With the global push towards carbon neutrality, the demand for efficient and safe energy storage solutions has surged. Lithium-ion batteries, particularly the lithium iron phosphate (LiFePO4) variant, have become a cornerstone technology for electric vehicles and large-scale stationary energy storage systems due to their intrinsic safety, long cycle life, and cost-effectiveness. However, as the energy density and capacity of individual LiFePO4 battery cells increase to meet market demands—now routinely exceeding 50 Ah and reaching capacities like 86 Ah—the potential severity of thermal runaway (TR) events escalates proportionally. Thermal runaway remains a critical failure mode, a self-sustaining exothermic reaction that can lead to fire, explosion, and the release of toxic gases. While extensive research exists on small-format cylindrical cells, the thermal runaway characteristics of large-format, prismatic LiFePO4 batteries under different states of charge (SOC) are not yet fully quantified. This knowledge gap poses a significant challenge for designing effective safety management systems for modern battery energy storage systems (BESS).
This study aims to bridge this gap by conducting a systematic experimental investigation into the thermal runaway behavior of a commercial 86 Ah prismatic LiFePO4 battery. The core objective is to elucidate the influence of the battery’s State of Charge (SOC) on the initiation, progression, and severity of thermal runaway triggered by external heating. Understanding how key parameters such as temperature, voltage, and mass evolve during a TR event for different SOC levels is paramount for developing predictive models and early warning strategies. We employ a purpose-built experimental platform to subject batteries at 25%, 50%, 75%, and 100% SOC to a constant external heat flux, meticulously recording their response.
1. Experimental Methodology
1.1 Battery Cell Specifications and Preparation
The subject of this investigation is a large-format, prismatic LiFePO4 battery cell with a nominal capacity of 86 Ah. The key physical and electrical parameters of this LiFePO4 battery are summarized in Table 1. Prior to thermal abuse testing, each test LiFePO4 battery underwent three full charge-discharge cycles using a professional battery cycler to ensure consistency and stabilize its electrochemical state. The cycling protocol involved constant-current discharge to a cut-off voltage of 2.5 V at a 0.5C rate (43 A), followed by constant-current constant-voltage (CC-CV) charge to 3.65 V with a termination current of 0.03C (2.58 A). Following this conditioning, the LiFePO4 batteries were charged to the target SOC levels (25%, 50%, 75%, 100%) and left to rest for 24 hours to achieve equilibrium before testing.
| Parameter | Value / Specification |
|---|---|
| Dimensions (L × W × H) | 174 mm × 48 mm × 132 mm |
| Nominal Capacity | 86 Ah |
| Nominal Voltage | 3.6 V |
| Charge/Discharge Voltage Range | 2.5 V – 3.65 V |
| Mass | 2100 ± 2 g |
| Cathode Chemistry | Lithium Iron Phosphate (LiFePO4) |
1.2 Thermal Runaway Test Platform and Instrumentation
All experiments were conducted inside a dedicated combustion chamber with dimensions of 3.0 m × 3.0 m × 2.5 m, providing a controlled environment for safety and data collection. The experimental setup, as illustrated in the schematic, consisted of several key components. The thermal abuse was induced by a 450 W electric heating plate, which was placed in full surface contact with one of the large lateral sides of the LiFePO4 battery. To minimize heat loss to the environment and ensure consistent heating, the assembly was insulated using high-temperature ceramic wool. A custom clamping fixture with screw bolts secured the LiFePO4 battery against the heater, preventing displacement due to possible swelling.
Comprehensive data acquisition was central to this study. Temperature was measured at eight critical locations using 1 mm diameter K-type sheathed thermocouples (TCs):
- TC1: Center of the heated surface (in contact with heater).
- TC2: Center of the surface opposite the heater.
- TC3, TC4, TC5, TC6: Four points distributed on the surface opposite the heater.
- TC7: Center of one adjacent side surface.
- TC8: Positioned 2 cm directly above the pressure relief valve (PRV).
The average temperature of the back surface \(T_{avg}\) is calculated as:
$$T_{avg} = \frac{T_2 + T_3 + T_4 + T_5 + T_6}{5}$$
This \(T_{avg}\) serves as a representative temperature for analyzing the bulk thermal behavior of the LiFePO4 battery. Cell voltage was monitored directly via probes connected to the terminals. The entire mass of the battery-heater assembly was placed on a high-precision electronic balance to record real-time mass loss. A digital camera recorded the visual phenomena throughout the event. All data streams were synchronized and recorded at a frequency of 1 Hz.
