I approach the thermal management of an electric vehicle battery pack as a coupled electrochemical, thermal, and fluid problem. The central motivation is straightforward: an electric vehicle battery pack must deliver high power and high energy over repeated charge–discharge cycles, yet its performance, safety, and lifetime are strongly constrained by temperature. In my work, I focus on a lithium-ion electric vehicle battery pack because lithium-ion cells provide high energy density, long cycle life, and acceptable power capability, but they remain sensitive to both high-temperature and low-temperature operation. I designed and evaluated a coupled microchannel–phase change material architecture that can cool, insulate, and preheat an electric vehicle battery pack within one integrated structure. The goal is not merely to reduce peak temperature, but also to improve temperature uniformity, suppress thermal runaway risk, shorten low-temperature preheating time, and preserve useful heat during cold parking.
From a thermal management perspective, an electric vehicle battery pack has three major requirements. First, the preferred operating window should remain approximately between 25°C and 40°C. Second, the maximum temperature difference among cells should remain below 5°C, because larger gradients accelerate capacity fade and increase cell-to-cell inconsistency. Third, in cold environments, the electric vehicle battery pack should reach a usable temperature quickly, ideally above 20°C, while maintaining acceptable uniformity. These criteria guided every simulation and parameter study that I performed. I treated the electric vehicle battery pack as a multi-domain system containing cylindrical cells, a composite phase change material, an aluminum frame, and embedded microchannels through which a liquid coolant or preheating fluid flows.
I used a simplified but physically meaningful heat-generation model for the lithium-ion cells. The total heat generation includes reversible reaction heat, ohmic heat, polarization heat, and secondary reaction heat. Under normal operating conditions, the secondary reaction heat is small and can be neglected. I therefore represented the volumetric heat source as a function of current, internal resistance, and entropic variation. The general energy balance for a cell can be written as:
$$q = \frac{I}{V_b}\left[E_{oc}-E-T\frac{\partial E_{oc}}{\partial T}\right]$$
where \(q\) is the volumetric heat generation rate, \(I\) is current, \(V_b\) is cell volume, \(E_{oc}\) is open-circuit voltage, \(E\) is operating voltage, and \(T\) is temperature. For engineering simulation, I simplified the internal resistance term so that the heat source became a function of temperature and discharge rate. The combined ohmic and polarization resistance was represented as:
$$R_{\Omega}+R_P=-0.0001T^3+0.0134T^2-0.5345T+12.407$$
This allowed me to compute the volumetric heat generation for different discharge rates. I then compared the predicted cell temperature rise with independent experimental data, and the maximum deviation was small enough to justify the model for system-level studies of an electric vehicle battery pack. The validation case used a 3C discharge, an ambient temperature of 20°C, and a discharge duration of 1200 s. The predicted final temperature difference relative to the reference was only 0.64°C, which I considered acceptable for design optimization.
For the cell domain, I solved a three-dimensional transient heat conduction equation with anisotropic properties. The governing equation is:
$$\rho C_p \frac{\partial T}{\partial t} = \lambda_r \frac{1}{r}\frac{\partial}{\partial r}\left(r\frac{\partial T}{\partial r}\right) + \lambda_\theta \frac{1}{r^2}\frac{\partial^2 T}{\partial \theta^2} + \lambda_z \frac{\partial^2 T}{\partial z^2} + q$$
Here, \(\rho\) is density, \(C_p\) is specific heat capacity, \(\lambda_r\), \(\lambda_\theta\), and \(\lambda_z\) are radial, circumferential, and axial thermal conductivities, and \(q\) is the volumetric heat source. The material properties that I assigned to the equivalent cylindrical cell are summarized in Table 1. These values were obtained by volume averaging and by applying series and parallel thermal resistance approximations for the layered internal structure.
| Property | Value | Unit |
| Average density | 2180 | kg/m³ |
| Average specific heat capacity | 1066 | J/(kg·K) |
| Radial thermal conductivity | 1.1 | W/(m·K) |
| Circumferential thermal conductivity | 1.1 | W/(m·K) |
| Axial thermal conductivity | 12.5 | W/(m·K) |
I selected 18650 cylindrical lithium iron phosphate cells for the electric vehicle battery pack model. Each cell has a nominal capacity of 1.5 Ah, a diameter of 18 mm, and a height of 65 mm. I arranged 16 cells in a 4×4 honeycomb-inspired layout. This arrangement increases packing efficiency and creates relatively regular pathways for heat transfer between the cells, the phase change material, and the microchannel network. The cell spacing and frame geometry are important because they determine whether heat can be removed effectively from the interior of an electric vehicle battery pack.
