In my study, I focus on the thermal management of an EV battery pack that uses high-energy-density lithium-ion cells. The EV battery pack is the core energy source of an electric vehicle, but during high-rate discharge it generates substantial heat. Without an effective heat dissipation path, the temperature of the battery pack rises rapidly, and the temperature distribution becomes highly non-uniform. Because lithium-ion cells are strongly temperature-sensitive, excessive temperature and temperature gradients can accelerate capacity fade, shorten cycle life, and even trigger thermal runaway. Therefore, the design of a high-efficiency cooling system is a key requirement for the safe and durable operation of an EV battery pack.
To solve this problem, I propose a hybrid cold plate that combines straight channels with pin-fin arrays. The pin-fin structures disturb the thermal boundary layer and enhance convective heat transfer, while the parallel channel layout keeps the pressure drop under control. I develop a coupled numerical model of the EV battery pack and the cold plate, validate it with experiments, and then carry out a multi-objective optimization. My aim is to obtain a set of design parameters that simultaneously reduce the maximum temperature, improve the temperature uniformity, and minimize the pumping power required by the EV battery pack cooling system.
1. Single-Cell Heat Generation Model
Before analyzing the full cooling system, I first establish a heat generation model for a single prismatic battery cell. The cell uses a 100 Ah lithium iron phosphate chemistry, with the following geometric and physical parameters.
| Parameter | Value | Unit |
| Nominal capacity | 100 | Ah |
| Operating voltage range | 2.5-3.65 | V |
| Cell dimensions | 160×116×50 | mm |
| Thermal conductivity | 22.5/22.5/1.5 | W/(m·K) |
| Cell mass | 1.65 | kg |
| Density | 2050 | kg/m³ |
| Specific heat capacity | 1088.745 | J/(kg·K) |
The heat generation rate inside the battery cell is calculated with the Bernardi model. The total volumetric heat generation rate is
$$q=\beta\frac{I}{V_c}\left[(E_0-U)-T\frac{dE_0}{dT}\right]$$
Equation includes irreversible Joule heat and polarization heat, together with the reversible entropic heat. In the engineering analysis considered here, polarization heat and entropic heat are small when compared with Joule heating at 1C discharge; therefore I simplify the heat generation expression to
$$q=\frac{I^2 R_T}{V_c}$$
where \(I\) is the discharge current, \(R_T\) is the battery internal resistance, and \(V_c\) is the cell volume. Because the internal resistance changes with the state of charge (SOC), I fit the measured resistance curve at 1C discharge to a polynomial. The resistance is relatively constant for SOC between 0.2 and 0.8, and increases sharply outside this range. To avoid the high heat generation region, I set the discharge end SOC to 0.2 in my simulation of the EV battery pack.
Using the fitted resistance and the heat generation formula, I perform a transient thermal simulation of a single cell. The cell is treated as a homogeneous solid with an internal heat source. The energy conservation equation in the solid is
$$\rho_b c_b\frac{\partial T}{\partial t}=k_{x,b}\frac{\partial^2 T}{\partial x^2}+k_{y,b}\frac{\partial^2 T}{\partial y^2}+k_{z,b}\frac{\partial^2 T}{\partial z^2}+Q$$
I use a multi-polyhedral mesh in Fluent, and the simulation predicts a maximum cell temperature of 318 K after a full 1C discharge from an initial temperature of 291 K. I compare the simulated temperature rise with experimental data, and the maximum difference at key time nodes is within 5%. This value confirms that the single-cell heat generation model is sufficiently accurate for the later coupled analysis of the EV battery pack and the cold plate.
2. Numerical Model of the Pin-Fin Channel Cold Plate
I design a battery module with 24 cells arranged in two rows. The cells are separated by thermally conductive silicone pads and cooled by three aluminum cold plates. Each cold plate has a central inlet and two side outlets, forming a symmetric parallel-channel network. The flow path is divided into eight channels. A pin-fin array is placed inside each channel to enhance heat transfer. I choose four configurations for comparison: straight channel (SC), square pin-fin (S-Pin-fin), circular pin-fin (C-Pin-fin), and triangular pin-fin (T-Pin-fin).