1.3 Test Matrix and Procedure
The experimental matrix comprised four distinct test conditions, differentiated solely by the initial SOC of the LiFePO4 battery, as shown in Table 2.
| Test Case | Initial State of Charge (SOC) |
|---|---|
| #1 | 25% |
| #2 | 50% |
| #3 | 75% |
| #4 | 100% |
The standardized test procedure was as follows:
- Instrumentation: The prepared LiFePO4 battery was instrumented with thermocouples and voltage leads, connected to the data acquisition system outside the chamber.
- Initiation: The heating plate power was switched on, simultaneously triggering the start of data recording for temperature, voltage, mass, and video.
- Observation & Termination: The experiment continued until all visible reactions ceased and the LiFePO4 battery began to cool. The heater was then de-energized, and the chamber’s exhaust fan was activated to remove smoke and gases.
2. Results and Analysis
2.1 Phenomenological Stages of Thermal Runaway
Visual observations revealed that the thermal runaway process for the large-capacity LiFePO4 battery, regardless of SOC, could be consistently segmented into four distinct stages. However, the duration and intensity of each stage were highly SOC-dependent.
Stage 1: Pre-heating and Gas Generation. Upon heater activation, the external plastic wrap and adhesive tapes on the LiFePO4 battery surface began to smolder, producing minor white smoke. As the temperature rose internally, endothermic and initial exothermic reactions commenced. These include electrolyte evaporation and, critically, the decomposition of the Solid Electrolyte Interphase (SEI) layer on the anode. The SEI decomposition, typically beginning around 80-120°C for graphite anodes, is a key early exothermic reaction that releases heat and generates gases (e.g., CO2, C2H4). The internal pressure builds, causing visible swelling of the LiFePO4 battery casing.
Stage 2: Pressure Relief Valve (PRV) Actuation and Venting. Once the internal gas pressure exceeds the mechanical threshold of the PRV, it ruptures. This event is marked by a sudden, forceful jet of white smoke/vapor, primarily consisting of electrolyte solvent droplets and decomposition gases like H2, CO, and light hydrocarbons. Following the initial jet, continuous but gradually slowing venting is observed.
Stage 3: Thermal Runaway and Intense Ejection. This is the critical stage. As temperature continues to rise (often accelerated by reactions between the lithiated anode and electrolyte), the internal separator begins to shrink and melt. When the separator fully collapses, a large-scale internal short circuit occurs, dumping the stored electrical energy almost instantaneously into heat. This triggers violent chain reactions (e.g., cathode decomposition, electrolyte oxidation) and leads to a second, much more intense ejection of smoke, flames (if combustible gas concentrations are right), and particulate matter. The temperature surges dramatically.
Stage 4: Reaction Quenching and Cooling. After the rapid exothermic reactions subside due to the consumption of active materials, the ejection ceases. The LiFePO4 battery enters a slow cooling phase, with residual smoldering possible.
The timing of these stages varied significantly with SOC. Key characteristic times are summarized in Table 3.
| SOC (%) | PRV Opening Time, tvent (s) | TR Onset Time, tTR (s) | TR Duration (s) | Time from PRV to TR (s) |
|---|---|---|---|---|
| 25 | 1003 | 2665 | N/A (No full TR) | 1662 |
| 50 | 856 | 2184 | ~45 | 1328 |
| 75 | 842 | 1599 | ~60 | 757 |
| 100 | 812 | 1249 | ~75 | 437 |
The data shows a clear trend: higher SOC accelerates the entire process. The PRV opened 19% earlier for the 100% SOC LiFePO4 battery compared to the 25% SOC one. More strikingly, the onset of full thermal runaway occurred 53.1% sooner. The interval between venting and TR also shortened dramatically from over 1600 seconds at 25% SOC to just 437 seconds at 100% SOC. This indicates that a higher SOC not only increases gas generation (earlier venting) but also drastically reduces the thermal stability margin after venting, leading to a much faster transition to catastrophic failure.