The heat generation rates that I used for the high-temperature cooling study are listed in Table 2. The values increase nonlinearly with discharge rate, which means that a high-rate electric vehicle battery pack can produce several times more heat at 4C than at 1C. This nonlinearity is one reason why passive cooling alone is often insufficient for aggressive fast-charging or high-power driving scenarios.
| Discharge rate | Current | Volumetric heat generation |
| 1C | 1.5 A | 5362 W/m³ |
| 2C | 3.0 A | 30586 W/m³ |
| 3C | 4.5 A | 60156 W/m³ |
| 4C | 6.0 A | 126758 W/m³ |
I designed the coupled thermal management system as an aluminum frame containing vertical circular microchannels, a composite phase change material filling the interstitial space, and the 18650 cells embedded inside the composite phase change material. The aluminum frame provides structural support and a high-conductivity path between the phase change material and the microchannel fluid. The phase change material stores latent heat and smooths transient temperature fluctuations. The microchannels provide active heat removal or active heat supply, depending on whether the electric vehicle battery pack is being cooled in a hot environment or preheated in a cold environment. In my baseline design, the microchannel diameter is 1 mm, and the frame height is 65 mm. The baseline layout contains 63 microchannels distributed along the six sides of the aluminum frame.
For the liquid domain, I solved the continuity, momentum, and energy equations for incompressible water flow. The continuity equation is:
$$\frac{\partial \rho_w}{\partial t}+\nabla\cdot(\rho_w\mathbf{v})=0$$
The momentum equation is:
$$\frac{\partial}{\partial t}(\rho_w\mathbf{v})+\nabla\cdot(\rho_w\mathbf{v}\mathbf{v})=-\nabla p+\nabla\cdot(\mu\nabla\mathbf{v})$$
The energy equation for the water is:
$$\frac{\partial}{\partial t}(\rho_w c_{p,w}T)+\nabla\cdot(\rho_w c_{p,w}\mathbf{v}T)=\nabla\cdot(\lambda_w\nabla T)$$
Here, \(\rho_w\), \(c_{p,w}\), \(\lambda_w\), \(\mu\), and \(\mathbf{v}\) are water density, specific heat capacity, thermal conductivity, dynamic viscosity, and velocity vector, respectively. Because the Reynolds number in the microchannels remained below the transition value for the cases I studied, I used a laminar flow model. I also activated the solidification and melting model to capture phase change in the composite phase change material.
For the phase change material, I used an enthalpy-porosity formulation. The energy equation is:
$$\rho_{PCM}\frac{\partial H}{\partial t}=\nabla\cdot(\lambda_{PCM}\nabla T)$$
The total enthalpy is split into sensible and latent contributions:
$$H=H_1+\Delta H$$
$$H_1=\int_{T_0}^{T}c_{p,PCM}dT$$
$$\Delta H=\beta\gamma$$
where \(\gamma\) is the latent heat of fusion, and \(\beta\) is the liquid fraction. I represented the liquid fraction as:
$$\beta=
\begin{cases}
0, & TT_l
\end{cases}$$
Here, \(T_s\) and \(T_l\) are the solidus and liquidus temperatures. This formulation allows the composite phase change material to absorb or release latent heat while maintaining a relatively stable temperature near the phase change point. For an electric vehicle battery pack, this behavior is valuable because it reduces temperature spikes during high-rate discharge and helps preserve warmth during cold soaking.
I used a commercial CFD solver with a pressure-based transient formulation. I activated the energy equation, the laminar viscous model, and the solidification/melting model. The cell domain was assigned a volumetric heat source corresponding to the selected discharge rate. For the cooling cases, the inlet boundary was a velocity inlet with a specified water temperature, and the outlet boundary was a pressure outlet. The exterior surfaces exchanged heat with the environment through convection. The solid–solid and solid–liquid interfaces used coupled wall conditions with a no-slip velocity boundary. I used the SIMPLE algorithm for pressure–velocity coupling, second-order upwind discretization for momentum and energy, and a first-order implicit scheme for time advancement. The convergence criteria were set to \(10^{-6}\).
