where \( Q \) is the total heat generation rate, \( I \) is the current, \( R \) is the internal resistance, \( \Delta H_i \) is the enthalpy change of reaction \( i \), and \( r_i \) is the reaction rate. When the heat generation exceeds dissipation, thermal runaway occurs. For lithium-ion batteries, this critical condition can be expressed as:
$$ \frac{dT}{dt} = \frac{Q – hA(T – T_{\text{ambient}})}{mC_p} > 0 $$
where \( \frac{dT}{dt} \) is the temperature rise rate, \( h \) is the heat transfer coefficient, \( A \) is the surface area, \( T \) is the battery temperature, \( T_{\text{ambient}} \) is the ambient temperature, \( m \) is the mass, and \( C_p \) is the specific heat capacity. This equation highlights the importance of temperature monitoring in lithium-ion battery systems.
To understand the features of thermal runaway in lithium-ion batteries, we analyze both internal and external characteristics. Internally, temperature and voltage are key parameters. Externally, gas and smoke emissions are critical indicators. Below, we summarize the characteristic parameters of lithium-ion battery thermal runaway.
| Parameter | Description | Typical Range or Value |
|---|---|---|
| Temperature | Surface temperature rise during thermal runaway | 50°C to 200°C |
| Voltage | Voltage drop or surge during abuse conditions | 0 V to 5.5 V |
| CO Concentration | Characteristic gas from electrolyte decomposition | Up to 200 ppm |
| H₂ Concentration | Gas produced during reactions | Up to 80 ppm |
| VOC Concentration | Volatile organic compounds from decomposition | Up to 300 ppm |
| Solid Particles | Smoke particles from thermal runaway | 0.1 μm to 10 μm in diameter |
The gas composition during thermal runaway of a lithium-ion battery is diverse. Based on experimental data, we can represent the volume fractions of major gases as follows:
$$ V_{\text{CO}} = 0.5105, \quad V_{\text{H}_2} = 0.1771, \quad V_{\text{CO}_2} = 0.1121, \quad V_{\text{HC}} = 0.1753 $$
where \( V \) denotes volume fraction. These gases are released rapidly after the safety valve opens, making them effective early warning indicators. For solid particles, the size distribution can be modeled using a log-normal distribution:
$$ f(d) = \frac{1}{d \sigma \sqrt{2\pi}} \exp\left( -\frac{(\ln d – \mu)^2}{2\sigma^2} \right) $$
where \( d \) is the particle diameter, \( \mu \) is the mean, and \( \sigma \) is the standard deviation. In our experiments, the median diameter was around 290 nm, indicating that nanoparticles are prevalent during thermal runaway of lithium-ion batteries.
Our proposed multi-parameter warning system integrates temperature, characteristic gases (CO, H₂, VOC), and solid particles (0.1–10 μm). The logic for triggering warnings is based on the fusion of these parameters. If one parameter shows abnormal behavior, a caution signal is issued; if two or more parameters are abnormal, a full warning is triggered, and fire prevention measures are activated. This approach reduces false alarms and improves reliability compared to single-parameter methods.
To validate the system, we designed an experimental platform simulating a typical energy storage station environment. The platform included a sealed chamber with dimensions of 2.35 m (width) × 2.39 m (height) × 3.5 m (length). We used lithium-ion battery samples, specifically LiFePO₄ cells with capacities of 50 Ah and 280 Ah, to conduct overcharge and overheating tests. The sensors deployed were:
- Infrared thermal imager (FLIR SE-187) for temperature monitoring.
- Gas sensors for CO, H₂, and VOC.
- Dust particle counter (CLJ-E type) for solid particle detection.
- Traditional smoke detectors and aspirating smoke detectors for comparison.
The experimental procedures followed standards such as GB/T 36276-2018. For overcharge tests, we charged the lithium-ion battery at 1C rate until the safety valve opened. For overheating tests, we heated the battery at 5°C/min to 130°C and maintained the temperature. Each test was repeated three times to ensure consistency.
The results from our experiments are summarized in the tables below. Table 1 shows the peak concentrations of characteristic gases during thermal runaway for different test conditions.
| Test Condition | CO Peak (ppm) | H₂ Peak (ppm) | VOC Peak (ppm) |
|---|---|---|---|
| 50 Ah Overcharge | 183–225 | 62–78 | 305–313 |
| 280 Ah Overcharge | Full scale | Full scale | Full scale |
| 50 Ah Overheat (50% SOC) | 173–190 | 74–81 | 318–332 |
| 50 Ah Overheat (100% SOC) | 188–193 | 74–83 | 318–327 |
Table 2 presents the solid particle counts for various particle sizes during thermal runaway. The data indicate a significant increase in particle numbers after the safety valve opens.
| Particle Diameter (μm) | Normal Mean (counts) | Peak During Thermal Runaway (counts) |
|---|---|---|
| 0.3 | 3,768–4,431 | 2,743,500–3,624,000 |
| 0.5 | 1,161–1,447 | 1,498,300–3,171,000 |
| 1.0 | 243–319 | 1,003,600–2,406,000 |
| 2.5 | 20–52 | 736,300–2,121,000 |
The temperature data during thermal runaway are critical. We observed that the lithium-ion battery surface temperature rose from room temperature to around 50°C during the initial stage, with bulging occurring. At approximately 80°C, the safety valve opened, releasing gases and particles. The temperature rise can be modeled using an exponential function:
$$ T(t) = T_0 + \Delta T \left(1 – e^{-kt}\right) $$
where \( T_0 \) is the initial temperature, \( \Delta T \) is the total temperature increase, and \( k \) is a rate constant dependent on the lithium-ion battery’s thermal properties.
