The efficient and safe operation of modern energy storage systems is fundamentally tied to the performance of their core component: the li ion battery. As the global push for renewable energy integration intensifies, the role of large-scale, high-capacity li ion battery packs becomes increasingly critical. However, the electrochemical processes within a li ion battery during charge and discharge cycles are inherently exothermic. Managing this generated heat is not merely a matter of performance optimization; it is a paramount safety imperative. Excessive temperature or significant temperature gradients within a li ion battery pack can accelerate degradation, reduce usable capacity, and in extreme cases, trigger thermal runaway—a catastrophic failure mode. Therefore, the design and optimization of advanced thermal management systems (TMS) are essential for unlocking the full potential and ensuring the longevity of energy storage solutions based on li ion battery technology.
This work focuses on the numerical investigation and structural improvement of cooling systems for a large-format energy storage li ion battery pack. Utilizing computational fluid dynamics (CFD), we model and compare the thermal performance of different cooling architectures, aiming to identify solutions that maintain optimal operating temperatures and minimize temperature differentials across the pack.
1. Mathematical Modeling and Simulation Framework
The thermal behavior of a li ion battery pack is governed by complex interactions between internal heat generation and external heat dissipation. To accurately simulate this, we establish a coupled model encompassing the electrochemical heat source and the fluid-thermal dynamics of the cooling system.
1.1 Thermal Model of the Lithium-Ion Battery
We treat each individual li ion battery cell as a homogeneous solid with anisotropic thermal properties. The transient heat conduction within the cell is described by the Fourier’s law, yielding the following energy conservation equation:
$$
\rho C_p \frac{\partial T}{\partial t} = \lambda_x \frac{\partial^2 T}{\partial x^2} + \lambda_y \frac{\partial^2 T}{\partial y^2} + \lambda_z \frac{\partial^2 T}{\partial z^2} + \dot{q}_{gen}
$$
where $\rho$ is the density, $C_p$ is the specific heat capacity, $T$ is the temperature, $\lambda_x$, $\lambda_y$, $\lambda_z$ are the thermal conductivities in the three principal directions, and $\dot{q}_{gen}$ is the volumetric heat generation rate within the li ion battery.
The heat generation rate is a critical input. We employ the widely-used Bernardi equation, which accounts for irreversible Joule heating and reversible entropic heating:
$$
\dot{q}_{gen} = \frac{I}{V_b} \left[ (E_{ocv} – U) – T \frac{dE_{ocv}}{dT} \right]
$$
Here, $I$ is the current, $V_b$ is the cell volume, $E_{ocv}$ is the open-circuit voltage, and $U$ is the terminal voltage. For system-level simulation of a large li ion battery pack comprising many cells, the average heat generation rate calculated from this formula for a representative discharge cycle (e.g., 1C) is applied uniformly to each cell domain, providing a robust basis for comparative cooling analysis.
The material properties for a commercial 280 Ah lithium iron phosphate (LFP) li ion battery cell used in this study are summarized below.
| Property | Value | Unit |
|---|---|---|
| Chemistry | Lithium Iron Phosphate (LFP) | – |
| Nominal Capacity | 280 | Ah |
| Nominal Voltage | 3.2 | V |
| Density ($\rho$) | 2120 | kg/m³ |
| Specific Heat ($C_p$) | 3660 | J/(kg·K) |
| In-plane Thermal Conductivity ($\lambda_x$, $\lambda_y$) | 21.6 | W/(m·K) |
| Through-plane Thermal Conductivity ($\lambda_z$) | 2.11 | W/(m·K) |
| Internal Resistance | 0.43 | mΩ |
1.2 Fluid Dynamics and Cooling System Models
The cooling performance is evaluated for two primary systems: indirect liquid cooling and immersion cooling. Their governing equations are based on the principles of fluid mechanics and heat transfer.
For the coolant flow (a 50/50 water-ethylene glycol mixture in indirect cooling or dielectric fluid in immersion cooling), we model it as an incompressible, single-phase fluid. The governing equations are the continuity, momentum (Navier-Stokes), and energy equations:
$$
\nabla \cdot \vec{v} = 0
$$
$$
\rho_f \left( \frac{\partial \vec{v}}{\partial t} + \vec{v} \cdot \nabla \vec{v} \right) = -\nabla p + \mu \nabla^2 \vec{v}
$$
$$
\rho_f C_{p,f} \left( \frac{\partial T}{\partial t} + \vec{v} \cdot \nabla T \right) = k_f \nabla^2 T
$$
where $\rho_f$, $C_{p,f}$, $\mu$, and $k_f$ are the density, specific heat, dynamic viscosity, and thermal conductivity of the coolant, respectively; $\vec{v}$ is the velocity vector, and $p$ is the pressure. Turbulent flow in the channels is modeled using the standard $k$-$\epsilon$ turbulence model.