The structural dimensions and thermal properties used in the numerical model are summarized below.
| Component | Density (kg/m³) | Specific heat (J/(kg·K)) | Thermal conductivity (W/(m·K)) | Viscosity (Pa·s) |
| Silicone pad | 2094.96 | 2684 | 4 | / |
| Cold plate (aluminum) | 2719 | 871 | 202.4 | / |
| Coolant (50% ethylene glycol) | 1065 | 3281 | 0.38 | 0.0069×(T/273)^-8.3 |
I assume that the coolant is incompressible and has no-slip wall conditions. The straight channel is simulated with the laminar model, while the pin-fin channels are simulated with the standard \(k\)-\(\epsilon\) turbulence model because the pins generate flow separation and vortex shedding. The governing equations for the fluid are:
$$\frac{\partial \rho_f}{\partial t}+\nabla\cdot(\rho_f \mathbf{v})=0$$
$$\frac{\partial}{\partial t}(\rho_f \mathbf{v})+\nabla\cdot(\rho_f \mathbf{v}\mathbf{v})=-\nabla p$$
$$\frac{\partial}{\partial t}(\rho_f c_f T_f)+\nabla\cdot(\rho_f c_f \mathbf{v} T_f)=\nabla\cdot(k_f \nabla T_f)$$
For the solid components, I use:
$$\rho_s c_s\frac{\partial T_s}{\partial t}=\nabla\cdot(k_s \nabla T_s)$$
The \(k\) and \(\epsilon\) transport equations are:
$$\frac{\partial(\rho k)}{\partial t}+\frac{\partial(\rho u_j k)}{\partial x_i}=\frac{\partial}{\partial x_j}\left[\left(\mu+\frac{\mu_t}{\sigma_k}\right)\frac{\partial k}{\partial x_j}\right]+P_k-\rho\epsilon$$
$$\frac{\partial(\rho \epsilon)}{\partial t}+\frac{\partial(\rho u_j \epsilon)}{\partial x_i}=\frac{\partial}{\partial x_j}\left[\left(\mu+\frac{\mu_t}{\sigma_\epsilon}\right)\frac{\partial \epsilon}{\partial x_j}\right]+C_{\epsilon1}\frac{\epsilon}{k}P_k-C_{\epsilon2}\rho\frac{\epsilon^2}{k}$$
The model constants are \(C_{\epsilon1}=1.44\), \(C_{\epsilon2}=1.92\), \(\sigma_k=1.0\), \(\sigma_\epsilon=1.2\). All exposed outer surfaces of the EV battery pack are assigned a convective heat transfer coefficient of 10 W/(m²·K). The inlet coolant temperature and ambient temperature are kept at 293 K. The mass flow rate at the inlet is varied from 0.01 kg/s to 0.03 kg/s.
I use three performance metrics to compare the cold plate designs:
$$T_{\max}=\max\{T_1,T_2,\ldots,T_n\}$$
$$\Delta T=T_{\max,mid}-T_{\min,mid}$$
$$\Delta P=P_{in}-P_{out}$$
2.1 Mesh Independence and Model Validation
I generate poly-hexcore meshes for the fluid domain and solid domains. The mesh is locally refined near the pin surfaces and the walls. I perform a mesh independence study using the straight channel and circular pin-fin configurations. The maximum temperature and pressure drop stabilize when the grid count reaches about 12.5 million for the straight-channel model and 13.2 million for the circular pin-fin model. Therefore, I use these grid resolutions in all subsequent simulations.
For model validation, I fabricate a small pin-fin cold plate and set up a closed-loop experimental platform. The pressure drop predicted by the standard \(k\)-\(\epsilon\) model is compared with measured data for mass flow rates between 0.01 kg/s and 0.03 kg/s. The average pressure-drop error is 7.1%, and the maximum heating-surface temperature error is within 0.8 K. The good agreement confirms that the numerical model is reliable for evaluating the thermal-hydraulic behavior of the EV battery pack cooling system.
2.2 Comparison of Different Pin-Fin Shapes
At the end of the 1C discharge, I compare the maximum temperature and temperature difference for the four cold plate configurations. The maximum temperature decreases as the mass flow rate increases, but the rate of decrease slows down at higher flow rates. This is because the coolant temperature rise reduces the driving temperature difference. The straight channel has the highest temperature because the laminar boundary layer continuously thickens along the channel. The three pin-fin channels all outperform the straight channel by periodically disturbing the boundary layer. Among them, the circular pin-fin configuration achieves the best balance between heat transfer enhancement and flow resistance.
