In my recent work, I focused on improving the performance of immersion liquid cooling systems for energy storage battery modules. The thermal management of energy storage battery systems is a critical challenge because lithium-ion batteries generate significant heat during operation, especially at high discharge rates. If this heat is not removed effectively, the battery temperature rises, leading to reduced efficiency, accelerated aging, and even thermal runaway. Among various cooling strategies, immersion liquid cooling is considered one of the most promising due to its direct contact between coolant and battery surfaces, which provides superior heat transfer compared with air cooling or indirect cold plates. However, traditional immersion-cooled modules often suffer from uneven flow distribution and high temperatures in the middle and rear sections of the pack. To solve these problems, I introduced a novel module design that incorporates a series of baffles placed between the batteries in the downstream half of the module. In this article, I present a detailed numerical analysis of the flow and heat transfer characteristics of this new design, compare it with the conventional module, investigate the effects of coolant flow rate and initial temperature, and develop a correlation for predicting the Nusselt number for the cooling process.
The importance of reliable thermal management for energy storage battery systems cannot be overstated. The performance of an energy storage battery is strongly influenced by its operating temperature. A high temperature accelerates side reactions inside the cell, increases internal resistance, and degrades the electrode materials. Non-uniform temperature distribution within a battery pack leads to unbalanced state-of-charge and state-of-health among cells, which shortens the overall pack lifetime. Therefore, both the maximum temperature and the temperature difference must be carefully controlled. Immersion cooling, where the cells are directly submerged in a dielectric coolant, offers excellent heat transfer coefficients and can reduce temperature gradients effectively. The coolant used in this study is Novec 7000, a fluorinated fluid with low dielectric constant, good material compatibility, and favorable thermophysical properties. The cells considered are prismatic lithium iron phosphate batteries, each with dimensions of 174 mm × 71 mm × 207 mm. The module contains eight cells in series, and the gap between adjacent cells as well as between the cells and the module walls is 35.5 mm, which equals half the battery thickness. The inlet and outlet are located on the bottom and top of the side walls, respectively. During operation, the coolant enters from the bottom inlet, floods the module, and flows over the battery surfaces, then exits through the top outlet.
In the conventional immersion-cooled module, I observed that the average temperature of each individual cell increases continuously during discharge, but the temperature rise is not uniform. The cells near the inlet are cooler than the cells in the middle and rear portions of the module. This is because the coolant absorbs heat as it travels, so its temperature rises and the heat transfer driving force between the coolant and the downstream cells diminishes. The highest average temperature occurs at cell 6 and cell 7, while cell 1 has the lowest. The flow field inside the conventional module also shows significant maldistribution. The coolant velocity is high near the side walls and low in the spaces between cells. In fact, there are almost stagnant regions between cells 4 and 8, which severely impairs heat transfer. The reason is that the battery surfaces block the flow, and the narrow gaps cause high flow resistance. Since the inlet and outlet are on opposite sides, the main flow bypasses the central region and prefers the open side channels. As a result, the rear cells are poorly cooled, leading to hot spots and a large maximum temperature difference within the module.
To overcome these drawbacks, I proposed a novel design with an internal baffle structure placed in the gaps between cells 3 to 8. Each baffle is located at the centerline of the gap and is oriented perpendicular to the main flow direction. The baffles are staggered in such a way that they redirect the coolant from the side channels into the inter-cell gaps, generating active cross-flow over the battery surfaces. This forced diversion enhances local convective heat transfer and helps to homogenize the flow distribution. My numerical simulations show that this design significantly improves both flow uniformity and thermal performance. A systematic comparison between the conventional and novel modules under the same operating conditions (coolant flow rate of 3 L/min, initial temperature of 25 °C, and 1C discharge rate) revealed the following benefits. The maximum average cell temperature decreased by 2.2% relative to the conventional module, and the cell that experiences the highest temperature shifted from cell 6 to cell 8. More importantly, the average temperature rise of all cells was reduced by 24.9%, indicating that the baffles effectively enhance the overall cooling capacity. The maximum temperature difference among cells in the module decreased by 28.3%, confirming improved temperature uniformity. Additionally, the local hotspot temperature of the hottest cell was reduced by 20% compared to the conventional design.
