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 based energy storage cells are widely adopted due to their high energy density, power density, efficiency, and modular scalability. However, the thermal management of these energy storage cells during operation is paramount to ensure safety, longevity, and performance. Inadequate temperature control can lead to thermal runaway, catastrophic failures, and significant safety hazards, as evidenced by numerous incidents in energy storage facilities. Therefore, developing an effective battery thermal management system (BTMS) is essential. Liquid cooling, with its superior heat transfer capabilities compared to air cooling, offers a promising solution for managing the thermal behavior of high-power energy storage cells. This article, from my research perspective, delves into the optimization design and analysis of a liquid cooling system tailored for energy storage cells under various operational scenarios. The focus is on enhancing temperature uniformity and controlling peak temperatures within energy storage cell modules through parametric studies and structural innovations.
The core of this work involves the design and simulation of liquid cooling configurations for a module comprising multiple energy storage cells. I begin by establishing a comprehensive three-dimensional model using finite element analysis software, COMSOL Multiphysics, to simulate the thermal and fluid dynamics. The energy storage cell considered is a prismatic lithium iron phosphate (LFP) cell, which is common in stationary energy storage applications. The module consists of 18 such cells arranged in a 3-series, 6-parallel configuration. Two primary liquid cooling plate layouts are conceptualized: a longitudinal placement (Design 1) where cooling plates are oriented along the length of the cells, and a transverse placement (Design 2) where plates are placed across the width. The fundamental geometric parameters are summarized in Table 1.
| Component | Dimensions (mm) | Material | Remarks |
|---|---|---|---|
| Single Energy Storage Cell | 148 × 26 × 92 | Lithium Iron Phosphate | Simplified as a homogeneous block for simulation |
| Cooling Plate (Design 1 & 2) | 221 × 6 × 92 | Aluminum | Each plate contains multiple microchannels |
| Cooling Microchannel | 5 × 5 (cross-section) | – | Spacing varies with channel count |
| Module Assembly | – | – | 18 cells, 3S6P, with integrated cooling plates |
The thermal behavior of an energy storage cell is governed by its internal heat generation, which is primarily influenced by electrochemical reactions and ohmic losses. To accurately simulate this, I employ the widely accepted Bernardi model for volumetric heat generation rate. The heat generation rate \( q \) for an energy storage cell is expressed as:
$$q = \frac{1}{V} \left[ I (U_0 – U) – I T \frac{dU_0}{dT} \right]$$
However, for constant current operations, it is often simplified to:
$$q = \frac{I}{V} \left[ (U_0 – U) – T \frac{dU_0}{dT} \right]$$
where \( V \) is the volume of the energy storage cell, \( I \) is the current, \( U_0 \) is the open-circuit voltage, \( U \) is the terminal voltage, \( T \) is the absolute temperature, and \( \frac{dU_0}{dT} \) is the temperature coefficient of the open-circuit voltage. This model accounts for both irreversible (ohmic and polarization) and reversible (entropic) heat effects. The governing energy conservation equation for the solid domain of the energy storage cell is given by:
$$\rho_b c_{p,b} \frac{\partial T_b}{\partial t} = \nabla \cdot (k_b \nabla T_b) + q$$
Here, \( \rho_b \), \( c_{p,b} \), and \( k_b \) represent the density, specific heat capacity, and thermal conductivity tensor of the energy storage cell, respectively. The thermal conductivity is anisotropic for typical laminated battery structures. For the aluminum cooling plate, which does not generate heat, the energy equation simplifies to:
$$\rho_c c_{c} \frac{\partial T_c}{\partial t} = \nabla \cdot (k_c \nabla T_c)$$
The cooling fluid, assumed to be water, flows through the microchannels. Its behavior is described by the incompressible Navier-Stokes equations and energy equation. The continuity and momentum equations are:
$$\nabla \cdot \mathbf{u}_w = 0$$
$$\rho_w (\mathbf{u}_w \cdot \nabla) \mathbf{u}_w = -\nabla p + \mu_w \nabla^2 \mathbf{u}_w$$
The energy conservation for the coolant is:
$$\rho_w c_{w} \left( \frac{\partial T_w}{\partial t} + \mathbf{u}_w \cdot \nabla T_w \right) = \nabla \cdot (k_w \nabla T_w)$$
The convective heat transfer between the cooling plate walls and the coolant is modeled using Newton’s law of cooling:
$$-k_c \left( \frac{\partial T}{\partial n} \right)_w = h (T_w – T_f)$$
where \( h \) is the convective heat transfer coefficient, \( T_w \) is the wall temperature, and \( T_f \) is the local coolant temperature. The boundary conditions for the simulations include an initial temperature of 25°C for both the energy storage cell module and the coolant, natural convection on external surfaces with a coefficient of 5 W/(m²·K), and no-slip conditions at channel walls. The coolant inlet is specified with a fully developed flow profile at a prescribed velocity and temperature, while the outlet is set to atmospheric pressure.
