In the context of global energy transition and the pursuit of carbon neutrality, energy storage systems have emerged as a critical component in modern power grids and renewable energy integration. Among various energy storage technologies, lithium-ion energy storage cells are widely adopted due to their high energy density, long cycle life, and flexible operation. However, the thermal management of these energy storage cells during charging and discharging processes poses a significant challenge, as excessive temperature rise or uneven temperature distribution can lead to performance degradation, safety hazards, and reduced lifespan. Therefore, developing efficient thermal management systems is paramount for ensuring the reliability and safety of energy storage systems.
As a researcher focused on thermal management for energy storage applications, I have extensively studied liquid cooling systems due to their superior heat transfer capabilities compared to air cooling. Liquid cooling leverages the high specific heat capacity and thermal conductivity of coolants to rapidly dissipate heat, thereby controlling the maximum temperature and improving temperature uniformity within energy storage cell modules. In this article, I present a comprehensive analysis and optimization of a liquid cooling system designed for energy storage cells, evaluating its performance under both constant-current and real-world operating conditions such as peak shaving and frequency regulation.
To begin, I established a thermal model for a prismatic lithium iron phosphate (LFP) energy storage cell with a rated capacity of 20 Ah. The cell dimensions are 148 mm × 26 mm × 92 mm. Using finite element simulation software, I modeled the cell and its module consisting of 18 cells in a 3-series 6-parallel configuration. The thermal behavior of the energy storage cell is governed by the heat generation rate, which can be expressed using Bernardi’s model:
$$ q = \frac{1}{V} \left[ I (U_0 – U) – I T \frac{dU_0}{dT} \right] $$
where \( q \) is the volumetric heat generation rate (W/m³), \( V \) is the cell volume, \( I \) is the current, \( U_0 \) is the open-circuit voltage, \( U \) is the terminal voltage, \( T \) is the temperature, and \( \frac{dU_0}{dT} \) is the temperature coefficient. For simulation purposes, I assumed uniform heat generation and isotropic material properties within the energy storage cell. The transient heat conduction equation for the cell is:
$$ \rho_b c_b \frac{\partial T_b}{\partial t} = \nabla \cdot (k_b \nabla T_b) + q $$
Here, \( \rho_b \), \( c_b \), and \( k_b \) denote the density, specific heat capacity, and thermal conductivity of the energy storage cell, respectively. The cooling system involves aluminum cold plates with micro-channels through which coolant (water) flows. The governing equations for the coolant include the continuity, momentum, and energy equations:
$$ \nabla \cdot \mathbf{u}_w = 0 $$
$$ \rho_w (\mathbf{u}_w \cdot \nabla) \mathbf{u}_w = -\nabla p + \mu_w \nabla^2 \mathbf{u}_w $$
$$ \rho_w c_w \frac{\partial T_w}{\partial t} + \rho_w c_w \mathbf{u}_w \cdot \nabla T_w = \nabla \cdot (k_w \nabla T_w) $$
where the subscript \( w \) refers to the coolant. The interface between the cold plate and coolant involves convective heat transfer, described by Newton’s law of cooling:
$$ -k_c \left( \frac{\partial T}{\partial n} \right) = h (T_w – T_c) $$
with \( h \) as the convective heat transfer coefficient. To validate the thermal model, I compared simulation results with experimental data under a 0.5 C constant-current charge followed by constant-voltage charge. The root mean square error (RMSE) was calculated as:
$$ \text{RMSE} = \sqrt{ \frac{1}{N} \sum_{i=1}^{N} (y_i – \hat{y}_i)^2 } $$
where \( y_i \) and \( \hat{y}_i \) are the experimental and simulated maximum temperatures, respectively. The RMSE was found to be 0.15°C, confirming the model’s accuracy.
I designed two liquid cooling configurations for the energy storage cell module: Design 1 with longitudinally placed cold plates and Design 2 with transversely placed cold plates. Each cold plate contains multiple micro-channels. The cooling performance was evaluated based on the maximum temperature (\( T_{\text{max}} \)) and maximum temperature difference (\(\Delta T_{\text{max}}\)) within the energy storage cell module under constant-current charge rates of 1 C and 3 C. The initial temperature and coolant temperature were set to 25°C, with a coolant flow rate of 1 cm³/s. The results are summarized in Table 1.
| Design | Charge Rate | \( T_{\text{max}} \) (°C) | \(\Delta T_{\text{max}} \) (°C) |
|---|---|---|---|
| Design 1 (Longitudinal) | 1 C | 34.93 | 2.44 |
| 3 C | 58.82 | 5.01 | |
| Design 2 (Transverse) | 1 C | 37.91 | 4.10 |
| 3 C | 52.58 | 8.54 |
Design 1 demonstrated superior cooling performance, as it better accommodates the anisotropic thermal conductivity of the energy storage cell. Specifically, the thermal conductivity in the Y-direction (through-plane) is significantly lower than in the X and Z directions, making transverse placement less effective at dissipating heat. Thus, I selected Design 1 for further parameter optimization.
Under constant-current conditions, I investigated the effects of coolant flow rate, coolant temperature, number of cooling channels, and inlet direction on the thermal performance of the liquid cooling system for the energy storage cell module. The goal was to minimize \( T_{\text{max}} \) and \(\Delta T_{\text{max}} \). For each parameter, I conducted simulations at charge rates of 1 C, 2 C, and 3 C, with other factors fixed at baseline values unless varied. The findings are presented below.