The improvement can be attributed to the flow guidance provided by the baffles. In the traditional module, the coolant tends to bypass the inter-cell gaps because the flow resistance is high. The baffles block the side channel partially and force a portion of the coolant to pass through the gaps, thereby increasing the velocity and turbulence in those regions. The longer flow path also increases the contact area between the coolant and the battery surfaces, which enhances the effective heat transfer area. However, it is important to note that adding baffles increases the pressure drop. In my simulation, the pressure loss increased from 47.21 Pa in the conventional module to 61.59 Pa in the novel module, a rise of 30.5%. While this increased pumping power is undesirable, the benefits in terms of temperature uniformity and reduced hot spots outweigh this downside, especially because improved temperature uniformity directly enhances the safety and cycle life of the energy storage battery system.
After confirming the superiority of the novel module, I proceeded to investigate the influence of operating parameters. I studied five different coolant flow rates: 1, 2, 3, 4, and 5 L/min, while keeping the initial temperature at 25 °C and the discharge rate at 1C. I also studied five different initial coolant temperatures: 15, 17.5, 20, 22.5, and 25 °C, with a fixed flow rate of 3 L/min. The results for the flow rate study show that increasing the coolant flow rate reduces the average temperature of the hottest cell. Specifically, when the flow rate was increased from 1 L/min to 5 L/min, the average temperature of the high-temperature cell dropped from 28.7 °C to 27.1 °C, a reduction of 5.7%. The local hotspot temperature of that cell decreased by 5.2% from 29.0 °C to 27.5 °C. The temperature rise of all cells also decreased significantly, with the maximum cell temperature rise being reduced by 48.9% and the average temperature rise of all cells by 44.1%. The maximum temperature difference within the module was reduced by 47.6%, from 1.89 °C to 0.99 °C. This is because higher coolant velocity enhances convective heat transfer and also reduces the temperature gradient between the front and rear cells, since the coolant does not heat up as much along the path.
On the other hand, increasing the coolant flow rate also increases pressure loss. The pressure drop rose by a factor of 4.7 when the flow rate was increased from 1 L/min to 5 L/min. In my comparison with a similar study from the literature, the pressure drop increase in their module was 7.4 times for the same flow rate range, which means my novel module has a 36.6% lower pressure drop increase. This indicates that the baffle design helps to distribute the flow more efficiently without incurring excessive pumping power. Nevertheless, the trade-off between heat transfer enhancement and pressure loss must be considered when selecting the optimal flow rate for practical energy storage battery thermal management systems. Lower flow rates are suitable for low-power applications where energy efficiency of the cooling system is critical, while higher flow rates are necessary for high-rate discharge or high ambient temperature conditions.
The influence of coolant initial temperature is also substantial. I observed that when the initial temperature was raised from 15 °C to 25 °C, the average temperature of the hottest cell at the end of discharge increased by 28.2% (from 19.92 °C to 27.74 °C). The local hotspot temperature increased by 33.2% (from 21.1 °C to 28.1 °C). This is straightforward: a higher coolant temperature reduces the temperature difference between the cell and the coolant, which lowers the heat transfer rate. However, a surprising result is that the temperature uniformity improved with increasing initial temperature. The maximum temperature difference between cells decreased by 59.5%, from 3.51 °C at 15 °C to 1.42 °C at 25 °C. The reason is that when the coolant is much colder than the cells, the cells near the inlet are over-cooled, creating a large temperature gradient. At a higher initial temperature, the entire module operates closer to the battery’s optimal temperature range, and the cooling effect is more uniform across all cells. This trade-off demonstrates that there is an optimal initial temperature that balances the absolute temperature and the temperature uniformity. For many energy storage battery systems, the recommended operating temperature is between 20 °C and 30 °C, so an initial coolant temperature around 20 °C may provide a good compromise.
