In the context of the global energy transition, the rapid growth of renewable energy penetration has brought significant challenges to the stability of power systems. Energy storage technologies, especially electrochemical energy storage, have become essential to smooth the intermittency of wind and solar power. Among various electrochemical storage devices, lithium-ion batteries are widely adopted due to their high specific energy, low self-discharge rate, long cycle life, and continuously decreasing manufacturing cost. However, the thermal behavior of lithium-ion batteries during charge and discharge is a critical issue. Excessive temperature rise and non-uniform temperature distribution within a battery pack can lead to capacity fade, internal resistance increase, and even thermal runaway. The recommended operating temperature range for lithium-ion batteries is typically 15 to 35 °C, and the maximum temperature difference among cells in a pack should be kept below 5 °C to ensure long-term reliability and safety.
As the demand for large-scale energy storage increases, the capacity of individual cells has risen from small consumer cells (e.g., 2–5 Ah) to large-format cells (e.g., 280 Ah). This shift from small power-type batteries to large energy-storage-type batteries implies significantly higher heat generation per cell under the same C-rate. Most existing studies on battery thermal management have focused on small-capacity cells used in electric vehicles, such as 18650 cylindrical cells or small pouch cells. However, the thermal characteristics of large-format energy storage batteries differ substantially: the larger volume reduces the surface-area-to-volume ratio, making internal heat dissipation more difficult, and the higher absolute heat generation demands more efficient cooling methods. Air cooling, phase change material cooling, and indirect liquid cooling (cold plate) each have limitations when applied to large-format energy storage packs. Direct liquid cooling, also known as immersion cooling, has attracted increasing attention as a promising solution because the cells are directly immersed in a dielectric coolant. This configuration eliminates contact thermal resistance, maximizes heat transfer area, and provides excellent temperature uniformity. Moreover, immersion cooling can effectively suppress thermal runaway propagation due to the large thermal mass of the coolant surrounding each cell.
In this work, I perform a numerical investigation of an immersion liquid cooling system applied to a 280 Ah large-format lithium iron phosphate (LFP) battery pack. The purpose of this study is to systematically evaluate the effects of several design and operating parameters on the cooling performance, including the spacing between cells, the location of the coolant inlet and outlet, the inlet velocity, and the coolant type. I also analyze the sensitivity of the cooling performance to the thermophysical properties of the coolant, namely density, specific heat capacity, thermal conductivity, and dynamic viscosity. The results aim to provide practical guidance for the design of immersion cooling systems in large-scale energy storage battery applications.
Physical and Mathematical Model
I establish a three-dimensional model of a battery pack consisting of 52 prismatic LFP cells arranged in a 4 × 13 matrix. Each cell has a nominal voltage of 3.2 V and a capacity of 280 Ah. The physical dimensions of a single cell are 204 mm in height, 174 mm in width, and 72 mm in thickness. The battery pack is placed inside an insulated container with a height of 230 mm. The distance between the battery module and the container walls is set to 25 mm. The vertical spacing between cells in the same column is fixed at 10 mm, while the horizontal spacing between adjacent columns is varied from 0 to 10 mm to investigate its influence on cooling performance.
The system operates as a closed-loop circulation: the coolant is pumped from a reservoir into the battery container, flows through the gaps between cells, absorbs heat, and then exits to an external heat exchanger where the heat is rejected. I assume the coolant is incompressible and Newtonian, with constant thermophysical properties except for the buoyancy effect in the momentum equation. The flow is assumed to be laminar or turbulent depending on the Reynolds number, and I use the standard \(k-\varepsilon\) turbulence model when necessary. However, for the velocity range investigated (0.2–1.6 m/s) and the characteristic dimensions involved, the flow is mostly in the transitional or turbulent regime. The numerical simulations are conducted using the commercial CFD solver ANSYS Fluent. The governing equations are solved with the finite-volume method, and the SIMPLE algorithm is used for pressure-velocity coupling.
