Lithium-ion energy storage cells have become widely adopted in power systems due to their high energy density, long cycle life, and environmental friendliness. However, the performance, lifespan, and safety of energy storage cells are highly sensitive to temperature. Effective thermal management is therefore critical to ensure the reliable operation of energy storage systems. In the field of utility-scale energy storage, air cooling and liquid cooling are the two primary heat dissipation technologies. While air cooling offers simplicity and low cost, it suffers from slow thermal response, large temperature gradients, and limited cooling capacity, especially as energy density increases. Liquid cooling, using a water-glycol mixture as the coolant, provides superior heat transfer efficiency and lower energy consumption, enabling lower temperature rise and better temperature uniformity for energy storage cells. This has made liquid cooling the dominant research direction for high-capacity energy storage thermal management.
In this study, I focus on the design and simulation of a liquid-cooled energy storage cell module based on a multi-channel liquid-cooled plate. The objective is to develop a module that achieves excellent temperature control during high-rate charge/discharge operations. I present a comprehensive numerical analysis of the coolant flow field and battery temperature field under 1 P power conditions, followed by experimental validation.
The liquid-cooled plate is fabricated from extruded aluminum alloy with high thermal conductivity. It features multiple parallel channels to maximize the contact area between the coolant and the plate, thereby enhancing heat transfer. At the downstream end of the plate, cylindrical turbulence-dot structures are arranged to improve flow uniformity by disrupting the coolant stream. The energy storage cell module consists of a housing, 52 individual prismatic lithium iron phosphate energy storage cells (each rated at 280 Ah, 3.2 V), a battery management system slave unit, and the liquid-cooled plate. The cells are configured in a 1-parallel, 52-series arrangement, yielding a nominal capacity of 280 Ah and a nominal voltage of 166.4 V. Twelve temperature sensors are distributed within the module to monitor surface temperatures of the energy storage cells.
To analyze the thermal performance, I built a fluid‑solid coupled simulation model using a commercial CFD tool. The model includes the housing, all 52 energy storage cells, cell fixing endplates, and the bottom liquid-cooled plate. The computational domain is divided into a fluid region (coolant), solid regions (cells, endplates, liquid‑cooled plate, and piping), and an air region inside the module. The coolant is a 50/50 water‑ethylene glycol mixture. The inlet boundary condition is a velocity inlet with a flow rate of 10 L/min, and the outlet is set as a pressure outlet with fully developed flow. The ambient temperature and initial module temperature are 35 °C, and the coolant inlet temperature is 22 °C. Each energy storage cell is modeled as a uniform heat source with a heat generation rate of 13 W under 1 P operation.
The governing equations for the fluid flow are based on the continuity, momentum, and energy conservation laws. The turbulence is modeled using the standard k‑ε model. The flow equations are:
$$
\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{u}) = 0
$$
$$
\rho \frac{D\mathbf{u}}{Dt} = -\nabla p + \mu \nabla^2 \mathbf{u} + \rho \mathbf{g}
$$
$$
\rho c_p \frac{DT}{Dt} = k \nabla^2 T + \Phi
$$
where ρ is density, u is velocity vector, p is pressure, μ is dynamic viscosity, T is temperature, k is thermal conductivity, and Φ accounts for viscous dissipation and heat generation in the energy storage cells. The k‑ε turbulence equations are:
$$
\frac{\partial (\rho k)}{\partial t} + \nabla \cdot (\rho \mathbf{u} k) = \nabla \cdot \left( \left( \mu + \frac{\mu_t}{\sigma_k} \right) \nabla k \right) + G_k – \rho \varepsilon
$$
$$
\frac{\partial (\rho \varepsilon)}{\partial t} + \nabla \cdot (\rho \mathbf{u} \varepsilon) = \nabla \cdot \left( \left( \mu + \frac{\mu_t}{\sigma_\varepsilon} \right) \nabla \varepsilon \right) + C_{1\varepsilon} \frac{\varepsilon}{k} G_k – C_{2\varepsilon} \rho \frac{\varepsilon^2}{k}
$$
Here, μt is the turbulent viscosity, σk and σε are turbulent Prandtl numbers, and C1ε, C2ε are constants. The heat transfer within the solid domains is governed by the heat conduction equation:
$$
\nabla \cdot (k_s \nabla T_s) + q”’ = 0
$$
where ks is the solid thermal conductivity and q”’ is the volumetric heat generation rate for the energy storage cells. The conjugate heat transfer at the fluid-solid interfaces is handled by ensuring continuity of temperature and heat flux.
