The imperative for energy storage solutions has become a cornerstone of global energy strategy in the face of climate challenges. Among the various technologies, lithium-ion battery-based energy storage stands as a core pillar due to its advantages in power and energy density, cycle life, and environmental profile. However, the rapid scaling of this technology brings forth significant safety concerns, with thermal runaway incidents in battery energy storage systems (BESS) posing serious risks of fire, explosion, and economic loss. Therefore, in-depth research into the thermal runaway propagation behavior within battery packs and the development of effective mitigation strategies are paramount for enhancing the safety and reliability of large-scale energy storage installations.
This investigation focuses on the thermal safety management of high-capacity lithium-ion battery systems. Combing theoretical analysis with experimental testing, an immersion liquid cooling scheme is employed to study the characteristics of thermal runaway behavior under overcharge conditions. The study explores the influence mechanisms of factors such as the type of immersion liquid and the spacing between cells on critical parameters like voltage, temperature, and heat propagation during a module-level thermal runaway event. The findings are systematically compared with the behavior of conventional non-immersion modules. The core objective is to provide a detailed, first-person account of the experimental process, results, and implications, utilizing tables and formulas for comprehensive summarization.
The fundamental issue stems from the complex electrochemistry of the lithium-ion battery. Under abusive conditions like overcharging, a cascade of exothermic reactions can be triggered within the cell. This sequence often begins with the breakdown of the solid-electrolyte interphase (SEI) layer, proceeds to reactions between the anode and electrolyte, and can culminate in the decomposition of the cathode material and electrolyte. The heat release rate (HRR) from these reactions can be described in a simplified form as the sum of individual reaction heats:
$$ \dot{Q}_{total} = \sum_i \dot{Q}_i = \sum_i A_i \exp\left(-\frac{E_{a,i}}{RT}\right) f_i(\text{SOC}) $$
where $\dot{Q}_i$ is the heat generation rate of reaction $i$, $A_i$ is the pre-exponential factor, $E_{a,i}$ is the activation energy, $R$ is the universal gas constant, $T$ is the absolute temperature, and $f_i(\text{SOC})$ is a function of the state of charge. When the heat generation within a lithium-ion battery surpasses its heat dissipation capacity to the surroundings, thermal runaway ensues, characterized by a rapid, uncontrollable temperature rise, often exceeding 500–700°C, accompanied by venting of flammable gases and potential combustion or explosion.
In a module or pack configuration, the primary danger is the propagation of thermal runaway from one initiating cell to its neighbors. The heat transfer driving this propagation can occur via three main paths: conduction through busbars and structural components, convection via ejected gases and particles, and radiation from the high-temperature surfaces of the failing cell. The propagation risk $P$ can be conceptually modeled as a function of the initiating cell’s energy release $\Delta H_{TR}$, the thermal coupling $\kappa$ between cells, and the cooling efficiency $\eta_{cool}$ of the system:
$$ P \propto \frac{\Delta H_{TR} \cdot \kappa}{\eta_{cool}} $$
Conventional air-cooled or cold-plate-based systems often struggle to provide sufficient $\eta_{cool}$ to arrest propagation once a single high-capacity lithium-ion battery enters thermal runaway due to the limited heat capacity and thermal conductivity of air and the interfacial resistance between the cell casing and the cooling plate.
Immersion cooling presents a paradigm shift in thermal management for lithium-ion battery systems. By submerging cells directly in a dielectric fluid, several intrinsic safety advantages are realized. Firstly, the intimate contact eliminates interfacial thermal resistance, maximizing heat transfer efficiency. Secondly, the high volumetric heat capacity ($\rho C_p$) of the fluid acts as a significant thermal buffer, absorbing heat from the cell. Thirdly, by submerging the lithium-ion battery, the system inherently creates an oxygen-depleted environment around the cell, which can suppress combustion even if flammable gases are vented. The cooling power in an immersion system can be approximated by:
$$ \dot{Q}_{cool} = h A_s (T_{cell} – T_{fluid}) + \dot{m} C_{p,fluid} (T_{out} – T_{in}) $$
where $h$ is the convective heat transfer coefficient (significantly higher than in air), $A_s$ is the cell surface area, $T_{cell}$ and $T_{fluid}$ are the cell and local fluid temperatures, $\dot{m}$ is the fluid mass flow rate, and $C_{p,fluid}$ is the fluid’s specific heat capacity.