Battery Heat Generation Model
To model the heat generation within the battery, I adopt the widely used Bernardi equation, which assumes uniform heat generation inside the cell. The volumetric heat generation rate \(q_v\) is expressed as:
$$q_v = \frac{1}{V_b}\left(I^2 R_t – I T \frac{\partial U_0}{\partial T}\right)$$
where \(V_b\) is the battery volume (in m³), \(I\) is the current (in A), \(R_t\) is the internal resistance (in Ω), \(T\) is the battery temperature (in K), and \(\partial U_0/\partial T\) is the temperature coefficient of the open-circuit voltage (in V/K). In this study, I set \(\partial U_0/\partial T = -0.4\) mV/K, which gives the reversible heat term approximately 11.16 mV multiplied by the current. The internal resistance is taken as 0.43 mΩ, which is a representative value for a 280 Ah LFP cell at room temperature and 100% state of charge.
For a 1 C discharge, the current is 280 A. The volumetric heat generation rates for different discharge rates used in the validation are listed in Table 1.
| Discharge rate (C) | Current (A) | Heat generation rate \(q_v\) (W/m³) |
|---|---|---|
| 0.25 | 70 | 1130 |
| 0.5 | 140 | 3906 |
| 1.0 | 280 | 14386 |
Governing Equations for Fluid Flow and Heat Transfer
I use the following conservation equations in the fluid domain. The continuity equation is:
$$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0$$
The momentum equation is:
$$\frac{\partial (\rho \mathbf{v})}{\partial t} + \nabla \cdot (\rho \mathbf{v} \mathbf{v}) = -\nabla p + \nabla \cdot (\mu \nabla \mathbf{v}) + \rho \mathbf{g}$$
The energy equation for the fluid is:
$$\rho c_p \left( \frac{\partial T}{\partial t} + \mathbf{v} \cdot \nabla T \right) = \nabla \cdot (\lambda \nabla T)$$
For the solid battery domain, the heat conduction equation is:
$$\rho_b c_b \frac{\partial T_b}{\partial t} = \lambda_x \frac{\partial^2 T_b}{\partial x^2} + \lambda_y \frac{\partial^2 T_b}{\partial y^2} + \lambda_z \frac{\partial^2 T_b}{\partial z^2} + q_v$$
where \(\rho_b\) is the battery density, \(c_b\) is the specific heat capacity of the battery, and \(\lambda_x\), \(\lambda_y\), \(\lambda_z\) are the thermal conductivities in the three principal directions. The battery is considered orthotropic, with in-plane thermal conductivity of 21.6 W/(m·K) and through-plane thermal conductivity of 2.1 W/(m·K). The specific heat capacity is 1029.4 J/(kg·K), and the density is 2118 kg/m³.
Initial and Boundary Conditions
I set the initial temperature of the entire domain to 25 °C. The ambient temperature is also 25 °C. The battery discharges at a 1 C rate from 100% state of charge to the cutoff voltage. The coolant enters the container with a uniform velocity profile normal to the inlet boundary, and the inlet temperature is fixed at 25 °C. The outlet boundary is a pressure outlet at atmospheric pressure (101325 Pa). The external walls of the battery container exchange heat with the ambient environment through natural convection, with a heat transfer coefficient of 5 W/(m²·K). I neglect thermal radiation and contact resistance between cells and the coolant, which is a reasonable assumption for direct immersion cooling where the coolant directly wets the cell surfaces.
Grid Generation and Independence Verification
I use polyhedral mesh for the fluid and solid regions, which provides a good balance between accuracy and computational cost. Figure 1 shows the mesh structure of the computational domain. To verify grid independence, I select a representative case with a cell spacing of 5 mm, an inlet velocity of 0.4 m/s, and a specific inlet/outlet configuration. I evaluate the maximum temperature \(T_{\max}\) and the maximum temperature difference \(\Delta T_{\max}\) of the battery pack for different mesh counts.