To provide a quantitative summary of the effects of different parameters, I have collected the key results in the following tables.
| Coolant flow rate (L/min) | Average temperature of hottest cell (°C) | Hotspot temperature (°C) | Maximum temperature difference (°C) | Pressure drop (Pa) |
|---|---|---|---|---|
| 1 | 28.7 | 29.0 | 1.89 | 17.3 |
| 2 | 27.9 | 28.2 | 1.41 | 31.5 |
| 3 | 27.4 | 27.8 | 1.18 | 61.6 |
| 4 | 27.2 | 27.6 | 1.05 | 73.2 |
| 5 | 27.1 | 27.5 | 0.99 | 98.4 |
| Initial coolant temperature (°C) | Average temperature of hottest cell (°C) | Hotspot temperature (°C) | Maximum temperature difference (°C) |
|---|---|---|---|
| 15 | 19.92 | 21.1 | 3.51 |
| 17.5 | 22.10 | 23.2 | 2.90 |
| 20 | 24.35 | 25.3 | 2.35 |
| 22.5 | 26.08 | 27.0 | 1.88 |
| 25 | 27.74 | 28.1 | 1.42 |
Another important aspect of my study is the development of a predictive correlation for the heat transfer process inside the novel energy storage battery module. In engineering practice, it is useful to have a simple formula that can estimate the Nusselt number as a function of operating conditions, without performing time-consuming CFD simulations. I focused on the convective heat transfer between the coolant and the battery surfaces. The relevant dimensionless numbers are the Reynolds number (Re) and the Nusselt number (Nu). The Reynolds number is defined as:
$$Re = \frac{\rho_f u l}{\mu}$$
where \(\rho_f\) is the coolant density, \(u\) is the characteristic velocity, \(l\) is the characteristic length, and \(\mu\) is the dynamic viscosity. The Nusselt number is defined as:
$$Nu = \frac{h_{conv} l}{\lambda}$$
where \(h_{conv}\) is the convective heat transfer coefficient and \(\lambda\) is the coolant thermal conductivity. The heat transfer coefficient is calculated from:
$$h_{conv} = \frac{\Phi}{A \Delta T}$$
where \(\Phi\) is the total heat transfer rate between the coolant and the batteries, \(A\) is the total surface area of the batteries, and \(\Delta T\) is the mean temperature difference between the battery surface and the coolant. Since the temperature range in this study is near room temperature, the thermophysical properties of the coolant are almost constant, so I ignored their variations. I then proposed a correlation of the form:
$$Nu = C_1 Re^{C_2} \left(\frac{T_0}{T_{\infty}}\right)^{C_3} + C_0$$
where \(T_0\) is the initial coolant temperature, \(T_{\infty}\) is a reference temperature (taken as 25 °C), and \(C_0, C_1, C_2, C_3\) are constants to be determined from the numerical results. Using a nonlinear regression of the simulated data, I obtained:
$$Nu = 56.4075\, Re^{0.187} \left(\frac{T_0}{T_{\infty}}\right)^{-3.5195} – 13.2641$$
This correlation is valid for coolant flow rates between 1 L/min and 5 L/min and initial temperatures between 15 °C and 25 °C. To assess the accuracy of the correlation, I compared its predictions with the numerical simulation results. The average relative error was 2.0%, and the maximum relative error was 5.4% within the investigated range. The agreement is excellent, which indicates that the correlation can be used to predict the heat transfer performance of the novel energy storage battery module with high confidence. The negative exponent on the temperature ratio reveals that increasing the initial coolant temperature reduces the Nusselt number, which aligns with the observed decrease in heat transfer performance. The positive but small exponent on the Reynolds number indicates that the flow is not fully developed turbulent, and the effect of velocity on heat transfer is moderate in this regime.