2.2 Thermal Dynamics and Temperature Signatures
The temperature profiles provide quantitative insight into the heat generation and dissipation during the failure of the LiFePO4 battery. Figure 3 (representative of 100% SOC) shows the spatial temperature distribution. TC1, in direct contact with the heater, shows the steepest initial rise. The back-surface average temperature \(T_{avg}\) offers a robust metric for overall cell thermal state.
Several characteristic temperatures define the LiFePO4 battery’s journey to thermal runaway:
- Vent Temperature (\(T_{vent}\)): The \(T_{avg}\) at the moment of PRV opening.
- Thermal Runaway Trigger Temperature (\(T_{TR}\)): The \(T_{avg}\) at which the temperature rise becomes self-accelerating and irreversible, often defined when \(dT_{avg}/dt\) exceeds a threshold (e.g., 0.5°C/s).
- Maximum Temperature (\(T_{max}\)): The peak \(T_{avg}\) achieved during the event.
- Jet Temperature (\(T_{jet,max}\)): The peak temperature measured by TC8 above the PRV.
These temperatures for all SOC levels are compiled in Table 4.
| SOC (%) | \(T_{vent}\) (°C) | \(T_{TR}\) (°C) | \(T_{max}\) (°C) | \(T_{jet,max}\) (°C) |
|---|---|---|---|---|
| 25 | 101.5 | N/A | 237.1 | 184.9 |
| 50 | 98.2 | 165.8 | 278.6 | 215.4 |
| 75 | 96.7 | 164.1 | 335.3 | 242.7 |
| 100 | 94.3 | 160.4 | 378.9 | 270.1 |
The analysis reveals significant SOC dependence:
- \(T_{vent}\) and \(T_{TR}\) decrease with increasing SOC. A higher SOC means more lithium is intercalated in the anode graphite. This lithiated carbon is more reactive, facilitating earlier and more vigorous reactions with the electrolyte, thus lowering the temperature required to generate enough gas for venting and to trigger the catastrophic chain reaction.
- \(T_{max}\) and \(T_{jet,max}\) increase with increasing SOC. This is a direct consequence of the greater chemical energy stored in the LiFePO4 battery. The 100% SOC cell reached a \(T_{max}\) 60% higher than the 25% SOC cell. The hotter jet from the higher SOC cells indicates more energetic ejection of hot gases and decomposition products.
The rate of temperature change, or temperature derivative \(dT/dt\), is a crucial indicator of reaction intensity. We calculate the average back-surface温升速率 as:
$$\frac{dT_{avg}}{dt} = \frac{1}{n}\sum_{i=1}^{n} \frac{dT_i}{dt}$$
where \(T_i\) are the back-surface thermocouples. Figure 5 shows the \(dT_{avg}/dt\) profiles. Two critical extrema are observed:
- Maximum Cooling Rate at Venting (\(dT/dt_{min}\)): A sharp negative spike occurs at PRV opening as the rapid ejection of hot gases cools the cell surface momentarily.
- Maximum Heating Rate during TR (\(dT/dt_{max}\)): A massive positive spike occurs at the onset of thermal runaway, indicating the explosive release of heat.
The values of these extrema are key severity metrics, shown in Table 5.
| SOC (%) | \(dT/dt_{min}\) at Venting (°C/s) | \(dT/dt_{max}\) during TR (°C/s) |
|---|---|---|
| 25 | -1.66 | 0.33 |
| 50 | -0.74 | 0.84 |
| 75 | -0.71 | 1.09 |
| 100 | -0.67 | 2.58 |
The trend is revealing: the cooling spike at venting becomes less negative with higher SOC. This suggests that for a highly charged LiFePO4 battery, the ongoing internal exothermic reactions are so powerful that they quickly compensate for the cooling effect of venting. Most critically, the maximum heating rate \(dT/dt_{max}\) increases dramatically with SOC. The 100% SOC LiFePO4 battery’s heating rate was over 7 times greater than that of the 50% SOC cell and nearly 8 times that of the 25% SOC cell. This quantitatively confirms that the violence and speed of the thermal runaway event in a LiFePO4 battery are strongly correlated with its state of charge.