The results are summarized in Table 2. When the mesh count increases from 4.5 million to 6.05 million, the changes in \(T_{\max}\) and \(\Delta T_{\max}\) become very small. Further refinement to 7.2 million cells changes the results by less than 1%. Therefore, I choose a mesh count of approximately 6.05 million for all subsequent simulations, which ensures both accuracy and computational efficiency.
| Mesh count | \(T_{\max}\) (°C) | \(\Delta T_{\max}\) (°C) |
|---|---|---|
| 4.5 × 10⁶ | 35.52 | 9.58 |
| 5.3 × 10⁶ | 35.41 | 9.44 |
| 6.05 × 10⁶ | 35.35 | 9.37 |
| 7.2 × 10⁶ | 35.33 | 9.35 |
Validation of the Battery Heat Generation Model
To validate the heat generation model, I compare the simulated average surface temperature of a single cell during 1 C discharge with experimental data from a previous study. The comparison is shown in Figure 2. At the beginning of discharge, the simulated values agree well with the experimental results. At the end of discharge, the simulated average temperature is 34.9 °C, while the experimental value is 33.2 °C. The maximum relative error is about 5.1%, which is acceptable for engineering analysis. This confirms that the heat generation model can adequately reproduce the actual thermal behavior of the 280 Ah cell.
Results and Discussion
Effect of Cell Spacing on Temperature Rise Characteristics
To investigate the influence of the horizontal cell spacing \(d_i\) on the cooling performance, I fix the inlet/outlet configuration as Case5 (inlet height 102 mm, outlet height 102 mm), the inlet velocity at 0.4 m/s, and the inlet/outlet diameter at 40 mm. I vary the spacing \(d_i\) from 0 to 10 mm in increments of 2.5 mm. The maximum temperature difference \(\Delta T_{\max}\) and maximum temperature \(T_{\max}\) of the battery pack at the end of the 1 C discharge are shown in Table 3.
| Spacing \(d_i\) (mm) | \(\Delta T_{\max}\) (°C) | \(T_{\max}\) (°C) |
|---|---|---|
| 0 | 10.94 | 37.19 |
| 2.5 | 10.64 | 36.61 |
| 5.0 | 9.37 | 35.35 |
| 7.5 | 9.62 | 35.58 |
| 10.0 | 9.86 | 35.75 |
The results reveal an interesting non-monotonic behavior. When the spacing is increased from 0 to 5 mm, both \(\Delta T_{\max}\) and \(T_{\max}\) decrease significantly. The maximum temperature difference drops from 10.94 °C to 9.37 °C (a reduction of 1.57 °C), and the maximum temperature drops from 37.19 °C to 35.35 °C (a reduction of 1.84 °C). This improvement is attributed to two factors. First, the increase in spacing expands the heat transfer area between the coolant and the battery surfaces, since the coolant can now flow around both sides of each cell. Second, the larger gaps reduce the flow resistance, allowing the coolant to penetrate more uniformly through the pack. However, when the spacing is further increased beyond 5 mm, the cooling performance deteriorates. At a spacing of 10 mm, \(\Delta T_{\max}\) increases to 9.86 °C and \(T_{\max}\) to 35.75 °C. This is because, at a constant volumetric flow rate, a larger cross-sectional area leads to a lower coolant velocity, which weakens the convective heat transfer coefficient. The trade-off between increased heat transfer area and decreased flow velocity results in an optimal spacing of 5 mm for the present configuration.
To further understand the flow behavior, I examine the velocity contours at a horizontal plane located at z = 102 mm (mid-height of the cells). The velocity fields for different spacings are shown in Figure 3. When the cells are in direct contact (spacing = 0), the coolant flows mainly around the periphery of the module, leaving the central region relatively stagnant. As the spacing increases, the coolant is able to flow through the gaps, creating a more uniform velocity distribution across the entire battery module. At 5 mm spacing, the flow pattern appears most uniform, with no large stagnant zones. At 10 mm spacing, the velocity in the gaps is noticeably lower due to the increased cross-sectional area, which reduces the local convective heat transfer.