| Property (at 20°C) | Water-Glycol (50/50) | Dielectric Fluid | Unit |
|---|---|---|---|
| Density ($\rho_f$) | 1075 | 820 | kg/m³ |
| Thermal Conductivity ($k_f$) | 0.375 | 0.142 | W/(m·K) |
| Specific Heat ($C_{p,f}$) | 3251 | 2200 | J/(kg·K) |
| Kinematic Viscosity (at 40°C) | 4.8 | 10.5 | cSt |
1.3 Simulation Setup and Model Validation
All simulations are performed using the commercial CFD software STAR-CCM+. The li ion battery pack consists of 52 prismatic cells connected in series (1P52S). A polyhedral mesh is generated for the entire domain, with refined layers applied to critical regions like fluid boundaries, thermal interface materials (TIMs), and thin gaps to ensure accuracy.
A grid independence study is conducted to eliminate the influence of mesh density on the results. Five mesh configurations with increasing cell counts are evaluated for a key parameter like system pressure drop. The results converge within a 2% variation beyond a certain mesh size, confirming independence. The final selected mesh balances computational cost and precision.
To validate the thermal model of the li ion battery itself, a single-cell simulation is compared against experimental data. A 280 Ah LFP cell is discharged at a 1C rate from 100% to 30% State of Charge (SOC) in a controlled 25°C environment, with temperatures recorded at multiple surface locations. The simulation, using the uniform heat generation rate from the Bernardi equation, shows excellent agreement with the experimental measurements, with a maximum error below 5%. This validates the approach for extrapolating to full-pack simulations.
| Parameter | Value | Unit |
|---|---|---|
| Ambient Temperature | 25 | °C |
| Coolant Inlet Temperature | 25 | °C |
| Coolant Inlet Flow Rate | 12 | L/min |
| Discharge Rate | 1 | C |
| Discharge Duration | 2500 | s |
| External Convection Coefficient (Air) | 6 | W/(m²·K) |
2. Analysis of Conventional and Immersion Cooling Strategies
We begin by evaluating the baseline indirect liquid cooling system, which is prevalent in many current li ion battery pack designs, and then introduce a basic immersion cooling concept.
2.1 Baseline: Indirect Liquid Cooling System
The conventional system features an aluminum cold plate with a long, serpentine flow channel attached to the bottom of the pack enclosure. The li ion battery cells are mounted above the cold plate with a layer of thermally conductive gap filler (TIM) in between to reduce contact resistance. Heat generated in the cells conducts down through the TIM into the cold plate, where it is carried away by the flowing coolant.
Simulation results for the 1C discharge reveal significant thermal challenges. The maximum temperature ($T_{max}$) of the li ion battery pack reaches 38.28°C. More critically, the temperature distribution is highly non-uniform. The cells’ bottoms, in direct thermal proximity to the cold plate, are coolest, while the tops are significantly hotter. This results in a large overall pack temperature difference ($\Delta T_{pack}$) of 9.52°C and an even more concerning maximum cross-sectional temperature difference ($\Delta T_{section}$) of 9.3°C within the cells themselves. Such gradients induce uneven aging and stress within the li ion battery, compromising its lifecycle and safety margin.
2.2 Initial Immersion Cooling System Design
To address these limitations, an immersion cooling approach is modeled. In this design, the entire li ion battery pack is filled with a dielectric coolant. The cells are in direct contact with the fluid, which has a high boiling point and excellent electrical insulation properties. To reject heat from the dielectric fluid to an external loop, two internal cold plates (or heat exchangers) are integrated into the pack walls. A primary manifold distributes a secondary coolant (water-glycol) through these plates, chilling the surrounding dielectric fluid which then circulates via natural convection, absorbing heat directly from every cell surface.
The immersion system shows a marked improvement in single-cell temperature uniformity. The direct and omnidirectional contact greatly reduces the thermal resistance between the heat source (the li ion battery) and the coolant. The simulation shows a reduced $T_{max}$ of 35.59°C. While the overall $\Delta T_{pack}$ (9.43°C) is similar to the indirect system—mainly due to a temperature gradient along the flow path of the internal cold plates—the $\Delta T_{section}$ plummets to 4.17°C. This represents over a 55% reduction in the internal temperature gradient of the li ion battery cells, a crucial benefit for longevity. However, the pack-level gradient indicates room for structural optimization.
