In the context of the global transition to renewable energy, battery energy storage systems (BESS) play an increasingly critical role in energy storage and distribution. Electrochemical energy storage systems, particularly lithium-ion battery-based BESS, have become essential for achieving power balance and ensuring grid stability due to their rapid response and flexible energy supply capabilities. By the end of 2023, the installed capacity of global power storage projects reached 289.2 GW, with lithium-ion batteries dominating the market, especially in new energy storage applications, where growth rates exceeded 100%. In China, the installed capacity of BESS has surpassed 86.5 GW, with an annual growth rate of 45%. Lithium-ion batteries are favored for their high energy density, long lifespan, and relatively low cost. However, battery performance is highly sensitive to temperature variations, requiring operation within a specific range to ensure optimal performance and safety. During operation, electrochemical reactions during charging and discharging generate significant heat. If not effectively dissipated, this heat can lead to temperature non-uniformity, localized hot spots, increased internal resistance, reduced capacity, and even thermal runaway. Thus, thermal management in BESS is paramount to mitigate these risks.
Battery thermal management systems (BTMS) are crucial for maintaining optimal operating temperatures in BESS and electric vehicles (EVs). Effective thermal management not only enhances battery performance but also extends lifespan. Research has focused on evaluating various cooling strategies, including air cooling, liquid cooling, and phase change materials (PCM). Liquid cooling systems, with their superior heat conduction efficiency, are widely used in EVs for compact battery packs. However, their complexity, high operational costs, and maintenance challenges limit their application in stationary BESS. PCM-based BTMS stabilize temperatures through latent heat absorption and release but suffer from low thermal diffusivity, making them suitable for smaller-scale battery packs. Air cooling systems, due to their simplicity and low cost, are commonly employed in BESS. However, as BESS scales up and operates at high power densities and charge-discharge rates,散热 issues become more pronounced. Optimizing air-cooled thermal management to improve efficiency is a key research focus.
The core of air-cooled thermal management in BESS lies in optimizing airflow organization. Factors such as duct design, inlet and outlet layouts directly impact thermal efficiency. Previous studies have proposed structures like “main ducts + risers” for precise airflow distribution to battery racks, with added baffles to improve flow. Others have optimized airflow distribution in data center-inspired BESS or introduced internal fin structures in battery packs to enhance temperature uniformity. J-type cooling channels combining U and Z-type designs have also reduced battery pack temperatures. These approaches demonstrate that air cooling optimization can effectively address thermal issues in BESS. However, most research focuses on small-scale or specific battery pack structures, with limited attention to container-level large-scale BESS. This study addresses this gap by developing a three-dimensional CFD model for a container-level BESS, investigating the impact of cold aisle structures, air supply modes, and outlet layouts on thermal management efficiency. Through CFD simulations, we aim to optimize airflow paths, enhance thermal management capabilities, and ensure safe and efficient operation of container-level BESS.
We modeled a container-level BESS with dimensions of 4000 mm × 2440 mm × 2590 mm. The cold aisle, located between two rows of battery cabinets, measures 2650 mm × 1180 mm × 1960 mm. When the top and sides of the cold aisle are enclosed, it is referred to as a closed cold aisle; otherwise, it is open. Each battery cabinet has dimensions of 265 mm × 340 mm × 1960 mm, with fan inlets of 250 mm × 150 mm per layer. Twelve batteries are centrally placed in each cabinet with 5 mm gaps between them. Four column-type air conditioners are evenly distributed between the two rows of cabinets. The batteries are commercial prismatic lithium iron phosphate (LiFePO4) cells, model LF50K, with a rated capacity of 50 Ah and voltage of 3.2 V. Key parameters are listed in Table 1.
| Parameter | Value |
|---|---|
| Thermal Conductivity λx (W/m·K) | 1.0 |
| Thermal Conductivity λy (W/m·K) | 10.0 |
| Thermal Conductivity λz (W/m·K) | 21.0 |
| Specific Heat Capacity Cp (J/kg·K) | 1399 (body), 871 (anode), 381 (cathode) |
| Rated Voltage (V) | 3.2 |
| Rated Capacity (Ah) | 50 |
| Density ρb (kg/m³) | 1934 (body), 2719 (anode), 8978 (cathode) |
| Internal Resistance (mΩ) | ≤1.5 |
| Anode Material | Graphite |
| Cathode Material | LiNiO2 + LiCoO2 + Li2MnO2 |
| Height (mm) | 185 |
| Width (mm) | 135 |
| Thickness (mm) | 30 |
| Terminal Diameter (mm) | 12 |
| Terminal Height (mm) | 5 |
To reduce computational load, the BESS model was simplified to a symmetric domain along the Z-direction. The cooling air enters through five inlets (each 530 mm × 530 mm) and exits via an outlet (1180 mm × 400 mm). The symmetric plane is indicated in the model. During operation, cooling air enters the cold aisle and is drawn by fans into the battery cabinets; heated air is expelled through the outlet.