I performed mesh independence and time-step independence studies before using the model for parameter optimization. Four mesh sizes were tested: 1,166,891; 1,971,858; 3,802,869; and 7,464,891 cells. The maximum temperature became nearly invariant after approximately 1.97 million cells. I therefore selected the 1,971,858-cell mesh for the main simulations. I also compared time steps of 0.5 s, 1 s, and 2 s. The differences in maximum temperature and maximum temperature difference were negligible, while the 2 s step required the least computation. The final simulations used a 2 s time step with up to 20 iterations per step. The mesh independence results are summarized in Table 3.
| Mesh cells | Maximum temperature trend | Decision |
| 1,166,891 | Noticeably higher | Coarse |
| 1,971,858 | Converged within acceptable tolerance | Selected |
| 3,802,869 | Nearly identical to selected mesh | Over-resolved |
| 7,464,891 | No meaningful improvement | Rejected |
To evaluate the advantage of coupling microchannels with phase change material, I first compared two configurations for a hot environment: a phase-change-only system and a coupled microchannel–phase-change system. The ambient temperature was 35°C, the discharge rate was 4C, and the inlet water temperature in the coupled case was 35°C. The inlet velocity was 0.1 m/s. The phase-change-only case replaced the aluminum frame and water channels with additional phase change material, so that the electric vehicle battery pack relied only on latent heat storage and passive conduction.
The results showed a clear benefit from active microchannel cooling. In the phase-change-only case, the maximum temperature reached 41.98°C, and the maximum temperature difference reached 3.50°C. In the coupled case, the maximum temperature was 39.83°C, and the maximum temperature difference was 2.95°C. The coupled system therefore reduced the peak temperature by approximately 2.15°C and reduced the temperature difference by about 0.55°C. The final liquid fraction in the phase-change-only case was 0.90, whereas in the coupled case it was only 0.33. This means that active cooling removed a large fraction of the heat before it could melt the phase change material. The phase change material was therefore not exhausted, leaving thermal buffer capacity for transients. For an electric vehicle battery pack, this residual latent heat capacity is important because it provides safety margin during unexpected load spikes.
| Cooling mode | Maximum temperature | Maximum temperature difference | Final liquid fraction |
| Phase change material only | 41.98°C | 3.50°C | 0.90 |
| Microchannel + phase change material | 39.83°C | 2.95°C | 0.33 |
| Improvement | 2.15°C lower | 0.55°C lower | More buffer retained |
I then analyzed the effect of microchannel number on cooling performance while keeping the total flow rate constant. I defined three layouts. The first layout contained 63 microchannels, the second contained 111 microchannels, and the third contained 174 microchannels. To compare them fairly, I adjusted the inlet velocity so that the total volumetric flow through the electric vehicle battery pack remained constant. The inlet velocities were 0.1 m/s, 0.0567 m/s, and 0.0362 m/s for the three layouts, respectively.
Increasing the number of microchannels increases the heat transfer area, but it also reduces the local velocity in each channel when the total flow rate is fixed. These two effects compete. I found that the maximum temperature first decreased slightly and then increased. The 111-channel case produced a maximum temperature of 39.82°C, almost identical to the 63-channel case, while the 174-channel case increased to 40.09°C. The maximum temperature difference increased from 2.94°C in the 63-channel case to 3.14°C in the 111-channel case and 3.51°C in the 174-channel case. The final liquid fraction was 0.33, 0.24, and 0.25, respectively. I concluded that adding channels beyond a moderate number does not necessarily improve the thermal performance of an electric vehicle battery pack when the total flow rate is fixed. The baseline 63-channel layout provided the best balance in this study.
| Microchannel layout | Number of channels | Inlet velocity for equal total flow | Maximum temperature | Temperature difference | Final liquid fraction |
| Case 1 | 63 | 0.1000 m/s | 39.83°C | 2.94°C | 0.33 |
| Case 2 | 111 | 0.0567 m/s | 39.82°C | 3.14°C | 0.24 |
| Case 3 | 174 | 0.0362 m/s | 40.09°C | 3.51°C | 0.25 |
Next, I studied the effect of coolant velocity in the baseline 63-channel layout. I tested inlet velocities of 0.05 m/s, 0.1 m/s, and 0.2 m/s. The maximum temperature after discharge decreased from 40.25°C at 0.05 m/s to 39.83°C at 0.1 m/s and 39.63°C at 0.2 m/s. The improvement from 0.05 m/s to 0.1 m/s was substantial, but the additional benefit from 0.1 m/s to 0.2 m/s was much smaller. This indicates a diminishing return. A higher coolant velocity also increases pumping power and may reduce the utilization of latent heat because the phase change material melts less. The liquid fraction decreased as velocity increased, meaning that the phase change material remained more solid and retained more latent capacity. However, higher velocity also reduced the uniformity benefit of phase change in some regions. I selected 0.1 m/s as the balanced cooling flow rate for the electric vehicle battery pack because it achieved a low peak temperature without excessive parasitic energy consumption.