3. Structural Optimization of the Immersion Cooling System
The analysis of the initial two-plate immersion design identifies a key issue: the dielectric fluid near the inlet of the internal cold plates is much cooler, creating a lateral temperature gradient across the pack. To homogenize the temperature field, we propose a structural optimization focused on distributing the cooling power more evenly.
3.1 Optimized Design: Multi-Branch Cooling Plate Configuration
The core optimization involves increasing the number of internal cooling plates and arranging them in a parallel flow network. The original two-plate design is expanded to a system with five internal cooling plates. These plates are strategically positioned: one at each end of the pack and three spaced evenly along its length. All plates are fed in parallel from a common inlet manifold. This design drastically increases the effective heat exchange area inside the pack and ensures that every li ion battery cell is in close proximity to a cold surface, minimizing the path for heat diffusion through the dielectric fluid.
3.2 Performance of the Optimized System
The simulation of the optimized five-plate immersion system demonstrates transformative improvements in the thermal management of the li ion battery pack.
Temperature Uniformity: The most significant gain is in temperature homogeneity. The $T_{max}$ is further reduced to 30.69°C. Crucially, both the $\Delta T_{pack}$ and $\Delta T_{section}$ are brought under strict control. The overall pack temperature difference is reduced to 4.88°C, and the internal cell cross-sectional difference is minimized to 3.29°C. This represents a reduction of approximately 49% in pack gradient and 65% in cell sectional gradient compared to the traditional indirect cooling system. The li ion battery cells now operate in a much more uniform thermal environment.
System Hydraulics: The parallel configuration of the cooling plates also benefits the hydraulic performance. The flow resistance (pressure drop, $\Delta p$) of the internal cooling loop is dramatically lower. While the serpentine channel in the indirect cooling plate had a high $\Delta p$ of 66.1 kPa, the optimized parallel immersion system exhibits a very low $\Delta p$ of only 1.38 kPa. This translates to significantly lower pumping power requirements for the external coolant circuit, enhancing the overall energy efficiency of the thermal management system for the li ion battery pack.
| Performance Metric | Indirect Liquid Cooling | Immersion Cooling (2 Plates) | Immersion Cooling (5 Plates, Optimized) |
|---|---|---|---|
| Maximum Pack Temperature ($T_{max}$) | 38.28 °C | 35.59 °C | 30.69 °C |
| Maximum Pack Temperature Difference ($\Delta T_{pack}$) | 9.52 °C | 9.43 °C | 4.88 °C |
| Maximum Cell Sectional Temperature Difference ($\Delta T_{section}$) | 9.30 °C | 4.17 °C | 3.29 °C |
| System Pressure Drop ($\Delta p$) | 66.10 kPa | 3.50 kPa | 1.38 kPa |
4. Conclusion and Implications
This comprehensive numerical study systematically evaluates and optimizes cooling solutions for a large-scale energy storage li ion battery pack. The journey from a conventional indirect liquid cooling system to an optimized immersion cooling architecture reveals clear and compelling advantages for advanced thermal management.
The traditional bottom-cooled indirect system, while functional, struggles with inherent thermal gradients. The li ion battery cells experience a large top-to-bottom temperature difference due to the one-sided cooling approach and the cumulative thermal resistance of interfaces and materials. This is suboptimal for the health and performance of the li ion battery.
Immersion cooling fundamentally changes the heat transfer paradigm. By directly enveloping each li ion battery cell in a dielectric coolant, it drastically reduces thermal resistance and enables multidirectional heat removal. Even a basic two-plate immersion design shows a profound improvement in internal cell temperature uniformity, a critical factor for cycle life.
The proposed structural optimization—increasing the number of parallel internal cooling plates to five—solves the pack-level temperature gradient issue present in the initial design. The results are remarkable: a significant reduction in both maximum temperature and all measured temperature differences, alongside a drastic drop in system flow resistance. The optimized immersion cooling system ensures that the li ion battery pack operates cooler, more uniformly, and with higher auxiliary efficiency.
In conclusion, immersion cooling, particularly with a well-distributed internal heat exchanger network, presents a superior thermal management strategy for high-capacity energy storage li ion battery packs. It addresses the key challenges of temperature homogeneity and hotspot prevention more effectively than conventional methods. This work provides a validated numerical framework and design insights that can guide the development of safer, more efficient, and longer-lasting li ion battery-based energy storage systems, ultimately supporting a more resilient and sustainable energy grid.