We measured battery surface temperatures using K-type thermocouples at points T1, T2, and T3 on a fully charged battery placed in a blast-proof incubator at 25°C. Discharge tests at 1C, 2C, and 3C rates (where C = discharge current / rated capacity) showed average temperature rises of 3.83°C, 8.47°C, and 12.47°C, respectively. The volumetric heat generation rates were calculated as 2819.9 W/m³, 14658.4 W/m³, and 37959.2 W/m³ using the formula:
$$ q = \frac{m C_p \Delta T}{t \cdot V} $$
where \( q \) is the volumetric heat source (W/m³), \( C_p \) is the specific heat capacity (J/kg·K), \( m \) is the battery mass (kg), \( \Delta T \) is the temperature difference (°C), \( t \) is the discharge time (s), and \( V \) is the battery volume (m³). For this study, we used the heat generation at 3C discharge (37959.2 W/m³) as the volumetric heat source for each battery to analyze optimization schemes under extreme conditions.
The heat transfer in the battery module was modeled using an anisotropic lumped capacitance approach. The prismatic battery cells were simplified as rectangular blocks, neglecting busbars and power modules. The governing equation for battery heat transfer is:
$$ \rho_b C_p \frac{\partial T}{\partial t} = \frac{\partial}{\partial x} \left( \lambda_x \frac{\partial T}{\partial x} \right) + \frac{\partial}{\partial y} \left( \lambda_y \frac{\partial T}{\partial y} \right) + \frac{\partial}{\partial z} \left( \lambda_z \frac{\partial T}{\partial z} \right) + q $$
where \( \rho_b \) is battery density, \( C_p \) is specific heat capacity, \( T \) is temperature, \( t \) is time, \( \lambda_x, \lambda_y, \lambda_z \) are thermal conductivities in x, y, z directions, and \( q \) is the volumetric heat source.
The transient three-dimensional Navier-Stokes equations were employed to simulate heat transfer dynamics. The continuity, momentum, and energy equations are:
$$ \frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{u}) = 0 $$
$$ \frac{\partial}{\partial t} (\rho \mathbf{u}) + \nabla \cdot (\rho \mathbf{u} \mathbf{u}) = -\nabla P + \rho \mathbf{g} $$
$$ \frac{\partial T}{\partial t} + \nabla \cdot (\mathbf{u} T) = \frac{\lambda}{\rho C} \nabla^2 T $$
where \( \rho \) is fluid density, \( \mathbf{u} \) is velocity vector, \( P \) is pressure, \( \mathbf{g} \) is gravitational acceleration, \( \lambda \) is fluid thermal conductivity, and \( C \) is fluid specific heat capacity.
The k-ε turbulence model was selected, with diffusion terms discretized using a second-order upwind scheme. Gravity was applied along the y-axis at 9.8 m/s². The outlet pressure was set to atmospheric, and no-slip conditions were assumed at walls. Initial temperatures for the fluid, battery modules, and environment were 25°C. Air properties are listed in Table 2.
| Medium | Density (kg/m³) | Specific Heat (J/kg·K) | Thermal Conductivity (W/m·K) | Viscosity (kg/m·s) | Thermal Expansion Coefficient (K⁻¹) |
|---|---|---|---|---|---|
| Air | 1.2 | 1006.4 | 0.024 | 1.8×10⁻⁵ | 3.4×10⁻³ |
Model validation was performed using experimental data from a data center散热 system, given the geometric similarities between chip servers and BESS in terms of heat generation and airflow distribution. The CFD model was configured to match the geometry and boundary conditions of the reference study. Comparisons of temperature differences between cabinet inlets/outlets and supply air showed good agreement, confirming model reliability for BESS applications.
We investigated the impact of cold aisle structure on battery temperature. The cold aisle is the space between two rows of battery cabinets, equal in height to the cabinets. In the closed configuration, the top and sides are sealed, restricting heat exchange to the cold aisle air. In the open configuration, the aisle is unenclosed. Table 3 compares airflow velocity and temperature in different cross-sections for open and closed cold aisles.
| Cold Aisle Structure | Plane Position (mm) | Max Velocity (m/s) | Max Temperature (°C) | Max Temperature Difference (°C) |
|---|---|---|---|---|
| Open | x = 1820 | 5.8 | 39.8 | 14.8 |
| Closed | x = 1820 | 6.4 | 39.5 | 14.5 |
| Open | y = 1000 | 6.8 | 33.4 | 8.4 |
| Closed | y = 1000 | 10.0 | 29.6 | 4.6 |
At the same locations, the closed structure exhibits higher maximum airflow velocities and lower temperatures than the open structure. This results in better cooling for batteries in the closed configuration. Table 4 summarizes battery surface temperatures for both structures.
| Cold Aisle Structure | Max Battery Surface Temperature (°C) | Max Temperature Difference on Single Cell (°C) | Average Battery Surface Temperature (°C) |
|---|---|---|---|
| Open | 44.6 | 6.1 | 29.4 |
| Closed | 43.7 | 5.2 | 28.6 |
The closed cold aisle reduces the maximum battery surface temperature by 0.9°C, decreases the single-cell temperature difference by 15% (0.9°C), and lowers the average surface temperature by 0.8°C, improving heat transfer efficiency by 2.7%. Velocity vector plots show vortices at the top of the cold aisle and near the outlet in both structures, but the closed design prevents mixing of cold and hot air, reducing energy loss and ensuring more uniform inlet temperatures for battery cabinets. Temperature contours confirm that the closed structure minimizes pre-cooling air leakage and enhances散热.