| Coolant velocity | Maximum temperature | Observation |
| 0.05 m/s | 40.25°C | Cooling is sufficient but peak temperature is higher |
| 0.10 m/s | 39.83°C | Best balance of cooling and energy use |
| 0.20 m/s | 39.63°C | Small additional cooling benefit |
I also examined the influence of graphite mass fraction in the composite phase change material. The composite consisted of paraffin and expanded graphite. Increasing the graphite fraction raises the effective thermal conductivity but reduces the specific heat capacity and latent heat. I tested mass fractions of 0%, 2%, 5%, 8%, 10%, 15%, and 20%. The relevant properties are summarized in Table 6. As the graphite mass fraction increased, the maximum temperature of the electric vehicle battery pack decreased because heat could be conducted more efficiently from the cells toward the microchannels and aluminum frame. However, the temperature difference increased slightly because the outer regions of the cells released heat faster than the core regions. The liquid fraction also decreased as the graphite fraction increased, which means that more heat was carried away by the liquid coolant and less was stored as latent heat.
| Graphite mass fraction | Thermal conductivity | Density | Specific heat capacity | Latent heat | Phase change temperature |
| 0% | 0.245 W/(m·K) | 800 kg/m³ | 2300 J/(kg·K) | 200.66 J/g | 37°C |
| 2% | 0.324 W/(m·K) | 800 kg/m³ | 2260 J/(kg·K) | 196.65 J/g | 37°C |
| 5% | 0.907 W/(m·K) | 800 kg/m³ | 2200 J/(kg·K) | 191.37 J/g | 37°C |
| 8% | 1.374 W/(m·K) | 800 kg/m³ | 2140 J/(kg·K) | 184.61 J/g | 37°C |
| 10% | 1.687 W/(m·K) | 800 kg/m³ | 2100 J/(kg·K) | 180.60 J/g | 37°C |
| 15% | 2.499 W/(m·K) | 800 kg/m³ | 2000 J/(kg·K) | 170.56 J/g | 37°C |
| 20% | 3.355 W/(m·K) | 800 kg/m³ | 1900 J/(kg·K) | 160.53 J/g | 37°C |
I found that a 5% graphite mass fraction is a practical optimum for the electric vehicle battery pack. With 5% graphite, the maximum temperature remained below 40°C, and the temperature difference remained well below 5°C. Further increases in graphite content produced only marginal improvements in peak temperature while reducing latent heat storage, increasing electrical conductivity, and raising safety concerns. The electrical conductivity issue is particularly important because a conductive phase change material could increase the risk of short circuits inside an electric vehicle battery pack. Therefore, I selected 5% graphite by mass as the preferred composite phase change material for both cooling and preheating studies.
I then investigated the effect of flow direction. In the co-current arrangement, all coolant entered from the top and exited from the bottom. In the counter-current arrangement, half of the coolant entered from the top and half entered from the bottom, creating opposing flow directions within the aluminum frame. The counter-current design reduced the maximum temperature only slightly, from 39.83°C to 39.80°C, but it reduced the maximum temperature difference from 3.15°C to 2.77°C and increased the final liquid fraction from 0.25 to 0.29. This means that counter-current flow improved temperature uniformity and used the phase change material more evenly. In an electric vehicle battery pack, better uniformity directly reduces cell-to-cell degradation differences.
| Flow arrangement | Maximum temperature | Maximum temperature difference | Final liquid fraction |
| Co-current | 39.83°C | 3.15°C | 0.25 |
| Counter-current | 39.80°C | 2.77°C | 0.29 |
The temperature contours explained why counter-current flow performed better. In the co-current case, the aluminum frame developed a clear axial temperature gradient from inlet to outlet. The cells near the inlet were cooler, while the cells near the outlet were warmer. In the counter-current case, the opposing flows produced a more symmetric temperature field. The high-temperature region became more centered and less skewed toward one side of the electric vehicle battery pack. This also improved the utilization of the phase change material. Regions near the inlet did not remain excessively solid, and regions near the outlet did not melt prematurely. The result was a more balanced thermal state across the entire electric vehicle battery pack.
