In the pursuit of a sustainable energy future, the global demand for green electricity has accelerated the deployment of large-scale industrial and commercial battery energy storage systems. These systems are pivotal for integrating renewable sources like solar and wind, enabling energy storage and dispatch. However, the core component—lithium-ion batteries—faces significant thermal challenges. Poor thermal stability, exacerbated side reactions at elevated temperatures, and the risk of thermal runaway pose serious threats to the safety, efficiency, and longevity of battery energy storage systems. Effective thermal management is thus not a luxury but a necessity to ensure these systems operate within the optimal temperature window of 20–35°C and maintain minimal temperature variations among cells. In this study, I employ a numerical simulation approach, calibrated with experimental data, to explore three critical factors influencing the thermal behavior of a battery pack within a battery energy storage system. The investigation focuses on the impact of different aspect ratio flow channel designs in liquid cooling plates, various layouts of these cooling plates, and the effect of inlet coolant flow rates. The overarching goal is to enhance cooling performance, improve temperature uniformity, and reduce the energy consumption of the thermal management system itself, thereby contributing to more reliable and efficient battery energy storage systems.
The importance of thermal management in battery energy storage systems cannot be overstated. Lithium-ion batteries generate heat during charge and discharge cycles due to irreversible ohmic losses and reversible entropic changes. If this heat is not adequately dissipated, it can lead to localized hot spots, accelerated aging, capacity fade, and in worst-case scenarios, thermal runaway—a cascading failure that can result in fire or explosion. For large-scale battery energy storage systems comprising thousands of cells, ensuring thermal homogeneity is as crucial as controlling the maximum temperature. A temperature differential exceeding 5°C between cells can lead to unbalanced aging and reduced overall system capacity. Traditional air-cooling methods often fall short for high-density battery energy storage systems due to the low heat capacity and thermal conductivity of air. Liquid cooling, particularly indirect cooling using cold plates, has become the industry standard for its superior heat transfer capability. However, design optimization is essential to maximize cooling efficiency while minimizing parasitic power consumption from pumps. This study delves into specific design parameters to push the boundaries of liquid cooling performance for battery energy storage systems.
My investigation is centered on a representative battery pack module, a fundamental building block of a larger battery energy storage system. The module configuration is 1P52S, meaning one cell in parallel and fifty-two in series. Each cell is a large-format semi-solid lithium-ion battery with a nominal capacity of 280 Ah. The physical dimensions and key properties are summarized below. The pack includes cells, insulating polycarbonate (PC) boards, thermal interface materials (TIM) like silica gel, and the liquid cooling plates. The cooling plate design features a serpentine flow channel with multiple parallel paths. The primary material properties used in the simulation are critical for accurate thermal modeling and are listed in a comprehensive table.
| Parameter | Value |
|---|---|
| Nominal Capacity | 280 Ah |
| Mass | 5.5 kg |
| Energy Density | ≥ 165 Wh/kg |
| DC Internal Resistance (25°C, 25% SOC) | 0.50 mΩ |
| Dimensions (L × W × H) | 173.8 mm × 71.6 mm × 207.2 mm |
| Charge Cut-off Voltage | 3.65 V |
| Discharge Cut-off Voltage | 2.5 V |
| Component | Density, ρ (kg/m³) | Specific Heat, c_p (J/(kg·K)) | Thermal Conductivity (W/(m·K)) |
|---|---|---|---|
| Cell (Anisotropic) | 2152 | 1051.1 | k_x = 1.04, k_y,z = 21.05 |
| Liquid Cooling Plate (Aluminum) | 2680 | 880 | 237 |
| Thermal Silicone Pad (TIM) | 2420 | 967 | 2.1 |
| Coolant (Water-Ethylene Glycol Mix) | 1065 | 3394 | 0.419 |
| PC Insulation Board | 1200 | 1340 | 0.194 |
The numerical model is built upon the fundamental laws of thermodynamics and fluid dynamics. The simulation makes several reasonable assumptions to balance complexity and computational cost: contact resistances between components are considered negligible due to the use of TIM; material properties are treated as temperature-independent within the operating range; and radiation heat transfer is excluded as its contribution is minimal compared to conduction and convection. The governing equations form the backbone of the thermal-fluid simulation for the battery energy storage system.
The energy conservation equation for the battery cell, which includes its internal heat generation, is given by:
$$ \rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + \dot{Q} $$
where \( \rho \) is density, \( c_p \) is specific heat capacity, \( k \) is thermal conductivity (a tensor for anisotropic cells), \( T \) is temperature, \( t \) is time, and \( \dot{Q} \) is the volumetric heat generation rate within the cell.