Before proceeding to system-level analysis, I validated the thermal model of a single energy storage cell against experimental data under a 0.5C constant current-constant voltage charge protocol. The root mean square error (RMSE) between simulation and experiment was calculated as:
$$\text{RMSE} = \sqrt{ \frac{1}{N} \sum_{i=1}^{N} (y_i – \hat{y}_i)^2 }$$
where \( y_i \) are experimental temperature values and \( \hat{y}_i \) are simulated values. The RMSE was found to be 0.15°C, confirming the model’s accuracy. A mesh independence study was also conducted for both the single cell and the full module with cooling structures. Three mesh sizes (coarse, fine, and finer) were tested, and the fine mesh was selected as it provided a balance between computational cost and result accuracy, with temperature variations less than 0.1°C compared to the finer mesh.
The first major investigation concerns the comparative performance of the two liquid cooling plate layouts: longitudinal (Design 1) and transverse (Design 2). The cooling performance is 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 conditions at 1C and 3C rates. The coolant flow rate is set to 1 cm³/s, temperature to 25°C, and each cooling plate has 7 channels. The simulation results are summarized in Table 2.
| Charge Rate | Design | \( T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) (°C) | Improvement in \( T_{\text{max}} \) (Design 1 vs 2) | Improvement in \( \Delta T_{\text{max}} \) (Design 1 vs 2) |
|---|---|---|---|---|---|
| 1C | Design 1 (Longitudinal) | 34.93 | 2.44 | 8.53% lower | 67.21% lower |
| Design 2 (Transverse) | 37.91 | 4.10 | |||
| 3C | Design 1 (Longitudinal) | 52.58 | 5.01 | 11.86% lower | 70.46% lower |
| Design 2 (Transverse) | 58.82 | 8.54 |
The longitudinal design (Design 1) consistently outperforms the transverse design. This is attributed to the anisotropic thermal conductivity of the energy storage cell. Typically, the through-plane thermal conductivity (direction through the electrode layers) is significantly lower than the in-plane conductivities. In Design 1, the cooling plates are aligned with the longer dimension of the cells, allowing heat to be extracted more effectively along the paths of higher thermal conductivity. In contrast, Design 2 places plates perpendicular to this direction, creating a thermal bottleneck that impedes efficient heat removal from the core of each energy storage cell. Consequently, Design 1 exhibits lower maximum temperatures and, more notably, superior temperature uniformity (smaller \( \Delta T_{\text{max}} \)). The temperature distribution contours at the end of charge clearly show a more uniform profile for Design 1, whereas Design 2 exhibits a pronounced temperature gradient from the coolant inlet side to the outlet side. Given its superior performance, Design 1 is selected as the baseline for all subsequent parametric optimization studies aimed at further enhancing the cooling of the energy storage cell module.
Having established the superior layout, I now focus on optimizing the liquid cooling system parameters under constant current conditions to achieve optimal thermal management for the energy storage cell module. The key parameters investigated are coolant flow rate, coolant inlet temperature, number of cooling channels per plate, and coolant inlet configuration. The performance metrics remain the module’s \( T_{\text{max}} \) and \( \Delta T_{\text{max}} \).