2.3 Electrochemical Response: Voltage Behavior
The cell voltage provides an electrical signature of the internal degradation. The voltage profiles for all SOC levels are shown in Figure 6. A common pattern is observed, but with shifted timelines. The voltage remains relatively stable during the initial heating phase. As internal damage accumulates (SEI decomposition, minor internal shorts), a slow, gradual voltage drop begins. This is followed by a rapid voltage plunge to zero, which signifies the occurrence of a massive internal short circuit due to separator failure.
The timing of this voltage plunge relative to other events is critical for early warning. We define \(t_{Vdrop}\) as the time when the voltage begins its rapid, irreversible descent. Table 6 compares \(t_{Vdrop}\) with the thermal runaway onset time \(t_{TR}\).
| SOC (%) | Voltage Drop Start, \(t_{Vdrop}\) (s) | TR Onset, \(t_{TR}\) (s) | Warning Lead Time, \(\Delta t = t_{TR} – t_{Vdrop}\) (s) |
|---|---|---|---|
| 25 | 2195 | 2665 | 470 |
| 50 | 1906 | 2184 | 278 |
| 75 | 1397 | 1599 | 202 |
| 100 | 1094 | 1249 | 155 |
A crucial finding is that the voltage drop consistently precedes the thermal runaway onset (defined by the temperature spike) by a significant interval, ranging from 155 to 470 seconds. This \(\Delta t\) represents a valuable window for early warning. Although the lead time decreases with higher SOC, it remains present. Furthermore, the onset of the slow voltage decay happens even earlier. Monitoring the derivative of voltage \(dV/dt\) could therefore serve as a highly effective, cell-internal precursor signal for impending thermal runaway in a LiFePO4 battery, complementing temperature-based detection.
2.4 Mass Loss and Energetic Severity
The mass loss during a thermal runaway event quantifies the ejection of volatile components (electrolyte, decomposed products) and is a direct measure of the severity of the cell’s disintegration. The real-time mass \(m(t)\) was recorded, and the mass loss rate \(\dot{m}\) was calculated as:
$$\dot{m} = \frac{dm}{dt}$$
The final mass loss \(\Delta m\) and the peak mass loss rate \(\dot{m}_{max}\) are summarized in Table 7.
| SOC (%) | Total Mass Loss, \(\Delta m\) (g) | \(\Delta m\) as % of 100% SOC loss | Peak Mass Loss Rate, \(\dot{m}_{max}\) (g/s) |
|---|---|---|---|
| 25 | 352.1 | 79.6% | 1.2 |
| 50 | 392.3 | 88.7% | 2.7 |
| 75 | 419.3 | 94.8% | 7.3 |
| 100 | 442.1 | 100% | 8.2 |
The data shows a strong positive correlation between SOC and both total mass loss and peak ejection rate. The 100% SOC LiFePO4 battery lost approximately 21% of its initial mass, significantly more than lower SOC cells. The peak mass loss rate \(\dot{m}_{max}\), which typically occurs during the intense ejection of Stage 3, increases by nearly an order of magnitude from the 25% SOC to the 100% SOC condition. This quantifies the dramatically more violent ejection process in a fully charged LiFePO4 battery. The mass curve also shows a momentary increase at the instant of PRV opening due to the reaction force of the high-speed jet on the scale, with this “force spike” being more pronounced at higher SOCs.
2.5 Post-Mortem Analysis
Physical inspection of the LiFePO4 batteries after testing revealed a clear gradient of damage severity aligned with SOC. The 100% SOC cell exhibited the most severe deformation: pronounced casing swelling, heavy sooting and burn marks covering the top and sides, and significant melting/distortion of the negative terminal. The 75% SOC cell showed similar but less extensive burn marks. The 50% SOC cell had localized ejecta deposition near the PRV. The 25% SOC LiFePO4 battery casing remained largely intact with only minor soiling from vented electrolyte droplets, confirming it underwent venting but not a full thermal runaway. This visual evidence reinforces the quantitative data: the higher the SOC, the greater the chemical energy released, leading to more complete and violent cell destruction.
3. Discussion: Implications for Safety Management of LiFePO4 Battery Systems
The experimental findings on this large-capacity LiFePO4 battery have several critical implications for the safety design and operation of battery energy storage systems.