Effect of Coolant Inlet and Outlet Positions
I further investigate the influence of the coolant inlet and outlet positions on the cooling performance. The nine configurations are defined by varying the heights of the inlet and outlet openings relative to the bottom of the battery container. The inlet height can be 182 mm (top), 102 mm (middle), or 20 mm (bottom), and the outlet height can likewise be 182 mm, 102 mm, or 20 mm. I label these configurations as Case1 through Case9, as shown in Table 4. For this parametric study, I fix the cell spacing at 5 mm, the inlet velocity at 0.4 m/s, and the inlet/outlet diameter at 40 mm.
| Case | Inlet height (mm) | Outlet height (mm) |
|---|---|---|
| Case1 | 182 | 182 |
| Case2 | 182 | 102 |
| Case3 | 182 | 20 |
| Case4 | 102 | 182 |
| Case5 | 102 | 102 |
| Case6 | 102 | 20 |
| Case7 | 20 | 182 |
| Case8 | 20 | 102 |
| Case9 | 20 | 20 |
Table 5 summarizes the simulated \(\Delta T_{\max}\) and \(T_{\max}\) values for all nine configurations. Comparing Case4, Case5, and Case6, which all have an inlet height of 102 mm, the outlet height has only a minor effect. The maximum temperature difference values are 9.32 °C, 9.38 °C, and 9.41 °C, respectively, corresponding to differences of less than 1%. The maximum temperature also varies by only 0.1 °C. This indicates that the outlet position is not a critical design parameter for immersion cooling as long as the outlet is located somewhere on the opposite side to prevent short-circuiting.
In contrast, the inlet position significantly affects the results. For a fixed outlet height of 182 mm (Case1, Case4, Case7), the \(\Delta T_{\max}\) changes from 9.87 °C (Case1) to 9.32 °C (Case4) and then to 9.73 °C (Case7). The middle inlet position (Case4) yields the best performance, with \(\Delta T_{\max}\) reduced by about 5.6% compared to the top inlet (Case1) and 4.2% compared to the bottom inlet (Case7). A similar trend is observed when the outlet is at other heights. The optimal inlet position is near the mid-height of the battery pack, because it creates a more symmetrical flow pattern and better coolant distribution around the cells. When the inlet is too high or too low, the coolant tends to flow preferentially through the upper or lower region, leaving the opposite region with poorer cooling.
| Case | \(\Delta T_{\max}\) (°C) | \(T_{\max}\) (°C) |
|---|---|---|
| Case1 | 9.87 | 35.88 |
| Case2 | 9.92 | 35.93 |
| Case3 | 9.95 | 35.96 |
| Case4 | 9.32 | 35.29 |
| Case5 | 9.38 | 35.35 |
| Case6 | 9.41 | 35.39 |
| Case7 | 9.73 | 35.72 |
| Case8 | 9.78 | 35.77 |
| Case9 | 9.81 | 35.80 |
To visualize the flow characteristics, I plot velocity contours and vorticity contours on a vertical plane passing through the center of the battery pack (x = 388 mm) for several representative cases. The results show that the inlet jet induces a strong vortex near the inlet region. The location and intensity of this vortex depend strongly on the inlet position. For Case1 (top inlet), the vortex forms below the inlet, recirculating the coolant in the upper part of the container. For Case7 (bottom inlet), the vortex forms above the inlet, confining the mixing to the lower part. For Case4 (middle inlet), the vortex is distributed on both sides of the inlet, creating a more balanced flow field. The average vorticity in Case4 is about 21% higher than in Case1 and 23.5% higher than in Case7. Higher vorticity enhances mixing and improves heat transfer between the coolant and the battery surfaces. The outlet position, on the other hand, has minimal influence on the vortex structure, as evidenced by similar velocity and vorticity fields for Case1, Case2, and Case3.