After establishing the cooling behavior, I evaluated the insulation and preheating performance of the same architecture. For insulation, I initialized the electric vehicle battery pack at 50°C and exposed it to an ambient temperature of -10°C for 10 hours. I compared cases without phase change material and with composite phase change materials having graphite mass fractions of 0%, 5%, 8%, 10%, 15%, and 20%. Without phase change material, the maximum cell temperature dropped below 0°C within about one hour. With phase change material, the electric vehicle battery pack remained above 0°C for a much longer period because the material released latent heat during solidification. As the graphite mass fraction increased, the insulation time shortened because heat could escape more easily to the environment. Pure paraffin gave the longest insulation because it had the lowest thermal conductivity and the highest latent heat. However, pure paraffin also has poor thermal conductivity, which is unfavorable for active cooling and preheating. Therefore, the 5% graphite composite again provided a balanced compromise between insulation and heat transfer capability.
| Insulation condition | Initial temperature | Ambient temperature | Observation |
| No phase change material | 50°C | -10°C | Falls below 0°C in about 1 hour |
| 0% graphite | 50°C | -10°C | Longest warmth retention |
| 5% graphite | 50°C | -10°C | Balanced insulation and conductivity |
| 20% graphite | 50°C | -10°C | Shorter warmth retention due to higher conductivity |
For low-temperature preheating, I set the initial temperature of the electric vehicle battery pack and the ambient temperature to -20°C, and I supplied warm fluid through the microchannels. The target was for the minimum cell surface temperature to reach 20°C. I evaluated the preheating time, the temperature rise rate, and the maximum temperature difference at the end of preheating. I defined the temperature rise rate as:
$$RTR=\frac{20-(-20)}{\tau}$$
where \(\tau\) is the time required to reach 20°C. If \(\tau\) is expressed in minutes, \(RTR\) has units of °C/min. This metric allowed me to compare different operating strategies for the electric vehicle battery pack under cold conditions.
I first varied the graphite mass fraction of the composite phase change material while using the baseline 63-channel layout, a preheating fluid velocity of 0.1 m/s, and a top-to-bottom flow direction. The preheating fluid temperature was 25°C. The time required to reach 20°C decreased from 405 s at 0% graphite to 360 s at 2%, 272 s at 5%, 256 s at 8%, 250 s at 10%, 240 s at 15%, and 235 s at 20%. The largest improvement occurred between 0% and 5% graphite. Beyond 5%, the additional benefit became small. The temperature difference increased as graphite content increased because heat moved more quickly from the aluminum frame into the outer regions of the electric vehicle battery pack. I selected 5% graphite as the preferred composition because it provided a strong reduction in preheating time without excessive temperature non-uniformity or safety risk.
| Graphite mass fraction | Preheating time to 20°C | Temperature difference trend | Assessment |
| 0% | 405 s | Lowest | Slow preheating |
| 2% | 360 s | Slightly higher | Improved but still slow |
| 5% | 272 s | Moderate | Best compromise |
| 8% | 256 s | Higher | Diminishing benefit |
| 10% | 250 s | Higher | Marginal improvement |
| 15% | 240 s | Higher | Safety and latent heat penalty |
| 20% | 235 s | Highest | Not recommended |
I then varied the preheating fluid velocity. Using 5% graphite, the baseline 63-channel layout, and top-to-bottom flow, I tested velocities of 0.05 m/s, 0.1 m/s, 0.25 m/s, 0.5 m/s, and 1.0 m/s. The preheating time decreased from 380 s at 0.05 m/s to 272 s at 0.1 m/s, 211 s at 0.25 m/s, 186 s at 0.5 m/s, and 169 s at 1.0 m/s. The largest reduction occurred between 0.05 m/s and 0.25 m/s. Beyond 0.25 m/s, the marginal benefit was much smaller. The temperature difference increased with velocity because the inlet region received heat more rapidly. I therefore selected 0.25 m/s as the balanced preheating velocity for the electric vehicle battery pack. It shortened preheating time significantly while keeping the temperature difference within a reasonable range and avoiding excessive pumping power.