Experimental Methodology and Setup
The experimental research was designed to quantitatively evaluate the efficacy of immersion cooling in suppressing thermal runaway propagation within a module of high-capacity cells. The test subject was a commercial 280 Ah lithium iron phosphate (LFP) prismatic lithium-ion battery. Its key specifications are summarized below.
| Parameter | Value |
|---|---|
| Nominal Capacity | 280 Ah |
| Nominal Voltage | 3.2 V |
| Charge/Discharge Cut-off Voltage | 3.65 V / 2.5 V |
| Dimensions (L x W x H) | 173.6 mm × 71.25 mm × 199.7 mm |
| Mass | Approx. 5.4 kg |
A custom immersion test chamber was constructed, designed to house a three-cell module. The chamber included inlet and outlet ports for fluid circulation, a pressure relief valve (burst disk) to safely manage gas generation, and sealed conduits for electrical and sensor feedthroughs. The three cells were positioned in series along the flow direction, labeled B1 (upstream), B2 (middle), and B3 (downstream). Non-conductive, flame-retardant spacers were placed between cells to maintain precise gaps of either 1.5 mm or 2.0 mm, allowing fluid to pass through.
Each lithium-ion battery was instrumented with four K-type thermocouples: two on the large faces (centered), one on the narrow side, and one at the venting port. Voltage taps were connected to each cell’s terminals. Additional thermocouples monitored the fluid temperature at the chamber inlet and outlet. Data was recorded at a high frequency using a National Instruments data acquisition system. The immersion cooling loop consisted of a pump, a flow meter, a reservoir/expansion tank, and a heat exchanger connected to a constant-temperature bath set to 25°C. For baseline comparison, an identical three-cell module was tested in an open, non-immersed configuration (air environment).
The trigger mechanism for thermal runaway was selected to be overcharging, a common field failure scenario. The central cell (B2) was subjected to a constant current overcharge at 0.5C (140 A) using a high-power battery cycler until thermal runaway was initiated. The adjacent cells (B1 and B3) were left at their initial state of charge (approximately 3.3 V). The key experimental variables were:
- Cooling Mode: Non-immersion (air) vs. Immersion.
- Immersion Liquid Type: Two different dielectric synthetic hydrocarbon oils (Fluid A and Fluid B) with distinct thermophysical properties.
- Cell Spacing: 1.5 mm vs. 2.0 mm within the immersion chamber.
The thermophysical properties of the two immersion fluids are critical and are listed in the following table.
| Property | Fluid A | Fluid B |
|---|---|---|
| Kinematic Viscosity @ 40°C (mm²/s) | 6.05 | 12.38 |
| Thermal Conductivity (W/m·K) | 0.13 | 0.14 |
| Specific Heat Capacity (kJ/kg·K) | 1.97 | 2.02 |
| Flash Point (°C) | 143 | 177 |
Results and Analysis: Non-Immersion Baseline
To establish a baseline for the severity of thermal runaway propagation, the experiment began with the non-immersed module. The results were stark and demonstrated the classical, dangerous propagation behavior.
Voltage and Temperature Response: The overcharged B2 cell reached a peak voltage of 15.92 V after absorbing 58.9 Ah of excess charge. Its voltage then rapidly collapsed to 0 V within 60 seconds, signaling the onset of full internal short circuit and thermal runaway. During this event, the surface temperature of B2 skyrocketed to a maximum recorded value of 362.9°C. The violent energy release involved intense jet flames and sustained combustion of the ejected materials.
Thermal Propagation: The heat from the failing B2 cell successfully propagated to its neighbor, B3. The temperature at the top surface of B3 (closest to B2) rose sharply, followed by its side temperature, which peaked at 171.6°C. Subsequently, the pressure relief device (PRD) of B3 activated at approximately 3,188 seconds. Post-test inspection confirmed that B3 had also undergone thermal runaway, with its voltage measured at 1.221 V. Cell B1, on the other side of B2, experienced significant heating (peak temperature of 157.5°C) and severe swelling, indicating it was pushed to the brink of failure, though its PRD did not open and its voltage remained at 3.321 V. This experiment clearly resulted in a propagating thermal runaway event (B2 → B3), with B1 critically damaged.
The energy balance during propagation can be analyzed. The heat flux $\dot{q}”$ from the runaway cell (B2) to an adjacent cell (B3) in air is dominated by radiation and hot gas convection:
$$ \dot{q}” = \epsilon \sigma (T_{B2}^4 – T_{B3}^4) + h_{gas} (T_{gas} – T_{B3}) $$
where $\epsilon$ is emissivity, $\sigma$ is the Stefan-Boltzmann constant, and $h_{gas}$ is the convective coefficient from ejected gases. For a lithium-ion battery like the LFP type used, the total energy released during thermal runaway $\Delta H_{TR}$ can be substantial, often estimated in the range of 200-300 kJ per cell for this capacity. In air, with minimal cooling ($\eta_{cool} \approx 0$), the thermal coupling $\kappa$ is high enough for $P$ (propagation risk) to exceed the critical threshold.
