The rapid integration of renewable energy sources like solar and wind into power grids has created a critical demand for large-scale energy storage solutions to address intermittency and balance grid loads. Among various electrochemical technologies, the lithium-ion battery stands as the predominant choice for modern energy storage systems (ESS) due to its superior energy density, long cycle life, and relatively low environmental footprint. These systems are often deployed in densely packed configurations within standardized containers or cabins to form megawatt-scale storage facilities. However, the inherent chemical instability of lithium-ion batteries presents a significant safety challenge. Under conditions of mechanical, electrical, or thermal abuse, a single lithium-ion battery cell can undergo a violent exothermic process known as thermal runaway. This event is characterized by the rapid release of heat, flammable gases, and toxic vapors. Within the confined space of an energy storage cabin, the heat from one failing lithium-ion battery module can readily propagate to adjacent modules, triggering a cascading failure that can escalate into a catastrophic fire or explosion. Therefore, developing and optimizing effective fire suppression systems specifically for lithium-ion battery energy storage cabins is paramount for ensuring the safe deployment of this critical technology.
Research into the hazards of lithium-ion battery thermal runaway and propagation is extensive. Studies have documented that a failing lithium-ion battery can eject flames and release substantial heat and combustible gas mixtures. Within a battery pack, thermal runaway can propagate swiftly, generating temperatures and pressures high enough to deform and melt module housing. Consequently, identifying suppression agents capable of halting this process is a key research focus. Various agents have been investigated, including water mist, gaseous agents like perfluorohexanone, aerogels, and liquid nitrogen. While water mist cools effectively, and gaseous agents inhibit flame chemically, liquid nitrogen offers a unique combination of mechanisms: extreme cooling due to its low boiling point (-196°C) and high latent heat of vaporization, coupled with oxygen dilution as it expands approximately 694 times in volume upon vaporization. Prior experimental work has shown that liquid nitrogen can significantly inhibit thermal runaway propagation in lithium-ion batteries. However, most research has been confined to single cell or small module levels. There remains a significant knowledge gap regarding full-scale fire dynamics and suppression efficacy within an actual energy storage cabin environment. This study employs advanced numerical simulation to bridge this gap, systematically evaluating the performance of liquid nitrogen fire suppression systems in a standard 20-foot lithium-ion battery energy storage cabin and comparing it with conventional water mist systems.
Numerical Modeling Methodology
This investigation utilizes Fire Dynamics Simulator (FDS), a Computational Fluid Dynamics (CFD) code developed by the National Institute of Standards and Technology (NIST), which is widely validated for simulating fire-driven fluid flow, heat transfer, and combustion. The effectiveness of FDS in modeling lithium-ion battery fires has been established in previous literature, making it a suitable tool for this cabin-scale study.
1.1 Geometry and Scenario
A 1:1 scale model of a standard 20-foot (6.096 m long) energy storage cabin was constructed. The internal dimensions are 6.00 m (L) × 2.50 m (W) × 2.30 m (H). Two symmetrical battery racks are placed along the longer walls, with a 1.2 m aisle between them. Each rack holds multiple lithium-ion battery modules. To simulate an internal short-circuit leading to thermal runaway, an ignition source is placed at the mid-height of a module on one rack. Fire suppression nozzles are installed 0.1 m below the cabin ceiling. Their activation temperature is set at 60°C, a common value in fire protection engineering. Key agent parameters, informed by standards and prior research, are summarized in Table 1.
| Parameter | Value for Liquid Nitrogen | Value for Water Mist |
|---|---|---|
| Nozzle Flow Rate | 0.8 kg/s | 0.8 kg/s |
| Droplet Diameter (SMD) | 500 μm | 500 μm |
| Initial Temperature | -196 °C | 20 °C |
| Activation Temperature | 60 °C | 60 °C |
1.2 Model Validation
The fidelity of the simulation hinges on accurately representing the fire source. A simplified model for a lithium-ion battery module was developed based on validated single-cell models. A single lithium-ion battery was modeled to undergo thermal runaway at 120°C with a peak Heat Release Rate (HRR) of 25 kW. A battery module was then constructed from four such cells, accounting for interstitial fixtures. The simulated HRR curve for a single lithium-ion battery showed excellent agreement with experimental data in terms of rise rate and peak. For the module, the peak HRR reached approximately 60 kW, consistent with experimental findings. Furthermore, the extinguishing sub-models were validated against separate experimental data for both liquid nitrogen and water mist application on a single lithium-ion battery fire. The simulated temperature decay trends matched the experimental results closely, confirming that liquid nitrogen induces a more rapid temperature drop compared to water mist. These multi-level validations confirm the reliability of the modeling approach for this study.