The governing equations used in my numerical model include the continuity, momentum, and energy equations for the coolant, as well as the heat conduction equation for the battery with a multi-scale multi-domain (MSMD) electro-thermal model. The shear-stress transport (SST) k-ω turbulence model was employed to close the Reynolds-averaged Navier-Stokes equations. The governing equations are:
Continuity equation:
$$\frac{\partial \rho_f}{\partial t} + \frac{\partial}{\partial x_i}(\rho_f u_i) = 0$$
Momentum equation:
$$\frac{\partial}{\partial t}(\rho_f u_i) + \frac{\partial}{\partial x_j}(\rho_f u_i u_j) = -\frac{\partial p}{\partial x_i} + \frac{\partial}{\partial x_j}\left[ \mu \left( \frac{\partial u_i}{\partial x_j} + \frac{\partial u_j}{\partial x_i} – \frac{2}{3}\delta_{ij}\frac{\partial u_k}{\partial x_k} \right) \right] + \frac{\partial}{\partial x_j}(-\rho_f \overline{u_i’ u_j’}) + \rho_f g_i$$
Energy equation for the fluid:
$$\frac{\partial (\rho_f c_f T_f)}{\partial t} + \nabla \cdot (\rho_f \mathbf{u} c_f T_f) = \nabla \cdot (h_{eff} \nabla T_f) + \tau_{eff} \cdot \mathbf{u}$$
Energy equation for the battery:
$$\frac{\partial (\rho_c c_c T_c)}{\partial t} = \nabla \cdot (k_c \nabla T_c) + \sigma_+ |\nabla \Phi_+|^2 + \sigma_- |\nabla \Phi_-|^2 + \dot{q}_{ECh} + \dot{q}_{short}$$
In the battery energy equation, \(\rho_c\), \(c_c\), and \(k_c\) are the equivalent density, specific heat, and thermal conductivity of the cell; \(\Phi_+\) and \(\Phi_-\) are the potentials of the positive and negative electrodes; \(\sigma_+\) and \(\sigma_-\) are their effective conductivities; and \(\dot{q}_{ECh}\) and \(\dot{q}_{short}\) are the heat generation rates due to electrochemical reactions and internal short circuits, respectively.
To model the electrical behavior of the battery, I used a second-order equivalent circuit model (ECM). The terminal voltage \(V\) is given by:
$$V = V_{ocv}(soc) – V_1 – V_2 – R_s(soc) I(t)$$
where \(V_{ocv}\) is the open-circuit voltage as a function of state of charge (SOC), \(R_s\) is the ohmic resistance, and \(V_1\) and \(V_2\) are the voltages across the two RC networks. The dynamics of these voltages are:
$$\frac{dV_1}{dt} = -\frac{1}{R_1(soc) C_1(soc)} V_1 – \frac{1}{C_1(soc)} I(t)$$
$$\frac{dV_2}{dt} = -\frac{1}{R_2(soc) C_2(soc)} V_2 – \frac{1}{C_2(soc)} I(t)$$
The state of charge evolves as:
$$\frac{d(soc)}{dt} = \frac{I(t)}{3600 Q_{ref}}$$
where \(Q_{ref}\) is the battery capacity in ampere-hours. This model allows accurate prediction of heat generation during the discharge process, which is essential for evaluating the thermal management system.
In my simulation setup, the initial battery temperature and ambient temperature were both set to 25 °C. The battery was discharged at a 1C rate. For the model validation, I compared the simulated surface center temperature with experimental data from the literature. The maximum relative error was 4.6%, which demonstrates that the numerical model is sufficiently accurate. This validation is crucial because it ensures the reliability of the subsequent parameter studies and correlation development. The mesh independence study was also conducted, and I selected a grid with 753,289 polyhedral cells as the baseline, as further refinement did not change the results significantly.
One of the interesting findings from the flow field analysis is the difference in the high-temperature cell location between the conventional and novel modules. In the conventional module, cell 6 is the hottest, while in the novel module, cell 8 becomes the hottest. The reason is that the baffles significantly enhance cooling in the central cells (cells 4–7), so the temperature peak shifts toward the outlet. Even so, the temperature of cell 8 in the novel module is lower than the temperature of cell 6 in the conventional module, indicating an overall improvement. The local hotspot temperature, which is a critical metric for safety, also decreases. The hotspot often occurs near the tabs or at the core of the cell, where heat generation is highest and heat dissipation is most difficult. By enhancing the heat transfer on the cell surfaces, the baffles effectively reduce the internal temperature gradient and mitigate hot spots.
A deeper insight into the flow distribution can be gained by examining the velocity fields. In the conventional module, the coolant velocity in the gaps between cells 4 and 8 was almost zero, leading to what I call “dead zones” where heat is removed only by natural convection or conduction through the coolant. In the novel module, these dead zones are eliminated. The baffles create a serpentine-like flow path, so the coolant is forced to travel alternately over the battery surfaces. This increases the velocity in the gaps by an order of magnitude, as observed in the simulations. The velocity distribution becomes much more uniform, and the coolant effectively sweeps the entire surface of each battery. The pressure drop penalty is a result of the reduced flow area and the increased path length, but it is acceptable when considering the substantial thermal benefits.