Results and Analysis: Immersion Cooling Efficacy
The immersion cooling tests presented a dramatically different outcome, fundamentally altering the failure scenario from a propagating disaster to a contained single-cell event.
Containment of Propagation: Across all immersion tests, regardless of fluid type or spacing, thermal runaway propagation was completely prevented. While the intentionally overcharged B2 cell underwent thermal runaway, the adjacent B1 and B3 cells remained intact. Post-test, both B1 and B3 exhibited no swelling, their PRDs remained sealed, and their voltages were stable at around 3.3 V. There were no incidents of fire or explosion within the immersion chamber. The visual observation was of gas bubbling through the fluid and a gradual discoloration of the fluid around the failed cell, but no open flames.
Drastic Reduction in Peak Temperature: The maximum temperature attained by the failing B2 cell was severely curtailed by the immersion liquid. Compared to the 362.9°C peak in air, the peak temperatures in the immersion tests were 249.6°C (Fluid A, 1.5mm gap) and 236.7°C (Fluid B, 1.5mm gap). This represents a reduction of approximately 85% in temperature rise above ambient. This suppression is a direct consequence of the high heat capacity of the fluid, which absorbs a massive amount of energy during the phase of rapid heat generation. The energy absorbed by the fluid $E_{abs}$ can be expressed as:
$$ E_{abs} = m_{fluid} C_{p,fluid} \Delta T_{fluid} + \dot{m} C_{p,fluid} \int (T_{out}(t) – T_{in}) dt $$
where $m_{fluid}$ is the static mass of fluid in contact with the cell, and $\Delta T_{fluid}$ is its temperature rise. This energy absorption directly subtracts from the energy available to heat the cell itself and its neighbors.
Influence of Immersion Fluid Properties
The choice of immersion fluid had a measurable impact on the thermal runaway dynamics of the initiating cell. The comparative data is summarized below.
| Parameter | Fluid A | Fluid B | Difference (B – A) |
|---|---|---|---|
| Overcharge Capacity to Trigger (Ah) | 60.7 | 67.4 | +6.7 |
| Peak Cell B2 Voltage (V) | 16.39 | 20.22 | +3.83 |
| Time from PRD Open to Peak Temp (s) | 909 | 1,059 | +150 |
| Peak Temperature of B2 (°C) | 249.6 | 236.7 | -12.9 |
| Peak Temperature of B1 (°C) | 195.6 | 178.2 | -17.4 |
Fluid B, with its higher specific heat capacity (2.02 vs. 1.97 kJ/kg·K) and slightly higher thermal conductivity, demonstrated superior performance. It extended the “gestation period” – the time between initial abuse (PRD opening) and the peak thermal runaway temperature – by 150 seconds (16.5%). It also resulted in lower peak temperatures for both the failing cell B2 (12.9°C lower) and the adjacent cell B1 (17.4°C lower). The higher heat capacity provides a greater thermal sink, as shown by modifying the earlier heat balance. The rate of temperature rise for the lithium-ion battery is governed by:
$$ \frac{dT_{cell}}{dt} = \frac{\dot{Q}_{gen} – \dot{Q}_{cool}}{m_{cell} C_{p,cell}} $$
With immersion, $\dot{Q}_{cool}$ is greatly enhanced. For a fluid with higher $C_{p,fluid}$, the term $\dot{m} C_{p,fluid} (T_{out} – T_{in})$ is larger for a given flow rate and temperature differential, leading to a more negative $dT_{cell}/dt$, thereby delaying and attenuating the temperature spike.
Influence of Inter-Cell Spacing
The spacing between cells within the immersed module also played a significant role in managing the thermal environment, particularly for the neighboring cells. The results for two different spacings using Fluid A are compared.