The flow of coolant within the liquid cooling plates is governed by the incompressible Navier-Stokes equations and continuity equation:
$$ \frac{\partial}{\partial t} (\rho \vec{v}) + \nabla \cdot (\rho \vec{v} \vec{v}) = -\nabla p + \mu \nabla^2 \vec{v} + \vec{F} $$
$$ \nabla \cdot (\rho \vec{v}) = 0 $$
Here, \( \vec{v} \) is the fluid velocity vector, \( p \) is pressure, \( \mu \) is dynamic viscosity, and \( \vec{F} \) represents body forces such as gravity.
The heat generation within a lithium-ion cell, \( \dot{Q} \), is modeled using the Bernardi equation, which accounts for both irreversible (Joule) heating and reversible (entropic) heating:
$$ \dot{Q} = \frac{1}{V_b} \left( I^2 R + I T \frac{dE_{oc}}{dT} \right) $$
where \( V_b \) is the cell volume, \( I \) is the current (positive for discharge), \( R \) is the total internal resistance, \( T \) is the absolute cell temperature, and \( \frac{dE_{oc}}{dT} \) is the entropy coefficient, taken as 0.00049 V/K. The internal resistance \( R \) is not constant; it varies with the cell’s state of charge (SOC) and temperature. This dependency is crucial for accurate modeling of a battery energy storage system under dynamic loads. The resistance values are obtained from experimental DCR (Direct Current Resistance) mapping. The SOC is updated during the simulation as:
$$ \text{SOC}(t) = 1 – \frac{I t}{C_N} $$
with \( C_N \) being the nominal capacity. A representative dataset of internal resistance versus SOC at different temperatures is provided in the following table to illustrate this complexity.
| SOC (%) | -10°C | 0°C | 25°C | 45°C |
|---|---|---|---|---|
| 0 | — | — | — | — |
| 10 | 3.47 | 1.63 | 0.53 | 0.33 |
| 30 | 2.76 | 1.41 | 0.46 | 0.31 |
| 50 | 2.24 | 1.29 | 0.43 | 0.29 |
| 70 | 2.13 | 1.24 | 0.43 | 0.30 |
| 90 | 2.00 | 1.17 | 0.39 | 0.28 |
| 100 | 1.97 | 1.17 | 0.39 | 0.29 |
The computational domain encompasses the entire battery pack, the surrounding air, and the internal fluid domain of the cooling plates. Boundary conditions are applied to replicate realistic operating conditions for a battery energy storage system. The initial temperature for all components is set to the ambient temperature of 25°C. The external surfaces of the battery pack exchange heat with the environment via natural convection, described by Newton’s law of cooling:
$$ -k_{xt} \frac{\partial T_{xt}}{\partial n} = h (T_{xt} – T_{amb}) $$
where \( h \) is the convective heat transfer coefficient, set to 5 W/(m²·K), a typical value for natural convection. For the liquid cooling system, a mass flow inlet boundary condition is specified at the coolant entrance, with a base flow rate of 0.1791 kg/s (approximately 10 L/min for the water-glycol mix) and an inlet temperature of 22°C. The outlet is set to a pressure outlet condition at atmospheric pressure. Gravity is included in the model to account for natural convection effects in the air domain. The discharge process is simulated at a constant 0.5C rate (140A) until 90% of the capacity is depleted, which corresponds to a simulation time of 6480 seconds.
A critical step in ensuring the fidelity of any numerical model for a battery energy storage system is experimental validation. I conducted discharge tests on a single cell at different rates (0.5C and 1C) in a temperature-controlled chamber. The cell surface temperature was monitored and compared against simulation results. The maximum relative error between the simulated and experimental temperature curves was found to be less than 3%, which confirms the accuracy and reliability of the numerical model for subsequent pack-level analysis.
The energy consumption of the liquid cooling system, an important metric for the overall efficiency of the battery energy storage system, is calculated based on the pressure drop across the cooling plates:
$$ E_{pump} = \int_0^t \Delta p \cdot \dot{V}_{in} \, dt $$
where \( \Delta p \) is the total pressure drop between inlet and outlet, and \( \dot{V}_{in} \) is the volumetric flow rate at the inlet.