Effect of Coolant Inlet Velocity
The inlet velocity is a key operating parameter that directly affects the convective heat transfer coefficient and the pressure drop. I select the optimal configuration from the previous section: Case4 (inlet height 102 mm, outlet height 182 mm), cell spacing 5 mm, and inlet/outlet diameter 40 mm. The inlet velocity is varied from 0.2 m/s to 1.6 m/s in steps of 0.2 m/s. The resulting \(\Delta T_{\max}\) and \(T_{\max}\) are presented in Table 6.
| Inlet velocity (m/s) | \(\Delta T_{\max}\) (°C) | \(T_{\max}\) (°C) | Pressure drop \(\Delta P\) (Pa) |
|---|---|---|---|
| 0.2 | 11.83 | 38.37 | — |
| 0.4 | 9.32 | 35.29 | — |
| 0.6 | 7.98 | 33.61 | — |
| 0.8 | 7.12 | 32.41 | — |
| 1.0 | 6.48 | 31.62 | — |
| 1.2 | 6.02 | 31.12 | — |
| 1.4 | 5.82 | 31.15 | — |
| 1.6 | 5.70 | 31.00 | — |
As expected, both \(\Delta T_{\max}\) and \(T_{\max}\) decrease monotonically with increasing inlet velocity. However, the rate of improvement diminishes as the velocity increases. When the velocity is raised from 0.2 m/s to 0.4 m/s, \(\Delta T_{\max}\) is reduced by 2.51 °C (21.2%) and \(T_{\max}\) by 3.08 °C (8.0%). In contrast, when the velocity is increased from 1.4 m/s to 1.6 m/s, the reductions are only 0.12 °C (2.0%) and 0.15 °C (0.5%). This indicates that there is a practical upper limit beyond which further increases in flow rate provide diminishing returns while consuming more pumping power. For the 280 Ah battery pack considered here, an inlet velocity of around 0.8–1.0 m/s appears to be a reasonable compromise between cooling performance and energy consumption, as the temperature difference at 1.0 m/s is already below the commonly recommended threshold of 5 °C? Actually, the value is 6.48 °C at 1.0 m/s, which is still above 5 °C. To achieve \(\Delta T_{\max}\) below 5 °C, velocities above 1.4 m/s are required. At 1.6 m/s, \(\Delta T_{\max}\) is 5.70 °C, still slightly above 5 °C. This suggests that the current design may need further optimization, such as increasing the number of inlet/outlet ports or using a coolant with better thermal properties.
I also examine the transient temperature evolution for different velocities. At low velocities (≤0.6 m/s), the battery temperature continues to rise throughout the discharge process, indicating that the cooling system is unable to remove heat at the same rate as it is generated. At higher velocities (≥0.8 m/s), the temperature reaches a quasi-steady state before the end of discharge. The time required to reach steady state decreases with increasing velocity. For example, at 0.8 m/s it takes about 3500 s, at 1.0 m/s about 3250 s, at 1.2 m/s about 3050 s, at 1.4 m/s about 2900 s, and at 1.6 m/s about 2750 s. This behavior is consistent with the enhanced convection at higher flow rates.
Figure 4 shows the temperature distribution on the battery pack at the end of discharge for various inlet velocities. At 0.2 m/s, the temperature field is highly non-uniform, with hot spots near the outlet side. As the velocity increases, the temperature distribution becomes more uniform, and the hot spots are reduced. The improvement from 0.2 to 0.4 m/s is particularly noticeable, which aligns with the quantitative results in Table 6.