Next, we examined the effect of air supply location on battery temperature, focusing on underfloor supply versus column-type air conditioner supply. In the underfloor mode, cooling air enters from below the floor, while in the column-type mode, air is supplied horizontally from units between cabinets. For the column-type supply, the underfloor inlets are set as adiabatic walls, and the air conditioner faces are velocity inlets. Other boundary conditions remain consistent with the closed cold aisle model.
Table 5 compares airflow velocity and temperature at the y = 1000 mm plane for both supply modes.
| Air Supply Mode | Max Velocity (m/s) | Max Temperature (°C) | Max Temperature Difference (°C) |
|---|---|---|---|
| Column-Type | 10.6 | 30.6 | 5.6 |
| Underfloor | 10.0 | 29.6 | 4.6 |
The column-type supply has a higher airflow velocity but also a higher outlet temperature. Battery surface temperatures are detailed in Table 6.
| Air Supply Mode | Max Battery Surface Temperature (°C) | Max Temperature Difference on Single Cell (°C) | Average Battery Surface Temperature (°C) |
|---|---|---|---|
| Column-Type | 44.1 | 5.6 | 28.7 |
| Underfloor | 43.7 | 5.2 | 28.6 |
The underfloor supply reduces the maximum battery temperature by 0.4°C and the single-cell temperature difference by 7.1% (0.4°C) compared to the column-type supply. The inferior performance of the column-type mode is attributed to vortex formation near the air inlets of cabinets adjacent to the air conditioners, which impedes cold airflow into some cabinets. This results in reduced cooling for affected cabinets and increased temperature non-uniformity. In contrast, underfloor supply ensures more consistent airflow distribution, leading to better thermal management in the BESS.
We further explored the impact of air outlet location on battery temperature. Three outlet positions were considered: Outlet I and II on the left and right sides (aligned with the cold aisle top), and Outlet III on the back (centered between six cabinets in the x-direction, at the same height as cabinet tops). All outlets have the same area.
Table 7 presents airflow velocity and temperature at the x = 1820 mm plane for different outlet locations.
| Outlet Location | Max Velocity (m/s) | Max Temperature (°C) | Max Temperature Difference (°C) |
|---|---|---|---|
| I (Left Side) | 6.4 | 39.5 | 14.5 |
| II (Right Side) | 6.5 | 39.6 | 14.6 |
| III (Back) | 23.8 | 39.1 | 14.6 |
Outlets I and II show similar patterns due to their symmetric positions, while Outlet III exhibits distinct characteristics. With Outlet III, the airflow path is shorter, reducing resistance and enhancing convective heat transfer, which lowers the back-side temperature of batteries. However, increased air density along the y-axis accelerates upward movement, potentially reducing cold air drawn by bottom fans and causing insufficient cooling at the base. This can be mitigated by adjusting inlet air velocity. Outlet III achieves the best cooling performance, with the lowest maximum temperature. Battery surface temperatures are summarized in Table 8.
| Outlet Location | Max Battery Surface Temperature (°C) | Max Temperature Difference on Single Cell (°C) | Average Battery Surface Temperature (°C) |
|---|---|---|---|
| I (Left Side) | 43.7 | 5.2 | 28.6 |
| II (Right Side) | 44.0 | 5.1 | 28.2 |
| III (Back) | 42.9 | 4.9 | 28.3 |
Outlet III yields the smallest maximum temperature difference on a single cell (4.9°C), which is 4.1% and 3.2% lower than Outlets I and II, respectively. It also has the lowest maximum battery surface temperature (42.9°C). Thus, placing the outlet at the back optimizes散热 efficiency in the BESS.
In conclusion, this study demonstrates the significance of cold aisle structure, air supply mode, and outlet location in thermal management of container-level BESS. The closed cold aisle design outperforms the open structure by minimizing air mixing and improving inlet temperature uniformity. Underfloor air supply is superior to column-type supply, as it avoids vortices and ensures consistent cooling. Back-side outlet placement enhances heat dissipation by shortening airflow paths and reducing resistance. These optimizations collectively improve the thermal performance and safety of battery energy storage systems, providing valuable insights for large-scale BESS design. Future work could explore dynamic control strategies and integration with renewable energy sources to further advance BESS capabilities.