1.3 Mesh Sensitivity Analysis
CFD simulation accuracy is critically dependent on grid resolution. An overly coarse mesh introduces significant errors, while an overly fine mesh makes computation prohibitively expensive. The characteristic fire diameter $D^*$ provides a guideline for mesh size:
$$D^* = \left( \frac{\dot{Q}}{\rho_{\infty} c_p T_{\infty} \sqrt{g}} \right)^{2/5}$$
where $\dot{Q}$ is the heat release rate (kW), $\rho_{\infty}$ is the ambient air density (1.2 kg/m³), $c_p$ is the specific heat of air (1.0 kJ/(kg·K)), $T_{\infty}$ is the ambient temperature (293 K), and $g$ is gravitational acceleration (9.8 m/s²). Considering the calculated $D^*$ and the size of the smallest obstructions (battery modules), four mesh sizes (0.15 m, 0.12 m, 0.10 m, 0.08 m) were tested. The temperature evolution at a fixed point inside the cabin was compared across these meshes. The results indicated that for mesh sizes of 0.12 m and smaller, the temperature profiles converged. Therefore, a mesh size of 0.10 m was selected for all subsequent simulations to ensure an optimal balance between computational accuracy and cost.
1.4 Simulation Matrix
A comprehensive set of simulation cases was designed to isolate the effects of key variables on fire suppression performance. The primary comparisons were between liquid nitrogen (LN2) and water mist (WM) under both normal atmospheric pressure (100 kPa) and a reduced-pressure environment (65 kPa), the latter representing high-altitude installations. Furthermore, the influence of liquid nitrogen system design was probed by varying the number of nozzles (1, 3, and 5) and the spray angle of the nozzles (15°–30°, 30°–45°, 45°–60°, 60°–75°). A control case with no suppression system was also included. The full simulation matrix is presented in Table 2.
| Case ID | Suppressant | Pressure (kPa) | Number of Nozzles | Spray Angle | Purpose |
|---|---|---|---|---|---|
| Ctrl | None | 100 / 65 | 0 | – | Baseline fire development |
| LN2-1 | Liquid Nitrogen | 100 | 1 | 60°–75° | Effect of nozzle count |
| LN2-3 | Liquid Nitrogen | 100 | 3 | 60°–75° | Effect of nozzle count |
| LN2-5-A | Liquid Nitrogen | 100 | 5 | 60°–75° | Baseline for angle study; Comparison with WM |
| LN2-5-B | Liquid Nitrogen | 100 | 5 | 45°–60° | Effect of spray angle |
| LN2-5-C | Liquid Nitrogen | 100 | 5 | 30°–45° | Effect of spray angle |
| LN2-5-D | Liquid Nitrogen | 100 | 5 | 15°–30° | Effect of spray angle |
| WM-100 | Water Mist | 100 | 5 | 60°–75° | Comparison with LN2 at 100 kPa |
| WM-65 | Water Mist | 65 | 5 | 60°–75° | Comparison with LN2 at 65 kPa |
| LN2-65 | Liquid Nitrogen | 65 | 5 | 60°–75° | Comparison with WM at 65 kPa |
Fire Suppression Performance and Fire Dynamics Analysis
2.1 Comparative Efficacy: Liquid Nitrogen vs. Water Mist
The effectiveness of the suppression systems was first evaluated by comparing the total Heat Release Rate (HRR) over time, as shown in the aggregated results. In the uncontrolled scenarios (Ctrl), the fire developed fully, with peak HRR reaching approximately 9 MW at 100 kPa and about 6 MW at 65 kPa. The lower peak in the reduced-pressure environment is attributed to oxygen-limited combustion, which stifles the burning rate of the failing lithium-ion battery modules.