1. SOC as a Primary Risk Factor: This study unequivocally establishes that the State of Charge is the dominant factor governing the severity and progression speed of thermal runaway in a LiFePO4 battery. A fully charged (100% SOC) cell presents a fundamentally different hazard profile compared to a partially charged one. Safety protocols, particularly for storage, transportation, and maintenance, should mandate lower SOC levels (e.g., below 30%) to intrinsically reduce the risk and potential consequences of a failure.
2. Early Warning Signals: The results identify multiple, tiered precursor signals that can be leveraged for early warning systems:
- Voltage Decline: The slow voltage drop beginning hundreds of seconds before TR onset is a highly reliable electrical precursor. Monitoring \(dV/dt\) could provide the earliest software-based warning.
- Pressure/Venting: The activation of the PRV is a clear physical indicator that the LiFePO4 battery is in a critical state. Gas/smoke detectors are essential for catching this event.
- Temperature Acceleration: The point where \(dT/dt\) sustainably exceeds a threshold (e.g., 0.5°C/s) marks the irreversible onset of TR and should trigger immediate countermeasures (e.g., deluge systems).
A multi-parameter protection strategy is recommended, as summarized in Table 8.
| Warning Stage | Primary Signal | Proposed Action |
|---|---|---|
| Early Alert | Abnormal voltage drop (\(dV/dt <\) threshold) | Increase monitoring frequency; Check adjacent cells; Consider pre-cooling. |
| Critical Alarm | PRV activation (detected by smoke/gas sensor or acoustic sensor) | Activate local fire suppression; Isolate the affected module/pack. |
| Emergency Mitigation | Rapid temperature rise (\(dT/dt > 0.5\) °C/s) | Trigger full emergency protocols: Deluge, ventilation, evacuation. |
3. Thermal Runaway Propagation Risk: The higher \(T_{max}\), \(T_{jet,max}\), and \(\dot{m}_{max}\) from a high-SOC LiFePO4 battery mean it can eject more energetic material over a greater distance. This significantly increases the risk of cascading thermal runaway in a densely packed module. Propagation mitigation design (e.g., thermal barriers, adequate spacing, directed venting channels) must account for this SOC-dependent ejection severity.
4. Material and Design Insights: The lower \(T_{vent}\) and \(T_{TR}\) at high SOC point to the thermal instability of the highly lithiated anode. This underscores the importance of ongoing research into more thermally stable anode materials, advanced separators with higher melt integrity, and electrolytes with wider stability windows to improve the intrinsic safety of the LiFePO4 battery chemistry itself.
4. Conclusion
This comprehensive experimental study has systematically characterized the thermal runaway behavior of an 86 Ah large-format LiFePO4 battery under external heating abuse, with a focus on the influence of the State of Charge. The key conclusions are as follows:
- The thermal runaway process accelerates dramatically with increasing SOC. The time to pressure relief valve opening and the time to full thermal runaway onset are significantly shortened. The 100% SOC LiFePO4 battery progressed to catastrophic failure 53.1% faster than the 25% SOC cell.
- The severity of thermal runaway, quantified by maximum temperature, temperature rise rate, mass loss, and ejection violence, scales positively with SOC. The fully charged LiFePO4 battery reached a peak temperature 60% higher and a mass loss rate nearly 7 times greater than the 50% SOC cell.
- Critical characteristic temperatures (\(T_{vent}\), \(T_{TR}\)) decrease with increasing SOC, indicating reduced thermal stability for a highly charged LiFePO4 battery due to the higher reactivity of the lithiated anode.
- The cell voltage provides a crucial early warning signal. A detectable voltage drop precedes the thermal runaway temperature spike by 155 to 470 seconds, offering a valuable window for preventive action. This lead time, while shorter for higher SOC, is consistently present.
- For the specific LiFePO4 battery tested, a SOC of 25% was insufficient to sustain a full thermal runaway chain reaction under the applied abuse condition, resulting only in safe venting. This highlights the critical risk-mitigation value of maintaining large-capacity LiFePO4 batteries at lower states of charge during periods of elevated risk.
These findings provide essential quantitative data for improving the safety design of battery packs and energy storage systems utilizing large-capacity LiFePO4 batteries. They strongly advocate for the implementation of SOC-aware safety protocols and the development of multi-parameter, model-based early warning systems that integrate voltage, temperature, and gas detection signals to reliably predict and mitigate thermal runaway events.