1. Influence of Coolant Flow Rate: The flow rate directly affects the convective heat transfer coefficient and the coolant’s temperature rise along the channels. Simulations are conducted for flow rates of 1, 2, and 3 cm³/s at 1C, 2C, and 3C charge rates. The coolant temperature is fixed at 25°C. The results are quantified in Table 3.
| Charge Rate | Flow Rate (cm³/s) | \( T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) (°C) | Reduction in \( T_{\text{max}} \) from previous rate |
|---|---|---|---|---|
| 1C | 1.0 | 33.05 | 2.31 | – |
| 2.0 | 31.27 | 2.33 | 5.38% | |
| 3.0 | 31.01 | 2.35 | 0.85% | |
| 2C | 1.0 | 42.79 | 3.55 | – |
| 2.0 | 40.86 | 3.58 | 4.52% | |
| 3.0 | 40.30 | 3.61 | 1.36% | |
| 3C | 1.0 | 49.66 | 4.53 | – |
| 2.0 | 46.90 | 4.57 | 5.55% | |
| 3.0 | 46.10 | 4.60 | 1.71% |
The data indicates that increasing the flow rate from 1 to 2 cm³/s yields a significant reduction in \( T_{\text{max}} \) (4.5-5.5%), but the marginal benefit diminishes sharply when increasing from 2 to 3 cm³/s (less than 2%). This is due to the logarithmic relationship between flow rate and convective heat transfer coefficient; beyond a certain point, increased pumping power does not translate to proportional cooling improvement. Notably, \( \Delta T_{\text{max}} \) increases slightly with flow rate because a higher flow rate reduces the coolant temperature more at the inlet region than at the outlet, slightly exacerbating the axial temperature gradient. Balancing cooling performance and pumping power, a flow rate of 2 cm³/s is deemed optimal for the energy storage cell module.
2. Influence of Coolant Inlet Temperature: The coolant temperature sets the baseline for heat rejection. Simulations are run with inlet temperatures of 20°C, 25°C, and 30°C, with a fixed flow rate of 2 cm³/s. The results are consolidated in Table 4.
| Charge Rate | Coolant Inlet Temp (°C) | \( T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) reduction for 10°C coolant temp drop |
|---|---|---|---|---|
| 1C | 20 | 29.58 | 2.22 | 0.57°C |
| 25 | 33.05 | 2.31 | ||
| 30 | 36.49 | 2.79 | ||
| 2C | 20 | 39.28 | 3.60 | 0.91°C |
| 25 | 42.79 | 3.55 | ||
| 30 | 46.39 | 4.51 | ||
| 3C | 20 | 46.11 | 4.56 | 1.22°C |
| 25 | 49.66 | 4.53 | ||
| 30 | 53.34 | 5.78 |
Lowering the coolant temperature effectively reduces \( T_{\text{max}} \) by approximately 3.5°C for every 5°C drop, irrespective of the charge rate. This linear relationship highlights the direct impact of the heat sink temperature. However, the effect on \( \Delta T_{\text{max}} \) is more complex and generally smaller. A 10°C reduction in coolant temperature reduces \( \Delta T_{\text{max}} \) by 0.57°C to 1.22°C, depending on the rate. While lower coolant temperatures are beneficial, they require additional energy for chilling. Considering typical operational constraints and efficiency, a coolant inlet temperature of 20°C is selected as optimal for maintaining the energy storage cell module within a safe temperature range.
3. Influence of Number of Cooling Channels: The number of channels affects the contact area between the coolant and the plate, as well as the hydraulic diameter. Designs with 5, 7, and 9 channels per plate are analyzed, keeping the total cross-sectional area per plate approximately constant by adjusting channel spacing. The flow rate per plate is maintained at 2 cm³/s, and coolant temperature is 20°C. The findings are presented in Table 5.
| Charge Rate | Channel Count | Channel Spacing (mm) | \( T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) (°C) | Reduction in \( T_{\text{max}} \) from 5 to 7 channels | Reduction in \( T_{\text{max}} \) from 7 to 9 channels |
|---|---|---|---|---|---|---|
| 1C | 5 | 20 | 32.58 | 2.15 | 9.23% | 3.00% |
| 7 | 15 | 29.58 | 2.22 | |||
| 9 | 10 | 28.69 | 2.28 | |||
| 2C | 5 | 20 | 43.85 | 3.48 | 10.41% | 3.39% |
| 7 | 15 | 39.28 | 3.60 | |||
| 9 | 10 | 37.95 | 3.67 | |||
| 3C | 5 | 20 | 51.78 | 4.41 | 10.95% | 3.38% |
| 7 | 15 | 46.11 | 4.56 | |||
| 9 | 10 | 44.55 | 4.63 |
Increasing the channel count from 5 to 7 significantly improves cooling, reducing \( T_{\text{max}} \) by about 10%. This is due to the increased heat transfer area and reduced thermal resistance. However, increasing from 7 to 9 channels yields only about a 3% further reduction, indicating diminishing returns. Moreover, \( \Delta T_{\text{max}} \) slightly increases with more channels, likely because the closer channel spacing may lead to more localized cooling and complex flow distributions. Considering manufacturing complexity and the marginal gain, 7 channels per plate is chosen as the optimal design for the energy storage cell module cooling system.