| Configuration | m (kg/s) | Tmax (K) | ΔT (K) | ΔP (kPa) |
| SC | 0.03 | 311.6 | 5.4 | 0.8 |
| S-Pin-fin | 0.03 | 308.4 | 4.7 | 4.5 |
| C-Pin-fin | 0.03 | 308.7 | 4.7 | 3.3 |
| T-Pin-fin | 0.03 | 309.1 | 4.8 | 3.9 |
At a mass flow rate of 0.03 kg/s, the circular pin-fin configuration reduces the maximum temperature by 2.9 K and the temperature difference by 0.7 K compared with the straight channel, while the pressure drop is much lower than that of the square pin-fin configuration. The velocity contour shows that the pin fins split the flow and create periodic acceleration. As a result, the heat transfer coefficient is increased not only near the inlet but also along the entire flow path.
I also visualize the vortex cores using the Q-criterion. The circular pin-fin produces regular ring-shaped vortices that are convected downstream. The high-turbulence region is mainly located in the vortex core and the near-wall shear layer. These vortices continuously destroy the thermal boundary layer and promote mixing. The square pin-fin creates fragmented vortices with strong turbulence dissipation, which leads to a high heat transfer coefficient but also a large pressure loss. The triangular pin-fin shows a combination of detachment at the leading edge and reattachment near the trailing edge. The circular pin-fin is therefore chosen for the subsequent multi-objective optimization of the EV battery pack cold plate.
3. Multi-Objective Optimization of the Circular Pin-Fin Cold Plate
3.1 Design Variables and Objective Functions
I choose three design variables that have a strong effect on the thermal-hydraulic performance of the cold plate: mass flow rate \(m\), channel height \(H\), and pin-fin pitch \(S\). The initial values and ranges are listed below.
| Design variable | Initial value | Minimum | Maximum |
| Mass flow rate, m (kg/s) | 0.02 | 0.01 | 0.03 |
| Channel height, H (mm) | 1.5 | 1.5 | 4 |
| Pin-fin pitch, S (mm) | 15 | 4 | 30 |
I initially consider three objective functions: maximum temperature \(T_{\max}\), temperature difference \(\Delta T\), and pressure drop \(\Delta P\). To reduce the computational cost of optimization, I first perform a sensitivity analysis and a correlation analysis.
3.2 Sensitivity and Correlation Analysis
The sensitivity of each objective to each design variable is calculated with
$$SA_i=\frac{f_{\max}(x_i)-f_{\min}(x_i)}{f_{\max}(x)-f_{\min}(x)}\times100\%$$
The sensitivity results indicate that the mass flow rate is the dominant operating parameter for \(T_{\max}\) and \(\Delta T\), with sensitivity values of 0.996 and 0.988, respectively. The channel height is the dominant geometric parameter for \(\Delta P\), with a sensitivity of 44.2%. The pin-fin pitch has only a small influence on pressure drop, with a sensitivity of 2.8%.
I also calculate the Pearson correlation coefficient between the objectives. The correlation between \(T_{\max}\) and \(\Delta T\) is 0.99, which means that these two objectives are highly positively correlated. Therefore, I remove \(T_{\max}\) from the optimization objectives and keep \(\Delta T\) and \(\Delta P\) as the two objectives. This improves the optimization efficiency without losing important information.
3.3 Kriging Surrogate Model
I use the optimal Latin hypercube sampling method to generate 31 sample points in the design space. Each sample is evaluated by the CFD model. A Kriging surrogate model is then built to replace the expensive CFD simulation. The coefficient of determination is calculated as
$$R^2=1-\frac{\sum_{i=1}^{n}(y_i-\hat{y}_i)^2}{\sum_{i=1}^{n}(y_i-\bar{y})^2}$$
The resulting coefficients of determination are 0.966 for \(\Delta T\) and 0.952 for \(\Delta P\). Both values are above 0.9, which indicates a high prediction accuracy. A representative portion of the sample points is shown in the table below.