Effect of Coolant Type
The choice of coolant is a fundamental decision for immersion cooling systems. I compare five different dielectric coolants commonly considered for battery thermal management: synthetic oil, ester-based MIVOLT-DF7, electronic fluorinated fluid FC-72, deionized water, and silicone oil. Their thermophysical properties at 25 °C are listed in Table 7. Note that deionized water is not typically dielectric, but it can be used in immersion cooling if the electrical insulation of cells is ensured; I include it for comparison because of its excellent thermal properties.
| Coolant | Density \(\rho\) (kg/m³) | Specific heat \(c_p\) (J/(kg·K)) | Thermal conductivity \(\lambda\) (W/(m·K)) | Dynamic viscosity \(\mu\) (Pa·s) | Dielectric constant |
|---|---|---|---|---|---|
| Synthetic oil | 807 | 2523 | 0.159 | 0.0070 | 2.20 |
| MIVOLT-DF7 | 916 | 1907 | 0.129 | 0.0150 | — |
| FC-72 | 1680 | 1100 | 0.057 | 0.0006 | 1.75 |
| Deionized water | 998 | 4182 | 0.598 | 0.0010 | 80.20 |
| Silicone oil | 965 | 1460 | 0.160 | 0.0965 | 16.00 |
I run simulations using the optimal geometric parameters identified earlier: Case4 configuration, cell spacing 5 mm, inlet diameter 40 mm, and an inlet velocity of 1.6 m/s. The simulated results for \(\Delta T_{\max}\) and \(T_{\max}\) are shown in Table 8.
| Coolant | \(\Delta T_{\max}\) (°C) | \(T_{\max}\) (°C) | Surface heat transfer coefficient (W/(m²·K)) | Pressure drop (Pa) |
|---|---|---|---|---|
| Deionized water | 3.61 | 28.64 | 22.69 | — |
| Synthetic oil | 5.72 | 30.93 | — | 2543 |
| FC-72 | 6.02 | 31.11 | — | 4865 |
| MIVOLT-DF7 | 6.22 | 31.74 | — | — |
| Silicone oil | 8.78 | 34.63 | 18.81 | — |
Deionized water provides the best cooling performance, with \(\Delta T_{\max} = 3.61\) °C and \(T_{\max} = 28.64\) °C. This is due to its high specific heat capacity (4182 J/(kg·K)), high thermal conductivity (0.598 W/(m·K)), and relatively low viscosity (0.001 Pa·s). These properties enable efficient heat absorption and transport. In contrast, silicone oil exhibits the worst performance, with \(\Delta T_{\max} = 8.78\) °C and \(T_{\max} = 34.63\) °C. Compared to deionized water, silicone oil yields a 143% higher temperature difference and a 20.9% higher maximum temperature. The high viscosity of silicone oil (0.0965 Pa·s) severely impedes flow and reduces convective heat transfer, despite its moderate density and heat capacity.
The performance of synthetic oil, MIVOLT-DF7, and FC-72 falls in between. Their \(\Delta T_{\max}\) values are 5.72 °C, 6.22 °C, and 6.02 °C, respectively. Synthetic oil performs slightly better than the other two due to its higher specific heat and reasonable viscosity. The pressure drop across the system varies significantly with coolant properties: synthetic oil yields the lowest pressure drop (2543 Pa), while FC-72 produces the highest (4865 Pa), which is 91% higher than synthetic oil. This is because FC-72 has a high density (1680 kg/m³) and very low viscosity, but the high density increases the kinetic energy required to accelerate the fluid.
The surface heat transfer coefficient \(h\) is inversely related to \(\Delta T_{\max}\). Deionized water has the highest \(h\) (22.69 W/(m²·K)), followed by synthetic oil and then silicone oil (18.81 W/(m²·K)). This confirms that the thermal boundary layer resistance is lowest for water, which has high thermal conductivity.
Sensitivity Analysis of Coolant Thermophysical Properties
Although the coolant type comparison is useful, it does not isolate the individual contribution of each thermophysical property. To quantify the influence of density, specific heat capacity, thermal conductivity, and dynamic viscosity on the cooling performance, I perform a sensitivity analysis. I take two baseline coolants: synthetic oil (Group A) and deionized water (Group B). For Group A, the inlet velocity is 1.2 m/s and the configuration is Case5 (middle inlet, middle outlet). For Group B, the inlet velocity is 1.6 m/s and the configuration is Case4 (middle inlet, top outlet). In each case, I vary one property from 60% to 140% of its baseline value while keeping the other three properties constant. The resulting \(\Delta T_{\max}\) values are recorded.