Regarding the parameter study, I also noted that the effect of flow rate is more pronounced at lower flow rates. For instance, going from 1 to 2 L/min yields a larger temperature reduction than going from 4 to 5 L/min. This suggests that there is diminishing returns with increasing flow rate. Therefore, for the design of energy storage battery cooling systems, one should choose a flow rate that provides adequate cooling without excessively increasing the pumping power. In my novel module, a flow rate of 3 L/min seems to be a good trade-off, as it already brings the maximum temperature difference below 1.2 °C, which is well within the recommended limit for lithium-ion batteries (typically less than 5 °C). Higher flow rates may only be needed for extreme fast charging or high ambient temperatures.
The initial temperature study reveals a different trade-off. If the coolant is too cold, the entire module operates at low temperatures, which increases the internal resistance of the cells and reduces the available capacity. It also creates a large temperature gradient, which is detrimental to cycle life. On the other hand, if the coolant is too warm, the absolute temperature may exceed the safe threshold. The optimal initial coolant temperature for a given application depends on the ambient conditions and the desired performance. In many practical energy storage battery containers, the coolant temperature is regulated by a chiller or a heat exchanger. My results suggest that setting the initial coolant temperature to around 20 °C provides a good balance: the hottest cell average temperature remains below 25 °C, and the maximum temperature difference is only 2.35 °C at the end of discharge.
I also compared my pressure drop results with a similar study from literature. In that study, the pressure drop increased by 7.4 times when the flow rate was increased over the same range, while my novel module’s pressure drop increased by only 4.7 times. This indicates that the baffle design not only improves heat transfer but also helps to manage the pressure drop growth rate. The reason is that the baffles convert some of the kinetic energy of the flow into useful cross-flow, rather than creating large recirculation zones with high viscous dissipation. However, it is important to note that the absolute pressure drop in the novel module is still higher than in the conventional module at the same flow rate, as mentioned earlier. The overall performance should be evaluated using a thermal-hydraulic efficiency factor, which considers both heat transfer and pressure drop. My calculations show that the new design has a higher efficiency factor than the conventional design, meaning that the additional pressure drop is justified by the increased heat transfer.
To further analyze the performance, I computed the average Nusselt number for each condition and used it to derive the correlation. The Nusselt number represents the ratio of convective to conductive heat transfer at the boundary. For the novel energy storage battery module, the Nusselt number ranged from about 40 to 60 in the studied parameter space. The correlation captures the trends correctly: Nu increases with Reynolds number and decreases with the dimensionless initial temperature ratio. The coefficient of determination (R²) was 0.98, indicating an excellent fit. The average relative error of 2% is very low for engineering purposes, and the maximum error of 5.4% occurs at the edge of the parameter range, which is acceptable.
The development of this correlation is beneficial for the design and optimization of immersion-cooled energy storage battery systems. It can be used to quickly estimate the heat transfer coefficient for different operating conditions, which can then be integrated into lumped thermal models or system-level simulations. This avoids the need for detailed CFD simulations during the initial design phase. Furthermore, the correlation can be used to optimize the coolant flow rate and initial temperature setpoints in real-time for a battery energy storage system, ensuring that the battery module operates within the desired temperature range while minimizing pumping energy.
In summary, my study demonstrates that the novel immersion-cooled energy storage battery module with internal baffles significantly improves flow and heat transfer characteristics compared to the traditional design. The baffles enhance coolant flow uniformity, reduce cell temperatures, improve temperature uniformity, and mitigate hot spots, at the cost of a 30% increase in pressure drop. Increasing the coolant flow rate improves both heat transfer and temperature uniformity, but with diminishing returns and a nonlinear increase in pressure loss. Raising the initial coolant temperature reduces heat transfer performance but improves temperature uniformity, requiring a trade-off. The proposed Nusselt number correlation accurately predicts the heat transfer performance within the investigated range, offering a valuable tool for engineering design and control of energy storage battery thermal management systems.
These findings provide useful insights for the continued development of high-performance and safe energy storage battery technologies. With the rapid growth of renewable energy integration and the need for large-scale energy storage, immersion cooling combined with advanced module designs such as the one I proposed can significantly enhance the reliability and durability of battery systems. In the future, I plan to extend this work by considering transient load profiles that represent real grid applications, such as frequency regulation and peak shaving, and to optimize the baffle geometry further using machine learning techniques. This will help to realize the full potential of immersion cooling for energy storage battery systems in practical engineering applications.