The simulation results for the flow field in the liquid-cooled plate indicate that the coolant enters the plate through the left inlet, divides into multiple channels, and flows uniformly toward the right outlet. The velocity distribution shows no recirculation zones, confirming the effectiveness of the turbulence-dot design in promoting uniform flow. The maximum coolant velocity is approximately 2.5 m/s within the channels, as summarized in Table 1.
| Parameter | Value |
|---|---|
| Inlet flow rate | 10 L/min |
| Maximum velocity in channels | 2.5 m/s |
| Average velocity | 1.8 m/s |
| Flow uniformity (standard deviation/mean) | 0.12 |
| Pressure drop across plate | 15 kPa |
The temperature field simulation under a continuous 1 P charge-discharge cycle reveals distinct thermal gradients within the energy storage cell module. The overall module temperature ranges from a minimum of 36.0 °C to a maximum of 49.0 °C. The temperature distribution shows that the lower portions of the energy storage cells near the liquid‑cooled plate are significantly cooler, while the upper regions are hotter. This vertical gradient is due to the direct contact between the cell bottoms and the cold plate, which efficiently removes heat via convection. In contrast, the cell tops rely solely on downward heat conduction, resulting in higher temperatures. Laterally, cells near the coolant inlet (left side) are slightly cooler than those near the outlet (right side) because the coolant temperature gradually increases as it absorbs heat along the flow path. However, the overall temperature uniformity is satisfactory, with a maximum cell‑to‑cell temperature difference of 3.0 °C among the monitored points.
In the simulation, I monitored twelve virtual temperature points placed on the top‑center surface of each energy storage cell group (four groups, three points per group). The temperature evolution over the 1 P cycle is plotted and key metrics are recorded in Table 2.
| Parameter | Value |
|---|---|
| Initial temperature | 35.0 °C |
| Maximum cell surface temperature | 49.0 °C |
| Minimum cell surface temperature | 46.0 °C |
| Maximum temperature rise | 14.0 °C |
| Maximum temperature difference between cells | 3.0 °C |
| Temperature uniformity (max-min)/mean | 0.063 |
To validate the simulation, I built a prototype liquid-cooled energy storage cell module and conducted charge‑discharge tests according to the standard GB/T 36276‑2023. The test setup included a bidirectional DC power supply (PSC900‑300) and a data acquisition system via the battery management system slave unit. The ambient temperature was controlled at 35 °C, and the coolant inlet temperature was maintained at 22 °C with a flow rate of 10 L/min. The module was subjected to a continuous 1 P charge‑discharge cycle between 2.80 V and 3.65 V.
The experimental results show that after the full cycle, the maximum energy storage cell surface temperature reaches 47.5 °C, the minimum is 44.3 °C, giving a maximum temperature rise of 12.5 °C and a maximum temperature difference of 3.2 °C. These values are in good agreement with the simulation predictions, confirming the accuracy of the numerical model. The slight discrepancies (approximately 1.5 °C lower in maximum temperature) can be attributed to simplifications in the model, such as uniform heat generation and perfect thermal contact assumptions. The measured temperature distribution is shown in Table 3.
| Parameter | Value |
|---|---|
| Initial temperature | 35.0 °C |
| Maximum cell surface temperature | 47.5 °C |
| Minimum cell surface temperature | 44.3 °C |
| Maximum temperature rise | 12.5 °C |
| Maximum temperature difference between cells | 3.2 °C |
| Average temperature rise rate | 0.017 °C/s |
| Heat dissipation power (calculated) | ≈ 676 W |
Further analysis of the temperature uniformity reveals that the liquid-cooled design effectively mitigates the hotspot issue common in air‑cooled energy storage cell modules. The Nusselt number for the coolant flow in the channels can be estimated using the Dittus‑Boelter correlation for turbulent flow:
$$
Nu = 0.023 \, Re^{0.8} \, Pr^{0.4}
$$
where the Reynolds number is calculated as:
$$
Re = \frac{\rho u D_h}{\mu}
$$
and Pr is the Prandtl number. For the given conditions (Re ≈ 5000 – 8000, Pr ≈ 7), the Nusselt number is approximately 40–50, indicating strong convective heat transfer. The overall heat transfer coefficient U between the energy storage cell surface and the coolant can be expressed as:
$$
\frac{1}{U} = \frac{1}{h_c} + \frac{t_p}{k_p} + \frac{1}{h_{gap}}
$$
where hc is the convective coefficient on the coolant side, tp and kp are the thickness and thermal conductivity of the liquid‑cooled plate wall, and hgap accounts for the thermal contact resistance between the cell and the plate. Using typical values, U is found to be around 800 W/(m²·K), which is an order of magnitude higher than an air‑cooled interface.