| Sample | m (kg/s) | H (mm) | S (mm) | Tmax (K) | ΔT (K) | ΔP (kPa) |
| 1 | 0.0260 | 2.16 | 23.6 | 307.2 | 4.4 | 2.2 |
| 2 | 0.0167 | 2.34 | 17.3 | 309.5 | 6.2 | 1.0 |
| 3 | 0.0127 | 1.92 | 22.2 | 311.3 | 7.6 | 0.9 |
| 4 | 0.0207 | 1.66 | 20.8 | 308.2 | 5.2 | 2.4 |
| 5 | 0.0300 | 2.76 | 19.4 | 306.7 | 4.0 | 2.1 |
| 6 | 0.0147 | 4.00 | 11.0 | 310.6 | 6.8 | 0.5 |
| 7 | 0.0100 | 3.34 | 8.9 | 313.3 | 8.7 | 0.3 |
| 8 | 0.0233 | 3.92 | 20.1 | 307.9 | 4.8 | 1.1 |
3.4 NSGA-II Optimization Result
I use the NSGA-II algorithm to search for the Pareto-optimal front. The population size is set to 120 and the number of generations is 50, so the algorithm generates 6000 candidate solutions. The Pareto front clearly shows a trade-off between \(\Delta T\) and \(\Delta P\). When the temperature difference is required to be very small, the pressure drop necessarily increases. I select three representative solutions from the Pareto front and validate each one with a CFD simulation.
| Model | m (kg/s) | H (mm) | S (mm) | Tmax (K) | ΔT (K) | Error | ΔP (kPa) | Error |
| Baseline | 0.0200 | 1.50 | 15 | 308.3 | 5.40 | / | 3.10 | / |
| Optimized model 1 (NSGA) | 0.0253 | 3.66 | 22 | / | 4.47 | 0.9% | 1.25 | 1.6% |
| Optimized model 1 (CFD) | 0.0253 | 3.66 | 22 | 307.5 | 4.51 | / | 1.27 | / |
| Optimized model 2 (NSGA) | 0.0285 | 3.14 | 19 | / | 4.02 | 3.1% | 1.77 | 0.6% |
| Optimized model 2 (CFD) | 0.0285 | 3.14 | 19 | 306.9 | 4.15 | / | 1.76 | / |
| Optimized model 3 (NSGA) | 0.0294 | 2.52 | 18 | / | 3.93 | 3.4% | 2.34 | 1.7% |
| Optimized model 3 (CFD) | 0.0294 | 2.52 | 18 | 306.7 | 4.07 | / | 2.30 | / |
I choose optimized model 2 as the final design because it gives the best compromise between thermal performance and energy consumption. Compared with the baseline model, the final optimized cold plate reduces the temperature difference by 23.1% from 5.40 K to 4.15 K, and reduces the pressure drop by 43.2% from 3.10 kPa to 1.76 kPa. The maximum prediction error of the Kriging model is less than 3.4%, which confirms the reliability of the optimization method.
3.5 Evaluation of the Optimized Cold Plate
To further evaluate the optimized cold plate, I calculate the thermal resistance components. The total thermal resistance is decomposed into convective, heat-capacity, and conductive resistances:
$$R_{conv}=\frac{T_{ave}-T_{liq}}{q}$$
$$R_{heat}=\frac{T_{out}-T_{in}}{q}$$
$$R_{cond}=\frac{L_b}{k_s A_{cont}}$$
$$R_{total}=R_{conv}+R_{heat}+R_{cond}$$
At 0.02 kg/s, the optimized cold plate reduces the total thermal resistance by 14% compared with the straight channel. At 0.03 kg/s, the reduction is 16%. The improvement is mainly contributed by the convective thermal resistance, while the heat-capacity and conductive resistances are nearly unchanged. This confirms that the pin-fin structure significantly enhances convective heat transfer in the EV battery pack cold plate.
I also use the effective heat transfer enhancement factor to compare the heat transfer benefit with the additional pressure-drop penalty:
$$Nu=\frac{h_{ave}D_h}{k_{liq}}$$
$$p_f=\frac{Nu/Nu_0}{(\Delta P/\Delta P_0)^{1/3}}$$
At a mass flow rate of 0.02 kg/s, the optimized circular pin-fin cold plate has a Nusselt number that is more than six times higher than that of the straight channel, and the effective enhancement factor \(p_f\) is about seven times higher. Compared with the baseline circular pin-fin cold plate, the optimized design improves \(Nu\) by 33.9% and \(p_f\) by 97.5%. This result demonstrates that the optimized structure achieves a large heat-transfer enhancement with a relatively small increase in flow resistance.