| Parameter | 1.5 mm Spacing | 2.0 mm Spacing | Difference |
|---|---|---|---|
| Peak Temperature of B2 (°C) | 248.7 | 254.5 | +5.8 |
| Peak Temperature of B1 (°C) | 195.6 | 155.3 | -40.3 |
| Peak Temperature of B3 (°C) | 173.9 | 145.1 | -28.8 |
While the peak temperature of the initiating B2 cell was marginally higher with the larger gap (likely due to a slightly less effective local fluid circulation or a minor variation in the trigger point), the benefit to the module’s safety was profound. The peak temperatures experienced by the adjacent cells B1 and B3 were drastically lower with the 2.0 mm spacing, showing reductions of 40.3°C and 28.8°C, respectively. This demonstrates that increasing the spacing reduces the thermal coupling $\kappa$ between cells. The heat transfer between two immersed cells is a complex function of conduction through the fluid gap and convective flow. A larger gap increases the thermal resistance $R_{gap}$ for conductive heat transfer:
$$ R_{gap} \approx \frac{L_{gap}}{k_{fluid} A} $$
where $L_{gap}$ is the spacing, $k_{fluid}$ is the fluid’s thermal conductivity, and $A$ is the cross-sectional area for heat flow. A larger $L_{gap}$ directly increases $R_{gap}$, thereby reducing the heat flux between cells and providing a larger channel for convective fluid flow to carry heat away, further protecting the neighboring lithium-ion battery from reaching critical temperatures.
Discussion and Implications for System Design
The experimental evidence strongly supports immersion cooling as a highly effective strategy for mitigating thermal runaway propagation in high-capacity lithium-ion battery energy storage modules. The transition from a propagating failure in air to a contained single-cell event in fluid represents a quantum leap in intrinsic safety. The mechanism is multifaceted:
- Enhanced Heat Extraction: The fluid’s high heat capacity and the excellent thermal contact provide immediate and massive heat sinking, capping the temperature rise of the failing cell.
- Atmosphere Inerting: The oxygen-free environment surrounding the submerged cell eliminates the possibility of sustained combustion of ejected materials, preventing fire.
- Dilution and Cooling of Ejecta: Vent gases and particles are forced to bubble through the fluid, which cools them and scrubs them of reactive species, preventing them from directly impinging on and heating adjacent cells.
The findings offer clear guidance for the design and optimization of immersion-based thermal safety management systems (TMS) for lithium-ion battery packs:
- Fluid Selection: Dielectric fluids with higher specific heat capacity and thermal conductivity are preferred, as they delay thermal runaway onset and reduce peak temperatures more effectively. Trade-offs with viscosity (affecting pumping power) and long-term material compatibility must be considered.
- Module Layout: Incorporating adequate spacing between cells, even in an immersed environment, is a simple yet powerful design lever to further decouple cells thermally and enhance fluid flow, providing an additional safety margin against propagation.
- System Integration: The immersion chamber must be engineered with robust pressure management (e.g., burst disks, vent paths) to handle gas generation. The external cooling loop (pump, heat exchanger) must be sized to reject the total heat load, including the occasional large thermal pulse from a failing cell, to prevent saturation of the fluid’s cooling capacity.
From a broader perspective, this work contributes to the fundamental understanding of failure containment in electrochemical energy storage. It validates that by drastically increasing the cooling efficiency term $\eta_{cool}$ in the propagation risk function, the hazard can be managed. For gigawatt-hour-scale grid storage, where thousands of high-capacity lithium-ion battery cells are colocated, such intrinsic containment strategies are not merely beneficial but may be essential for achieving acceptable risk levels.
Conclusion
This comprehensive experimental study investigated the thermal runaway behavior of a high-capacity lithium iron phosphate lithium-ion battery module under an immersion cooling regime and compared it directly with conventional non-immersion conditions. The key conclusions are:
- Immersion cooling fundamentally contains thermal runaway. While the abuse-triggered cell still undergoes failure, propagation to adjacent cells is prevented entirely. This is in stark contrast to the non-immersed module, which exhibited severe thermal runaway propagation from the initiating cell to its neighbor.
- Immersion cooling drastically mitigates the severity of the single-cell event. The peak temperature of the failing lithium-ion battery was reduced by approximately 85% compared to failure in air, with no associated fire or explosion observed within the immersion chamber.
- The thermophysical properties of the immersion fluid are critical. A fluid with higher specific heat capacity and thermal conductivity (Fluid B) was shown to extend the thermal runaway gestation period by 150 seconds and further reduce peak temperatures by 12-17°C compared to a fluid with lower properties (Fluid A).
- Inter-cell spacing remains a relevant design parameter within an immersed module. Increasing the gap from 1.5 mm to 2.0 mm significantly reduced the peak temperatures experienced by the cells adjacent to the failing one by up to 40.3°C, enhancing the module’s resilience.
The results underscore that immersion cooling is not merely an efficient method for operational thermal management of lithium-ion battery packs but also a profoundly effective engineered safety barrier against catastrophic failure propagation. It addresses the core thermal challenges by maximizing heat transfer, providing a large thermal buffer, and creating an inert local environment. These findings provide valuable empirical data and design insights critical for advancing the safety and reliability of the next generation of large-scale lithium-ion battery energy storage systems.