4. Influence of Coolant Inlet Configuration: To address the temperature gradient along the flow direction, I propose and test an alternating inlet configuration. In the baseline design (same-side inlets), all cooling plate channels receive coolant from the same end. In the alternating design (opposite-side inlets), adjacent plates have inlets on opposite ends. This aims to create a more symmetric temperature field. The comparison is made at the optimal parameters identified so far: flow rate 2 cm³/s, coolant temp 20°C, 7 channels. Results are in Table 6.
| Charge Rate | Inlet Configuration | \( T_{\text{max}} \) (°C) | \( \Delta T_{\text{max}} \) (°C) | Reduction in \( \Delta T_{\text{max}} \) with Alternating Inlets |
|---|---|---|---|---|
| 1C | Same-side | 29.58 | 2.22 | 0.09°C (4.1%) |
| Alternating | 29.48 | 2.13 | ||
| 2C | Same-side | 39.28 | 3.60 | 0.29°C (8.1%) |
| Alternating | 39.15 | 3.31 | ||
| 3C | Same-side | 46.11 | 4.56 | 0.42°C (9.2%) |
| Alternating | 45.93 | 4.14 |
The alternating inlet configuration has a minimal effect on \( T_{\text{max}} \) (reduction less than 0.2°C) but provides a measurable improvement in temperature uniformity, reducing \( \Delta T_{\text{max}} \) by up to 9.2% at high rates. This is because the strategy counteracts the inherent coolant temperature rise along the flow path, leading to a more balanced heat extraction from the energy storage cell module. Therefore, the alternating inlet design is adopted as part of the optimized configuration.
Synthesizing the above parametric studies, the optimized liquid cooling system for the energy storage cell module under constant current conditions features: longitudinal cooling plates (Design 1), a coolant flow rate of 2 cm³/s per plate, a coolant inlet temperature of 20°C, 7 cooling channels per plate, and an alternating coolant inlet configuration. With this setup, the module’s thermal metrics at the end of charge are: for 1C, \( T_{\text{max}} = 29.48°C \), \( \Delta T_{\text{max}} = 2.13°C \); for 2C, \( T_{\text{max}} = 39.15°C \), \( \Delta T_{\text{max}} = 3.31°C \); for 3C, \( T_{\text{max}} = 45.93°C \), \( \Delta T_{\text{max}} = 4.14°C \).
While constant current analysis provides foundational insights, real-world energy storage systems for grid applications, such as peak shaving and frequency regulation, operate under dynamic and irregular current profiles. To validate the robustness of the optimized cooling design, I subject the energy storage cell module to two representative duty cycles derived from actual grid service data. The first is a frequency regulation (FR) profile characterized by frequent charge-discharge switches every ~60 seconds. The second is a peak shaving (PS) profile with sustained periods of charging or discharging interspersed with volatile periods. The initial module temperature is 25°C, and the cooling system operates with the optimized parameters (2 cm³/s, 20°C, etc.). For comparison, a natural convection scenario (no active cooling, only ambient air at 25°C with 5 W/(m²·K) convection) is also simulated.