The temperature uniformity can be quantified by the dimensionless temperature difference:
$$
\Delta T^* = \frac{T_{max} – T_{min}}{T_{avg} – T_{coolant,in}}
$$
For the simulated case, ΔT* ≈ 0.09, while the experimental value is 0.10. Both values are low, demonstrating excellent thermal management capability. The energy storage cell module’s temperature rise curve follows an exponential approach to steady state, as described by:
$$
T(t) = T_{initial} + (T_{steady} – T_{initial})\left(1 – e^{-t/\tau}\right)
$$
where τ is the thermal time constant. From the experimental data, τ is approximately 1200 s, indicating a relatively fast thermal response.
To further investigate the sensitivity of the design, I performed parametric studies on coolant flow rate and channel geometry. Table 4 summarizes the effect of varying the flow rate on the maximum temperature and temperature difference at steady state.
| Flow rate (L/min) | Max cell temperature (°C) | Max temperature difference (°C) | Pressure drop (kPa) |
|---|---|---|---|
| 6 | 51.2 | 4.5 | 8 |
| 8 | 49.8 | 3.8 | 12 |
| 10 | 49.0 | 3.0 | 15 |
| 12 | 48.5 | 2.6 | 20 |
| 14 | 48.1 | 2.3 | 26 |
As the flow rate increases, the maximum temperature and temperature difference decrease, but the pressure drop rises nonlinearly. A flow rate of 10 L/min offers a good balance between thermal performance and pumping power. The channel width and height also influence the heat transfer. A wider channel reduces flow velocity and heat transfer coefficient, while a narrower channel increases pressure drop. The optimal channel aspect ratio (height/width) is found to be around 0.5 for the given dimensions.
In addition, I examined the impact of the turbulence-dot array. Removing the dots increases the maximum temperature difference to 4.1 °C and worsens flow maldistribution. The dots create local mixing that enhances heat transfer near the downstream region, compensating for the coolant temperature rise. This design feature is critical for maintaining uniform cooling across all energy storage cells.
From the experimental validation, the prototype module demonstrates reliable thermal performance. The cycle life of energy storage cells is known to degrade significantly when operating above 45 °C or with large temperature gradients. Our module maintains the maximum cell temperature below 48 °C and the temperature difference below 4 °C under continuous 1 P operation, which is well within the recommended operating range for lithium iron phosphate energy storage cells. This is a marked improvement over typical air‑cooled modules, which often experience temperature rises exceeding 20 °C and differences greater than 10 °C under similar conditions.
The simulation model also enables prediction of the module performance under partial load conditions. For example, at 0.5 P power, the heat generation per energy storage cell is 6.5 W, and the maximum temperature rise is only 6 °C with a temperature difference of 1.5 °C. Conversely, at 1.5 P overload conditions (19.5 W per cell), the maximum temperature reaches 56 °C and the difference increases to 5 °C, indicating the need for higher flow rates or more advanced cooling strategies for extreme duty cycles.
In summary, the proposed liquid-cooled energy storage cell module based on a multi‑channel cold plate with turbulence‑dot structures provides outstanding thermal management. The simulation and experimental results demonstrate maximum temperature rise of 12.5 °C and maximum temperature difference of 3.2 °C under nominal 1 P operation. The design ensures that all energy storage cells operate within a narrow temperature window, thereby enhancing the longevity and safety of the overall energy storage system. This work offers a reliable reference for the thermal design of high‑capacity liquid‑cooled energy storage systems.
The key contributions of this research include:
- A detailed fluid‑solid coupled model that accurately predicts the temperature field of the energy storage cell module.
- Quantitative evidence of the effectiveness of the multi‑channel liquid‑cooled plate with turbulence dots.
- Experimental confirmation of the simulation, with less than 2 °C deviation in maximum temperature.
- A parametric analysis that guides the selection of optimal coolant flow rates and channel geometries for future designs.
Future work will focus on integrating the liquid‑cooled module into a full‑scale energy storage system, evaluating the impact of varying ambient temperatures, and exploring novel coolants such as dielectric fluids for enhanced safety. Additionally, transient thermal behavior during fast‑charge events (2 P or higher) will be investigated to push the boundaries of liquid‑cooled thermal management for next‑generation energy storage cells.