The cooling efficiency factor is used to relate the heat removed by the coolant to the pumping power consumed:
$$Q_{liq}=c\,m\,\Delta T$$
$$\eta=\frac{Q_{liq}/t}{\Delta P\,q_v}$$
The optimized cold plate reaches a maximum cooling efficiency factor of 15452 at 0.02 kg/s, which is much higher than the baseline circular pin-fin model (4875) and the straight channel (7250). This indicates that the optimized cold plate can remove a large amount of heat from the EV battery pack while consuming only a small amount of pumping energy.
4. Experimental Study of the Circular Pin-Fin Cold Plate
4.1 Effect of Channel Height
I further analyze the influence of channel height on the cooling performance and pressure drop. The channel height directly determines the thickness of the cold plate and therefore affects the compactness of the EV battery pack thermal management system. I keep the optimized mass flow rate and pin-fin spacing constant and vary the channel height from 1 mm to 4 mm.
| Channel height (mm) | Tmax (K) | ΔT (K) | ΔP (kPa) |
| 1.0 | 306.6 | 4.18 | 13.0 |
| 1.5 | 306.9 | 4.15 | 5.3 |
| 2.0 | 307.0 | 4.23 | 2.9 |
| 3.0 | 307.1 | 4.28 | 1.4 |
| 4.0 | 307.2 | 4.31 | 0.9 |
The maximum temperature increases by only 0.3 K when the channel height increases from 1 mm to 4 mm. In contrast, the pressure drop drops dramatically. When the channel height increases from 1 mm to 1.5 mm, the pressure drop is reduced by 59.2%, from 13.0 kPa to 5.3 kPa. Further increases in channel height give a smaller additional pressure-drop reduction. I therefore select a channel height of 1.5 mm as the best compromise between thermal performance and structural compactness. This cold plate is denoted as C-Pin-fin (BCS) and is used in the experimental study.
4.2 Fabrication and Experimental Setup
The C-Pin-fin (BCS) cold plate is fabricated with subtractive manufacturing using CNC milling. The cold plate is made of aluminum. The overall dimensions are 732 mm × 191 mm × 12 mm, with an inlet/outlet diameter of 6 mm, a channel height of 1.5 mm, a pin-fin spacing of 19 mm, and a pin-fin diameter of 3 mm. The cold plate is sealed with O-rings and assembled with threaded fasteners.
I build a closed-loop experimental platform to investigate the flow and heat transfer characteristics of the cold plate. The platform consists of a gear pump, a liquid turbine flow meter, PT100 temperature sensors, two pressure transmitters, T-type thermocouples, an infrared thermal camera, a constant-temperature water bath, a DC power supply, a data acquisition system, and a computer. Four heating films are attached to the cold plate surface to simulate the heat source of the EV battery pack. The coolant is 50% ethylene glycol aqueous solution, commonly used in electric vehicle thermal management systems.
| Equipment | Accuracy | Range |
| Multi-channel DC power supply | / | 0-200 W |
| Infrared thermal camera | ±2% | -20-2000 ℃ |
| Liquid turbine flow meter | ±0.5% | 0-10 L/min |
| Constant-temperature water bath | ±0.75% | -20-100 ℃ |
| PT100 temperature sensor | ±1% | -100-280 ℃ |
| Inlet pressure transmitter | ±0.5% | 0-500 kPa |
| Outlet pressure transmitter | ±0.5% | 0-130 kPa |
4.3 Data Reduction and Uncertainty Analysis
The heating power supplied by the heating films is calculated from the voltage and current:
$$P_h=8UI_h$$
The effective heat flux on the heating surface is
$$q=\frac{c_p m \Delta T}{A}$$
where \(A\) is the total surface area of the heating films. The pumping power is calculated as
$$P_p=\Delta P\cdot V$$
The uncertainty analysis follows the error propagation theory. If a calculated parameter \(R\) is a function of directly measured variables \(v_1,v_2,\ldots,v_n\), the uncertainty is
$$\delta R=\sqrt{\sum_{i=1}^{n}\left(\frac{\partial R}{\partial v_i}\delta v_i\right)^2}$$
$$\xi_R=\frac{\delta R}{R}$$
The maximum relative uncertainties of the measured and derived parameters are listed below.