Table 9 presents the results for Group A (synthetic oil), and Table 10 for Group B (deionized water). The data show that density has the most significant influence on \(\Delta T_{\max}\). When the density is varied from 60% to 140% of the baseline, \(\Delta T_{\max}\) decreases from 7.65 °C to 5.20 °C in Group A, a reduction of 32%. For Group B, the decrease is from 4.65 °C to 3.06 °C, a reduction of 34%. The reason is that higher density increases the momentum of the coolant jet and intensifies the disturbance inside the battery container, leading to better mixing and heat transfer. Additionally, higher density enhances the volumetric heat capacity (ρ·cp), allowing the fluid to absorb more heat per unit volume. However, a higher density also increases the pumping power requirement, as the pressure drop grows roughly linearly with density.
Specific heat capacity is the second most influential property. For Group A, increasing \(c_p\) from 60% to 140% of baseline reduces \(\Delta T_{\max}\) from 7.42 °C to 5.26 °C (a 29% reduction). For Group B, the reduction is from 4.29 °C to 3.24 °C (a 24% reduction). A higher specific heat capacity means that more heat can be removed per unit temperature rise of the coolant, thus lowering the battery temperature. Thermal conductivity ranks third. In Group A, \(\Delta T_{\max}\) decreases from 6.89 °C to 5.48 °C (20% reduction) when \(\lambda\) is increased from 60% to 140%. In Group B, it decreases from 4.02 °C to 3.35 °C (17% reduction). Thermal conductivity directly affects the heat transfer coefficient at the cell surface; higher \(\lambda\) reduces the thermal boundary layer resistance.
Dynamic viscosity has the opposite effect: increasing viscosity degrades the cooling performance. For Group A, raising \(\mu\) from 60% to 140% increases \(\Delta T_{\max}\) from 5.70 °C to 6.20 °C (a 9% increase). For Group B, the increase is from 3.22 °C to 3.75 °C (a 16% increase). This is because higher viscosity leads to a thicker boundary layer and lower Reynolds number for the same inlet velocity, both of which reduce convective heat transfer. However, the sensitivity to viscosity is smaller than to the other three properties within the range considered.
The overall ranking of the influence of thermophysical properties on the cooling performance is as follows: density > specific heat capacity > thermal conductivity > dynamic viscosity. This finding suggests that when selecting a coolant for immersion cooling of energy storage batteries, the density and specific heat capacity should be prioritized, followed by thermal conductivity, while the dynamic viscosity is a secondary factor. Nevertheless, viscosity also influences the pressure drop and pumping power, so it cannot be ignored in the system design.