Coolant Flow Rate: I tested flow rates of 1, 2, and 3 cm³/s. As the flow rate increased, \( T_{\text{max}} \) decreased, but the improvement diminished at higher rates. For instance, at 3 C, increasing the flow rate from 1 to 2 cm³/s reduced \( T_{\text{max}} \) by 5.55%, whereas from 2 to 3 cm³/s, the reduction was only 1.71%. \(\Delta T_{\text{max}} \) showed negligible changes. A flow rate of 2 cm³/s was deemed optimal considering both cooling efficiency and energy consumption.
Coolant Temperature: Coolant inlet temperatures of 20°C, 25°C, and 30°C were evaluated. Lower coolant temperatures significantly reduced \( T_{\text{max}} \), with approximately a 3.5°C drop per 5°C decrease, independent of charge rate. \(\Delta T_{\text{max}} \) was only slightly affected. Given the trade-off with cooling power, 20°C was selected as the optimal coolant temperature.
Number of Cooling Channels: I varied the number of channels per cold plate to 5, 7, and 9, while maintaining a constant cross-sectional area per channel. Increasing channels from 5 to 7 notably lowered \( T_{\text{max}} \) (e.g., 10.95% reduction at 3 C), but further increase to 9 yielded marginal benefits (3.38% reduction). Interestingly, \(\Delta T_{\text{max}} \) slightly increased with more channels, likely due to flow distribution effects. Thus, 7 channels per cold plate were chosen.
Inlet Direction: To mitigate temperature unevenness caused by coolant warming along the flow path, I compared a single-inlet design (all inlets on one side) with a staggered-inlet design (inlets on opposite sides). The staggered design slightly reduced \( T_{\text{max}} \) (up to 0.18°C at 3 C) and more appreciably lowered \(\Delta T_{\text{max}} \) (up to 0.42°C at 3 C), enhancing temperature uniformity across the energy storage cell module.
The optimal parameter combination derived from constant-current simulations is: coolant flow rate of 2 cm³/s, coolant temperature of 20°C, 7 cooling channels per cold plate, and staggered inlet configuration. Under these conditions, the energy storage cell module exhibits the following thermal performance:
| Charge Rate | \( T_{\text{max}} \) (°C) | \(\Delta T_{\text{max}} \) (°C) |
|---|---|---|
| 1 C | 29.58 | 2.22 |
| 2 C | 39.28 | 3.60 |
| 3 C | 46.11 | 4.56 |
To validate the practicality of this optimized liquid cooling system, I applied it to real-world operating conditions typical of energy storage systems engaged in peak shaving and frequency regulation. These scenarios involve dynamic charge-discharge profiles with rapidly changing currents, as illustrated in Figure 12 of the original text. I simulated both frequency regulation (frequent switching between charge and discharge every ~60 s) and peak shaving (sustained charge or discharge periods) profiles. The results were compared against natural convection cooling to quantify the improvement offered by liquid cooling for the energy storage cell module.
Under frequency regulation, the liquid cooling system reduced \( T_{\text{max}} \) by 5.28°C (10.74% reduction) and \(\Delta T_{\text{max}} \) by 2.79°C (44.1% reduction) compared to natural convection. For peak shaving, the reductions were 5.13°C (10.86%) in \( T_{\text{max}} \) and 1.91°C (37.45%) in \(\Delta T_{\text{max}} \). These results underscore the effectiveness of the liquid cooling system in maintaining the energy storage cell module within safe temperature limits during demanding grid services.
Building on this, I proposed a variable flow rate strategy to further optimize energy efficiency without compromising cooling performance. Since the current profiles in real-world operations are not constant, maintaining a fixed coolant flow rate may lead to unnecessary pumping power consumption. The variable strategy adjusts the flow rate based on the instantaneous charge-discharge rate: lowering it to 0.5 cm³/s during low-rate periods and raising it to 2.5 cm³/s during high-rate periods, compared to the baseline constant flow rate of 2 cm³/s.
Applying this strategy to the peak shaving profile, I observed that the variable flow rate reduced the overall coolant consumption by 10.3% (from 49,846.15 cm³ to 44,707.52 cm³) while slightly improving temperature control. Specifically, \( T_{\text{max}} \) decreased by 2.27°C (5.39% reduction) relative to the constant flow rate case, and \(\Delta T_{\text{max}} \) remained nearly unchanged (3.27°C vs. 3.19°C). This demonstrates that adaptive flow control can enhance the thermal management efficiency of liquid cooling systems for energy storage cells, aligning with energy-saving goals in practical deployments.
In conclusion, my investigation into liquid cooling systems for energy storage cells has yielded several key insights. First, the orientation of cold plates significantly impacts cooling performance; longitudinal placement outperforms transverse placement due to the anisotropic thermal properties of energy storage cells. Second, systematic parameter optimization under constant-current conditions reveals that a coolant flow rate of 2 cm³/s, a coolant temperature of 20°C, 7 cooling channels per plate, and staggered inlets collectively minimize both maximum temperature and temperature differences. Third, when applied to real-world peak shaving and frequency regulation duties, this optimized liquid cooling system substantially improves thermal management compared to natural convection, ensuring the safety and longevity of energy storage cells. Finally, implementing a variable flow rate strategy further boosts energy efficiency without sacrificing cooling efficacy, offering a pragmatic approach for dynamic operational environments.
The mathematical models and simulation frameworks developed here provide a foundation for designing and evaluating liquid cooling systems tailored to specific energy storage cell configurations and duty cycles. Future work could explore advanced coolants, hybrid cooling methods, or machine learning-based control algorithms to achieve even greater performance. As energy storage systems continue to expand their role in grid stabilization and renewable integration, robust thermal management solutions like the one presented here will be indispensable for harnessing the full potential of energy storage cells.