When suppression systems were activated, the HRR profiles changed dramatically. Both liquid nitrogen and water mist successfully prevented full-scale conflagration, reducing the peak HRR to around 350 kW—a reduction of over 95%. However, nuanced differences were evident. Under normal pressure, liquid nitrogen achieved a marginally lower peak HRR than water mist. More significantly, after reaching its peak, the HRR in the liquid nitrogen case decayed to zero more rapidly. This superior performance is due to the combined physical mechanisms of liquid nitrogen: its profound cooling effect directly extracts heat from the hot lithium-ion battery surfaces and the surrounding gases, while the massive volume of inert nitrogen gas released upon vaporization displaces oxygen, creating a local inert atmosphere that suppresses flame chemistry.
This advantage became more pronounced in the 65 kPa environment. The lower ambient pressure enhances the vaporization rate and expansion of liquid nitrogen, leading to faster oxygen dilution and more effective smothering of the fire. The HRR curve for liquid nitrogen at low pressure showed an even quicker decay post-peak compared to water mist at the same pressure. The analysis of temperature slices taken at 100 seconds after ignition (a time when fires were typically brought under control) visually corroborates these findings. In uncontrolled fires, large regions exceeding 200°C were present, indicating widespread thermal runaway propagation among lithium-ion battery modules. With suppression, these high-temperature zones were contained. Notably, in the low-pressure liquid nitrogen case, the cabin temperature was reduced to near-ambient levels more uniformly and quickly than with water mist.
The maximum gas temperature recorded inside the cabin during each simulation provides a single-point metric for severity. The results are summarized in Table 3. The data clearly shows that while both agents drastically reduce the maximum temperature compared to the uncontrolled fire, liquid nitrogen consistently achieves a lower peak temperature than water mist under identical pressure and nozzle configuration conditions. This confirms its superior cooling and inerting capability for lithium-ion battery cabin fires.
| Case Description | Pressure (kPa) | Suppressant | Max Cabin Temp. (°C) |
|---|---|---|---|
| Uncontrolled Fire | 100 | None | ~989 |
| Uncontrolled Fire | 65 | None | ~960 (est.) |
| 5 Nozzles, Standard Angle | 100 | Water Mist | ~620 |
| 5 Nozzles, Standard Angle | 100 | Liquid Nitrogen | ~597 |
| 5 Nozzles, Standard Angle | 65 | Water Mist | ~580 |
| 5 Nozzles, Standard Angle | 65 | Liquid Nitrogen | < 500 |
2.2 Influence of Nozzle Count on Liquid Nitrogen Performance
Having established the comparative advantage of liquid nitrogen, we investigated the optimization of its delivery system. The number of activated nozzles significantly impacts fire outcomes. With only a single nozzle (Case LN2-1), the fire was mitigated but not contained. The HRR curve showed a reduced peak of about 7 MW, indicating that the fire eventually involved most lithium-ion battery modules in the cabin, though with lower intensity than the uncontrolled case. The temperature slice at 100s showed a tall, localized high-temperature plume.
Increasing the nozzle count to three (LN2-3) and five (LN2-5-A) proved to be a critical threshold for successful containment. Both configurations prevented cascading failure, limiting the HRR peak to 420 kW and 360 kW, respectively. The temperature slices confirmed effective containment, with the 5-nozzle arrangement producing a more uniformly cooled environment. The relationship between nozzle count and maximum cabin temperature is clearly negative, as shown in Table 4. More nozzles deliver a greater total mass flow rate of liquid nitrogen per unit time. This enhances the total heat extraction capacity and increases the rate of nitrogen gas generation, leading to faster oxygen depletion and more effective suppression of the lithium-ion battery fire.
| Number of Nozzles | Peak HRR (kW) | Maximum Cabin Temp. (°C) | Containment Achieved? |
|---|---|---|---|
| 0 (Control) | ~9000 | ~989 | No |
| 1 | ~7000 | ~877 | No |
| 3 | ~420 | ~669 | Yes |
| 5 | ~360 | ~597 | Yes |
2.3 Influence of Nozzle Spray Angle on Liquid Nitrogen Performance
With five nozzles ensuring containment, the effect of spray angle was examined. The HRR profiles for different angles revealed a consistent trend: a smaller spray angle (a narrower, more focused spray cone) resulted in better fire suppression. The peak HRR decreased from approximately 350 kW for the 60°–75° angle to about 150 kW for the 15°–30° angle—a reduction of nearly 57%. Furthermore, the characteristic “rebound” in HRR observed with wider angles was less pronounced or absent with narrower angles. This rebound is attributed to the counter-flow between buoyant, hot combustion products rising from the lithium-ion battery fire and the descending cold nitrogen gas/ droplets. With wide angles, a significant portion of the agent is sprayed onto unignited modules above the fire, allowing time for vaporization and reducing the momentum and liquid fraction reaching the core fire zone.