With the validated model in hand, I proceeded to systematically investigate the three key factors. The first factor involves the geometry of the flow channels within the liquid cooling plate. Three different channel cross-sections were studied, all with a fixed hydraulic characteristic but varying aspect ratios (width-to-height ratio, W/H). The cross-sectional area was maintained approximately constant to ensure a fair comparison of flow resistance, but the shape was varied with W/H = 7, 5, and 3 (height H = 6 mm). The results for cell temperature, system pressure drop, and cooling energy consumption are summarized below.
| Parameter | W/H = 7 | W/H = 5 | W/H = 3 |
|---|---|---|---|
| Max Cell Temperature at 6480s (°C) | 32.71 | 32.58 | 32.50 |
| Max Cell Temp. Rise from Ambient (°C) | 7.71 | 7.58 | 7.50 |
| Global Max. Temperature Difference in Pack (°C) | 8.85 | 8.80 | 8.75 |
| Steady-State Pressure Drop, Δp (kPa) | 12.8 | 9.5 | 6.9 |
| Pump Energy Consumption, E_pump (J) @6480s | ~138,000 | ~102,000 | ~74,000 |
The data reveals that the aspect ratio has a modest influence on the final cell temperature. The maximum temperature decreases slightly as W/H decreases, with W/H=3 offering only about a 0.2°C improvement over W/H=7. This is because the heat transfer performance is a trade-off between convection coefficient and wetted surface area. A lower aspect ratio (squarer channel) increases fluid velocity for the same flow rate, enhancing the convection coefficient \( h_{conv} \), which can be approximated by Dittus-Boelter type correlations for turbulent flow:
$$ Nu = \frac{h_{conv} D_h}{k_f} = C Re^{0.8} Pr^{0.4} $$
where \( Nu \) is Nusselt number, \( D_h \) is hydraulic diameter, \( k_f \) is fluid thermal conductivity, \( Re \) is Reynolds number, and \( Pr \) is Prandtl number. However, a lower W/H also reduces the contact surface area between coolant and plate. These opposing effects largely cancel out regarding net heat transfer. In contrast, the hydraulic performance is strongly affected. The pressure drop \( \Delta p \) is significantly lower for the W/H=3 design, approximately 46% less than the W/H=7 design. This leads to a proportional reduction in pump energy consumption, a vital consideration for the net efficiency of a battery energy storage system. Therefore, from a system-level perspective favoring lower parasitic loads, a lower aspect ratio channel (W/H=3) is preferable.
The second and potentially more impactful factor is the layout of the liquid cooling plates within the battery energy storage system module. The conventional approach employs cooling only at the bottom of the cell stack. I investigated an enhanced layout where, in addition to the bottom cold plate, two vertical cold plates are placed along the sides of the pack, contacting the cells at their mid-to-upper height region. This design aims to address the significant vertical temperature gradient inherent in bottom-only cooling. For a fair comparison, the total coolant flow rate was kept constant. In the combined layout, the flow was split equally between the bottom and side plates (5 L/min each, summing to 10 L/min total). The performance comparison is stark.
| Performance Metric | Bottom Cooling Only | Bottom + Two-Side Cooling | Improvement |
|---|---|---|---|
| Maximum Cell Temperature (°C) | 32.62 | 25.51 | ↓ 7.11°C (21.8%) |
| Maximum Cell Temperature Rise (°C) | 7.62 | 0.51 | ↓ 7.11°C |
| Global Max. Temp. Difference in Pack (°C) | 8.80 | 3.00 | ↓ 5.80°C (65.9%) |
| Average Heat Transfer Rate from Pack (W) | ~280 | ~520 | ↑ ~86% |
The results are transformative for battery energy storage system thermal management. The bottom+side cooling configuration drastically reduces the maximum cell temperature, keeping it barely above the inlet coolant temperature. More importantly, it dramatically improves temperature uniformity, slashing the global temperature difference from an unacceptable 8.8°C down to an excellent 3.0°C. This enhancement stems from the multi-directional heat extraction, which mitigates the thermal stratification along the cell height. The side plates effectively remove heat from the upper portions of the cells, which are farthest from the bottom cooler and traditionally the hottest zones. The total heat transfer power from the battery pack to the coolant nearly doubles, confirming the superior cooling capacity of this layout. This approach directly tackles one of the most persistent challenges in battery energy storage system design: achieving spatial temperature homogeneity.
The third factor analyzed is the inlet coolant flow rate for the bottom cooling plate (in the baseline configuration). Flow rate is a primary operational variable for any liquid-cooled battery energy storage system. I tested five different flow rates: 5.0, 7.5, 10.0, 12.5, and 15.0 L/min. The objective was to identify the point of diminishing returns, where increasing flow yields negligible thermal benefit but substantially increases pumping power.