| Parameter | Maximum relative uncertainty |
| Inlet pressure | 2.4% |
| Outlet pressure | 0.6% |
| Volumetric flow rate | 2.5% |
| Mass flow rate | 2.5% |
| Inlet/outlet coolant temperature | 1.6% |
| Pressure drop | 2.5% |
| Heating-film heat flux | 3.1% |
| Thermal resistance | 3.6% |
4.4 Flow Resistance and Heat Transfer Results
I first measure the pressure drop of the C-Pin-fin (BCS) cold plate as a function of flow rate with no heating power. The pressure drop increases monotonically with the flow rate, but the slope changes in three stages. In the low-flow-rate region, the pressure drop increases slowly because viscous forces dominate and the flow remains attached around the pin fins. In the medium-flow-rate region, the pressure-drop growth rate increases as flow separation and wake regions appear behind the pins. In the high-flow-rate region, the pressure-drop curve rises sharply because inertial forces dominate, and the vortex shedding behind the pins produces strong turbulent dissipation. The pressure-drop values remain acceptable over the whole tested range.
I then carry out heat transfer experiments under different combinations of heating power and coolant flow rate. The heating power is varied from 200 W to 440 W. At a fixed flow rate, the heating-film temperature rises as the heat flux increases, because the coolant has a limited cooling capacity at that flow rate. At a fixed heating power, an increase in the flow rate reduces the heating-film temperature, because a higher velocity increases the convective heat transfer coefficient and thins the thermal boundary layer.
The infrared thermal images show a clear temperature gradient along the flow direction. The heating film near the inlet has the lowest temperature, while the downstream heating film has the highest temperature. This occurs because the coolant heats up as it flows through the channel, reducing the local temperature difference and thus the local heat transfer. The pin-fin array disrupts the boundary layer and creates secondary flow, but the overall temperature field still follows a general trend of increasing temperature from inlet to outlet.
The thermal resistance decreases when the flow rate increases. At the same time, the pump power increases in an approximately exponential manner because both flow rate and pressure drop rise together. The thermal resistance also decreases when the heating power increases, but the reduction gradually becomes smaller and eventually saturates. This behavior is caused by the temperature-dependent thermophysical properties of the coolant. At higher temperatures, the viscosity of the coolant decreases and its thermal conductivity increases, which improves heat transfer. However, this property-related improvement is limited, so the thermal resistance reaches a nearly constant value at high heating power.
The experimental results confirm that the numerical model is reliable. The maximum relative error between the predicted and measured pressure drop is 8%, and the maximum error for the average surface temperature is 1.2 K. This agreement proves that the C-Pin-fin (BCS) cold plate can be accurately modeled and optimized for EV battery pack thermal management.
5. Conclusion
In this work, I systematically studied the single-phase flow and heat transfer characteristics of pin-fin channel cold plates for EV battery pack cooling. The major conclusions are as follows.
First, the single-cell heat generation model based on the Bernardi equation and experimentally fitted internal resistance is accurate within 5% of measured temperature rise. The coupled battery-cold-plate numerical model can therefore be used for the thermal design of an EV battery pack.
Second, among the straight, square, circular, and triangular pin-fin channels, the circular pin-fin configuration provides the best trade-off between heat transfer enhancement and pressure drop. At 0.03 kg/s, the circular pin-fin cold plate reduces the maximum temperature by 2.9 K and the temperature difference by 0.7 K compared with the straight channel. The circular pin-fin generates regular ring-shaped vortices that efficiently use turbulent kinetic energy and avoid excessive dissipation.
Third, the Kriging-based NSGA-II multi-objective optimization yields a promising cold plate design with a mass flow rate of 0.0285 kg/s, a channel height of 3.14 mm, and a pin-fin spacing of 19 mm. Compared with the baseline model, the optimized cold plate reduces the temperature difference by 23.1% and the pressure drop by 43.2%. The thermal resistance is reduced by up to 16% relative to the straight channel, and the cooling efficiency factor reaches 15452.
Fourth, the experimental study shows that the channel height of 1.5 mm is optimal for both thermal performance and structural compactness. The pressure drop of this cold plate is 59.2% lower than that of the 1 mm channel-height design, while the maximum temperature is only 0.3 K higher. The experiments also demonstrate that an increase in coolant flow rate reduces the heating-film temperature and thermal resistance but increases pump power. The thermal resistance decreases with increasing heat flux and eventually stabilizes because of the temperature-dependent thermophysical properties of the coolant.
Overall, the circular pin-fin channel cold plate is an efficient and practical solution for the thermal management of an EV battery pack. The combination of numerical simulation, multi-objective optimization, and experimental validation provides a reliable framework for designing high-performance cooling systems that can maintain the EV battery pack within a safe and uniform temperature range at low energy cost.