To compare the effectiveness of our multi-parameter warning method, we measured the response times of different detection systems. Table 3 shows the response times for various methods during overcharge tests.
| Warning Method | Response Time (s) in Overcharge Test |
|---|---|
| Our Multi-parameter Method | 2,291–2,983 |
| Traditional Smoke Detector | 2,570–3,106 or no alarm |
| Battery Management System (BMS) | 2,365–3,091 |
| Gas Detector Alone | 2,371–3,099 |
Our multi-parameter method demonstrated earlier warnings by up to 123 seconds compared to traditional smoke detectors and up to 74 seconds compared to BMS. This highlights the advantage of integrating multiple parameters for lithium-ion battery thermal runaway detection.
Furthermore, we analyzed the voltage characteristics during thermal runaway. For overcharge conditions, the voltage of the lithium-ion battery increases steadily until a peak, then drops to zero. For mechanical abuse, such as nail penetration, the voltage drops abruptly. However, voltage alone is unreliable due to fluctuations from external disturbances. Therefore, our method does not rely solely on voltage but incorporates it as a supplementary parameter when available.
The solid particle analysis revealed that particles in the range of 0.1–10 μm are highly indicative of thermal runaway in lithium-ion batteries. The particle count increase follows a sigmoidal curve, which can be described as:
$$ N(t) = \frac{N_{\text{max}}}{1 + e^{-a(t – t_0)}} $$
where \( N(t) \) is the particle count at time \( t \), \( N_{\text{max}} \) is the maximum count, \( a \) is a growth rate constant, and \( t_0 \) is the time of safety valve opening. This model helps in setting thresholds for early warnings.
In terms of gas dynamics, the release of characteristic gases from a lithium-ion battery during thermal runaway can be approximated by a diffusion-controlled process. The concentration \( C \) of a gas at a distance \( x \) from the battery can be estimated using Fick’s law:
$$ \frac{\partial C}{\partial t} = D \frac{\partial^2 C}{\partial x^2} + S(t) $$
where \( D \) is the diffusion coefficient, and \( S(t) \) is the source term representing gas emission from the lithium-ion battery. By monitoring \( C \) at strategic points, we can detect thermal runaway early.
Our experimental platform also allowed us to study the effects of state of charge (SOC) on thermal runaway. For lithium-ion batteries at 100% SOC, the critical temperature for thermal runaway is lower than at 50% SOC. This can be expressed by an Arrhenius-type relationship:
$$ t_{\text{TR}} = A e^{\frac{E_a}{RT}} $$
where \( t_{\text{TR}} \) is the time to thermal runaway, \( A \) is a pre-exponential factor, \( E_a \) is the activation energy, \( R \) is the gas constant, and \( T \) is the temperature. Higher SOC reduces \( E_a \), accelerating thermal runaway in lithium-ion batteries.
Based on our findings, we propose a comprehensive warning algorithm that fuses data from temperature, gas, and particle sensors. The algorithm uses weighted sums or machine learning techniques to compute a risk score \( R \):
$$ R = w_T \cdot \Delta T + w_G \cdot \sum C_i + w_P \cdot \Delta N $$
where \( w_T \), \( w_G \), and \( w_P \) are weights for temperature, gas, and particle parameters, respectively; \( \Delta T \) is the temperature rise rate; \( C_i \) are gas concentrations; and \( \Delta N \) is the particle count increase. If \( R \) exceeds a threshold, a warning is issued. This approach minimizes false positives and ensures timely alerts for lithium-ion battery systems.
To enhance the practicality of our method, we considered integration with existing battery management systems. The multi-parameter sensors can be deployed in a distributed network within energy storage stations, communicating via IoT protocols. Data fusion centers can process information in real-time, providing actionable insights for operators.
In conclusion, our research on thermal runaway warning technology for lithium-ion batteries in energy storage stations demonstrates that a multi-parameter approach based on temperature, characteristic gases, and solid particles is highly effective. The method offers superior timeliness, reliability, and accuracy compared to traditional single-parameter systems. Future work will focus on optimizing sensor placement, improving algorithm efficiency, and scaling the system for large-scale lithium-ion battery installations. By advancing these technologies, we can significantly enhance the safety of energy storage stations and support the sustainable growth of the lithium-ion battery industry.
The importance of lithium-ion battery safety cannot be overstated, especially as global adoption increases. Our study contributes to this field by providing a robust framework for early thermal runaway detection. We encourage further research into advanced materials and sensors to mitigate risks associated with lithium-ion batteries. Through collaborative efforts, we can develop smarter, safer energy storage solutions that leverage the full potential of lithium-ion battery technology.