The simulation results for the frequency regulation duty cycle are summarized in Table 7. The metrics reported are the maximum values of \( T_{\text{max}} \) and \( \Delta T_{\text{max}} \) observed over the entire cycle.
| Cooling Condition | Maximum \( T_{\text{max}} \) during Cycle (°C) | Maximum \( \Delta T_{\text{max}} \) during Cycle (°C) | Reduction vs. Natural Convection |
|---|---|---|---|
| Natural Convection | 49.18 | 6.33 | – |
| Optimized Liquid Cooling | 43.90 | 3.54 | \( T_{\text{max}} \): 10.74% lower, \( \Delta T_{\text{max}} \): 44.1% lower |
For the peak shaving duty cycle, the results are presented in Table 8.
| Cooling Condition | Maximum \( T_{\text{max}} \) during Cycle (°C) | Maximum \( \Delta T_{\text{max}} \) during Cycle (°C) | Reduction vs. Natural Convection |
|---|---|---|---|
| Natural Convection | 47.24 | 5.10 | – |
| Optimized Liquid Cooling | 42.11 | 3.19 | \( T_{\text{max}} \): 10.86% lower, \( \Delta T_{\text{max}} \): 37.45% lower |
The optimized liquid cooling system demonstrates substantial efficacy in real-world scenarios. It reduces the peak temperature of the energy storage cell module by over 10% and improves temperature uniformity by reducing the maximum temperature difference by 37-44% compared to passive cooling. This confirms the design’s capability to maintain the energy storage cell module within safer and more efficient operational bounds during demanding grid services.
While the constant-flow strategy is effective, it may not be energy-optimal for highly variable profiles. Therefore, I propose and evaluate a variable flow rate strategy tailored to the instantaneous demand on the energy storage cell module. The principle is to modulate the coolant flow rate in response to the module’s thermal load, approximated here by the absolute value of the operational current. A simple rule is implemented: for current magnitudes below a threshold (e.g., corresponding to 0.5C), the flow rate is reduced to 0.5 cm³/s; for high current magnitudes (above 2C), it is increased to 2.5 cm³/s; otherwise, it remains at the baseline 2 cm³/s. This strategy aims to save pumping energy during low-load periods while enhancing cooling during high-stress periods. The peak shaving duty cycle is used for this evaluation, comparing the variable flow strategy against the constant 2 cm³/s strategy. The results are detailed in Table 9.
| Flow Strategy | Maximum \( T_{\text{max}} \) (°C) | Maximum \( \Delta T_{\text{max}} \) (°C) | Total Coolant Volume Consumed per Cycle (cm³) | Change vs. Constant Flow |
|---|---|---|---|---|
| Constant Flow (2 cm³/s) | 42.11 | 3.19 | 49,846.15 | – |
| Variable Flow (0.5 to 2.5 cm³/s) | 39.84 | 3.27 | 44,707.52 | \( T_{\text{max}} \): 5.39% lower, Coolant Use: 10.3% lower |
The variable flow strategy achieves a further 5.39% reduction in the maximum temperature of the energy storage cell module while consuming 10.3% less coolant volume over the cycle. The \( \Delta T_{\text{max}} \) remains virtually unchanged, with a slight increase of only 0.08°C (2.51%). This demonstrates that an adaptive cooling control strategy can significantly enhance both thermal performance and system efficiency for energy storage cell modules operating under dynamic conditions.
In this comprehensive study, I have explored the design and optimization of a liquid cooling system for energy storage cell modules. Through detailed three-dimensional coupled thermal-fluid simulations, several key conclusions are drawn. First, the orientation of cooling plates relative to the anisotropic thermal conductivity of the energy storage cell is crucial. A longitudinal placement (Design 1) outperforms a transverse placement (Design 2), offering significantly lower maximum temperatures and superior temperature uniformity for the energy storage cell module. Second, parametric optimization under constant current conditions reveals that a coolant flow rate of 2 cm³/s, an inlet temperature of 20°C, 7 cooling channels per plate, and an alternating inlet configuration form an effective combination for balancing performance and energy expenditure. This optimized design reduces the energy storage cell module’s peak temperature and temperature difference substantially compared to natural convection. Third, when applied to realistic frequency regulation and peak shaving duty cycles, this liquid cooling system proves robust, maintaining the energy storage cell module within safer thermal limits. Finally, introducing a variable flow rate strategy that responds to the operational current further enhances thermal performance while reducing coolant consumption, showcasing a path toward more intelligent and efficient thermal management for energy storage systems. These findings contribute valuable insights for the engineering of advanced thermal management solutions that ensure the safety, reliability, and longevity of lithium-ion based energy storage cells in grid-scale applications.