A narrower spray angle creates a more targeted jet, penetrating the buoyant plume more effectively and delivering a higher concentration of liquid phase agent directly onto the surfaces of the thermally runaway lithium-ion battery modules. This leads to more efficient direct cooling. Temperature slices at 100 seconds vividly illustrate this: with the narrowest angle, the cabin temperature was reduced to sub-zero levels in many areas, indicating a massive and direct application of cryogenic liquid. The quantitative relationship is summarized in Table 5.
| Spray Angle Range | Approx. Peak HRR (kW) | Maximum Cabin Temp. (°C) | Observation |
|---|---|---|---|
| 60° – 75° | 350 | ~597 | Fire contained, noticeable HRR rebound |
| 45° – 60° | 280 | ~520 | Reduced rebound, faster decay |
| 30° – 45° | 200 | ~450 | Minimal rebound, rapid suppression |
| 15° – 30° | 150 | < 400 | Fastest suppression, extensive sub-zero zones |
It is crucial to note that this finding is specific to the fire ignition location at the mid-height of the rack. For fires originating at the top or bottom of a rack, the optimal spray angle for targeting the flames might differ. This highlights the importance of considering fire scenario variability in system design.
Conclusions and Implications for Safety Design
This comprehensive numerical simulation study provides critical insights into the fire dynamics of lithium-ion battery energy storage cabins and the performance optimization of liquid nitrogen suppression systems. The key conclusions are as follows:
- Superior Efficacy of Liquid Nitrogen: For fires involving lithium-ion battery modules in a confined cabin, liquid nitrogen demonstrates superior suppression performance compared to water mist under both normal and reduced-pressure conditions. Its dual mechanism of extreme cooling and rapid oxygen dilution is more effective at interrupting the chain reactions involved in lithium-ion battery thermal runaway and propagation.
- Critical Role of Nozzle Density: The number of nozzles is a decisive factor. A single nozzle is insufficient to contain a propagating lithium-ion battery fire, only reducing its severity. A minimum of three nozzles in the modeled configuration is necessary to achieve containment, and five nozzles provide enhanced cooling and faster suppression. The relationship between nozzle count and key hazard parameters (Peak HRR, Max Temp) can be expressed conceptually as a negative correlation, emphasizing the need for adequate system coverage.
- Optimization via Spray Angle: For a given nozzle count and a fire located at mid-rack height, a narrower spray angle significantly improves suppression effectiveness. Focused application delivers more liquid-phase agent directly to the fire source, enhancing direct cooling and minimizing wasteful vaporization in the upper cabin space. The reduction in peak HRR with decreasing spray angle can be substantial, highlighting spray angle as a key design variable.
The findings offer direct guidance for engineering the fire protection of lithium-ion battery energy storage systems. Designers should consider liquid nitrogen as a highly effective suppression agent, particularly for facilities in low-pressure environments or where rapid inertion is critical. The system must be designed with sufficient nozzle density to ensure the fire is blanketed and contained. Furthermore, the spray pattern should be engineered to ensure targeted delivery to potential fire locations, which may involve using nozzles with different angles or staged activation protocols based on thermal detection zones.
Future work should explore the synergistic effects of combined agents (e.g., liquid nitrogen with water mist or additives) to potentially enhance efficiency and reduce the total agent requirement. Additionally, economic and environmental lifecycle analyses comparing liquid nitrogen systems with other solutions are necessary for holistic decision-making. Finally, experimental validation at this cabin scale remains the essential next step to confirm these numerical findings and refine the design guidelines for safeguarding the growing infrastructure of grid-scale lithium-ion battery energy storage.