| Property variation | Density effect \(\Delta T_{\max}\) (°C) | Specific heat effect \(\Delta T_{\max}\) (°C) | Thermal conductivity effect \(\Delta T_{\max}\) (°C) | Dynamic viscosity effect \(\Delta T_{\max}\) (°C) |
|---|---|---|---|---|
| 60% | 7.65 | 7.42 | 6.89 | 5.70 |
| 80% | 6.40 | 6.22 | 6.08 | 5.82 |
| 100% | 5.72 | 5.72 | 5.72 | 5.72 |
| 120% | 5.42 | 5.44 | 5.58 | 5.96 |
| 140% | 5.20 | 5.26 | 5.48 | 6.20 |
| Property variation | Density effect \(\Delta T_{\max}\) (°C) | Specific heat effect \(\Delta T_{\max}\) (°C) | Thermal conductivity effect \(\Delta T_{\max}\) (°C) | Dynamic viscosity effect \(\Delta T_{\max}\) (°C) |
|---|---|---|---|---|
| 60% | 4.65 | 4.29 | 4.02 | 3.22 |
| 80% | 4.02 | 3.82 | 3.68 | 3.40 |
| 100% | 3.61 | 3.61 | 3.61 | 3.61 |
| 120% | 3.28 | 3.38 | 3.46 | 3.68 |
| 140% | 3.06 | 3.24 | 3.35 | 3.75 |
Practical Implications for Energy Storage Battery Thermal Management
The findings of this study provide several useful insights for designing immersion cooling systems for large-capacity energy storage battery packs. First, the cell spacing should be carefully optimized. For the 280 Ah cells studied, a horizontal spacing of 5 mm yields the best balance between heat transfer area and flow velocity. Too small a spacing leads to poor coolant access, while too large a spacing reduces the local velocity and diminishes the cooling efficiency. Second, the coolant inlet should be positioned near the mid-height of the pack to ensure a symmetric flow distribution. The outlet position is less critical, but it should be placed on the opposite side to promote cross-flow. Third, higher inlet velocities improve the cooling performance but with diminishing returns. A velocity of 0.8–1.0 m/s may be sufficient for many applications, but if the target is to keep \(\Delta T_{\max}\) below 5 °C, a velocity above 1.4 m/s is necessary for the present design with deionized water. Fourth, the choice of coolant is extremely important. Deionized water offers the best thermal performance but may require additional electrical insulation measures due to its high dielectric constant. For applications where dielectric properties are mandatory, synthetic oil or ester-based fluids are reasonable compromises, while silicone oil is less favorable due to its high viscosity. fifth, the sensitivity analysis reveals that density and specific heat capacity dominate the cooling performance. Therefore, when evaluating new coolants, one should prioritize fluids with high density and high specific heat capacity. Meanwhile, additives that increase thermal conductivity and reduce viscosity can further improve performance, but their effects are secondary.
Conclusion
In this work, I present a comprehensive numerical study of an immersion liquid cooling system for a 280 Ah large-format energy storage battery pack. The effects of cell spacing, inlet/outlet positions, inlet velocity, and coolant type on the maximum temperature and maximum temperature difference are systematically investigated. The key conclusions are as follows:
(1) The cell spacing has a non-monotonic effect on cooling performance. The optimal horizontal spacing for the present pack is 5 mm, which reduces the maximum temperature difference by 1.57 °C and the maximum temperature by 1.84 °C compared to a zero spacing configuration. Further increasing the spacing degrades the cooling performance due to reduced flow velocity.
(2) The inlet position has a more significant influence than the outlet position. The optimal inlet height is near the mid-height of the battery pack. Inlet heights at the top or bottom increase the maximum temperature difference by about 4–6% compared to the optimal middle position. The outlet position affects the results by less than 1% for the range studied.
(3) Increasing the inlet velocity always improves the cooling performance, but with diminishing returns. The maximum temperature difference and maximum temperature decrease by 21.2% and 8.0% when the velocity is increased from 0.2 to 0.4 m/s, but only by 2.0% and 0.5% when the velocity is increased from 1.4 to 1.6 m/s.
(4) Among the five coolants evaluated, deionized water provides the best cooling performance, with a maximum temperature difference of 3.61 °C and a maximum temperature of 28.64 °C at an inlet velocity of 1.6 m/s. Silicone oil performs the worst, with corresponding values of 8.78 °C and 34.63 °C. The order of cooling performance is deionized water > synthetic oil > FC-72 > MIVOLT-DF7 > silicone oil.
(5) The sensitivity analysis shows that the influence of coolant thermophysical properties on the cooling performance follows the order: density > specific heat capacity > thermal conductivity > dynamic viscosity. This ranking can be used as a guideline for selecting or developing new immersion coolants for energy storage batteries.
These findings are directly relevant to the design and optimization of liquid cooling systems for large-format energy storage batteries. Future work will explore the effect of different discharge rates, transient thermal behavior during fast charging, and the influence of flow distribution manifolds on the temperature uniformity of the battery pack.