| Flow Rate (L/min) | Max Cell Temp. (°C) | Max Temp. Rise (°C) | Global Max ΔT (°C) | Pressure Drop, Δp (kPa) | Pump Power (W) @ steady Δp |
|---|---|---|---|---|---|
| 5.0 | 34.85 | 9.85 | 2.80 | ~3.2 | ~0.27 |
| 7.5 | 33.25 | 8.25 | 2.09 | ~5.1 | ~0.64 |
| 10.0 | 32.62 | 7.62 | 1.75 | ~9.5 | ~1.58 |
| 12.5 | 32.25 | 7.25 | 1.58 | ~14.8 | ~3.08 |
| 15.0 | 32.05 | 7.05 | 1.41 | ~21.5 | ~5.38 |
The data shows that increasing the flow rate consistently lowers the maximum cell temperature and improves pack temperature uniformity. However, the rate of improvement decays. For instance, increasing flow from 10 to 12.5 L/min reduces the max temperature by 0.37°C, but a further increase to 15 L/min only yields an additional 0.20°C reduction. The temperature uniformity follows a similar trend. In contrast, the pressure drop rises approximately with the square of the flow rate, as suggested by the Darcy-Weisbach equation for turbulent flow:
$$ \Delta p = f \frac{L}{D_h} \frac{\rho v^2}{2} $$
where \( f \) is the friction factor, \( L \) is channel length, and \( v \) is fluid velocity. Consequently, the pump power, proportional to \( \Delta p \times \dot{V} \), increases sharply. The pump power at 15 L/min is roughly 20 times higher than at 5 L/min. Therefore, an optimal flow rate exists that balances cooling performance and energy penalty. For this specific battery energy storage system module with bottom cooling, flow rates in the 10–12.5 L/min range offer a favorable compromise, maintaining cell temperatures well within limits without excessive parasitic load.
Finally, I conducted a multi-factor optimization study to identify the best overall configuration for the battery energy storage system thermal management. Based on the individual factor analyses, the optimal choices are: Flow channel aspect ratio W/H = 3 (for low pressure drop), Cooling plate layout: Bottom + Two Sides (for superior cooling and uniformity), and Coolant flow rate: 12.5 L/min (balanced performance). I simulated this combined configuration and compared its performance against the best single-factor configurations (e.g., W/H=3 alone, Bottom+Side alone at 10 L/min total, and 12.5 L/min flow alone with baseline design).
| Configuration | Max Cell Temp. (°C) | Max Temp. Rise (°C) | Global Max ΔT (°C) | Pressure Drop (kPa) / Pump Energy |
|---|---|---|---|---|
| Baseline (W/H=7, Bottom, 10 L/min) | 32.71 | 7.71 | 8.85 | 12.8 kPa / ~138 kJ |
| Single Factor: W/H=3 only | 32.50 | 7.50 | 8.75 | 6.9 kPa / ~74 kJ |
| Single Factor: Bottom+Side only (10 L/min total) | 25.51 | 0.51 | 3.00 | ~5.2 kPa* / ~55 kJ* |
| Single Factor: 12.5 L/min flow only (Bottom, W/H=7) | 32.25 | 7.25 | ~8.70 | ~14.8 kPa / ~185 kJ |
| Multi-Factor Optimal (W/H=3, Bottom+Side, 12.5 L/min total) | 25.40 | 0.40 | 2.90 | ~8.1 kPa* / ~101 kJ* |
*Estimated total for combined plate network.
The multi-factor optimal design achieves the best possible thermal performance: a maximum cell temperature rise of only 0.4°C and an exceptional global temperature uniformity of 2.9°C. It is noteworthy, however, that the performance of the single factor “Bottom+Side cooling” is remarkably close to the multi-factor optimum, with only a 0.1°C difference in maximum temperature rise. This underscores the dominant role that cooling layout plays in determining the thermal behavior of the battery energy storage system. The addition of the optimal channel shape and slightly higher flow rate in the multi-factor design provides only marginal incremental benefit in this case. The primary leverage point is clearly the introduction of side cooling to create a more three-dimensional heat extraction network.
In conclusion, this numerical investigation provides clear guidelines for enhancing the thermal management of battery energy storage systems. The key findings are: First, the aspect ratio of liquid cooling plate channels has a minor effect on cell temperature but a major impact on system pressure drop and pumping energy. A lower aspect ratio (W/H=3) is recommended to reduce parasitic losses. Second, the layout of cooling plates is the most critical design factor. Augmenting traditional bottom cooling with side-mounted plates dramatically reduces maximum cell temperature (by over 21%) and improves temperature uniformity (by over 65%), directly addressing core thermal challenges in battery energy storage systems. Third, there is an optimal range for coolant flow rate (10–12.5 L/min in this study) that provides effective cooling without the severe energy penalty associated with excessively high flow. Finally, a multi-factor optimized design incorporating all these elements yields the best overall performance, but the dominant contributor is unequivocally the multi-sided cooling layout. Implementing these findings can lead to battery energy storage systems that operate safer, last longer, and utilize energy more efficiently, thereby strengthening the foundation for a grid-scale renewable energy ecosystem. Future work could explore the integration of these optimized cold plates with phase change materials for peak heat absorption or investigate advanced control strategies to dynamically adjust flow rates based on real-time thermal load, further pushing the efficiency frontier of battery energy storage system thermal management.