| Preheating fluid velocity | Preheating time to 20°C | Temperature difference trend | Assessment |
| 0.05 m/s | 380 s | Lowest | Slow |
| 0.10 m/s | 272 s | Moderate | Good |
| 0.25 m/s | 211 s | Moderate | Best balance |
| 0.50 m/s | 186 s | Higher | Diminishing benefit |
| 1.00 m/s | 169 s | Highest | Excess pumping and non-uniformity |
I also studied the effect of channel number on preheating. Keeping the total flow rate constant, I compared the 63-channel, 111-channel, and 174-channel layouts. The corresponding inlet velocities were 0.25 m/s, 0.1418 m/s, and 0.0905 m/s. The preheating time decreased from 211 s for 63 channels to 192 s for 111 channels and 181 s for 174 channels. However, the maximum temperature difference increased from 3.32°C to 3.65°C and then to 4.50°C. More channels increased the heat transfer area and allowed heat to enter the electric vehicle battery pack more quickly, but they also caused the inlet-side region to warm faster than the outlet-side region. I concluded that the 111-channel layout offered a useful compromise for preheating: it shortened the preheating time by about 19 s compared with the baseline while keeping the temperature difference below 5°C.
| Microchannel layout for preheating | Number of channels | Inlet velocity for equal total flow | Preheating time to 20°C | Temperature difference |
| Case 1 | 63 | 0.2500 m/s | 211 s | 3.32°C |
| Case 2 | 111 | 0.1418 m/s | 192 s | 3.65°C |
| Case 3 | 174 | 0.0905 m/s | 181 s | 4.50°C |
I then examined the effect of flow direction during preheating using the 111-channel layout, 5% graphite, and an inlet velocity of 0.1418 m/s. The co-current arrangement required 192 s to reach 20°C, while the counter-current arrangement required 194 s. The difference was only 2 s, so preheating speed was essentially unchanged. However, the maximum temperature difference decreased from 3.65°C in the co-current case to 3.25°C in the counter-current case. The counter-current design therefore improved temperature uniformity even though it slightly increased the preheating time. For an electric vehicle battery pack, this trade-off is acceptable because uniformity is critical for long-term durability and safe operation.
| Preheating flow direction | Preheating time to 20°C | Temperature difference | Observation |
| Co-current | 192 s | 3.65°C | Faster by 2 s but less uniform |
| Counter-current | 194 s | 3.25°C | Better uniformity |
Finally, I investigated the influence of preheating fluid temperature and ambient temperature on low-temperature performance. I considered ambient temperatures of -40°C, -30°C, -20°C, and -10°C, and preheating fluid temperatures of 25°C, 35°C, 40°C, and 45°C. As expected, higher fluid temperature and higher ambient temperature both reduced preheating time. However, the fluid temperature had a much stronger effect than ambient temperature in the range I studied. For example, at -40°C ambient, increasing the preheating fluid from 25°C to 35°C reduced the preheating time from 223 s to 146 s. In contrast, changing only the ambient temperature from -40°C to -30°C reduced the time from 223 s to 208 s. This confirms that the preheating fluid temperature is a dominant factor for cold-start performance of an electric vehicle battery pack.
| Ambient temperature | Preheating fluid temperature | Preheating time | Observation |
| -40°C | 25°C | 223 s | Baseline cold case |
| -40°C | 35°C | 146 s | Large improvement |
| -40°C | 40°C | Shorter | Phase change begins |
| -40°C | 45°C | Shortest | High temperature difference penalty |
| -30°C | 25°C | 208 s | Ambient effect is smaller |
| -20°C | 25°C | About 190 s | Moderate cold |
| -10°C | 25°C | About 170 s | Mild cold |
I observed that the phase change temperature of the composite material, approximately 37°C, creates an important thermal buffer. When the preheating fluid exceeds this temperature, part of the thermal energy is absorbed as latent heat, which limits the rate at which the maximum temperature rises. This can improve uniformity because the phase change region remains near 37°C while transferring heat to the colder solid regions. However, it also means that increasing the preheating fluid temperature above 37°C produces diminishing returns in preheating time reduction. At the same time, the temperature difference can increase sharply if the fluid temperature is too high. The results showed that a fluid temperature above 40°C can cause the maximum temperature difference to exceed 14°C, which is undesirable. Therefore, I recommend selecting the preheating fluid temperature according to the ambient severity: a moderate temperature for mild cold conditions and a higher temperature only when fast preheating is essential, with careful attention to uniformity.
Based on the complete set of simulations, I can summarize the optimal parameter combinations for the coupled microchannel–phase change material electric vehicle battery pack thermal management system. For high-temperature cooling at 35°C ambient and 4C discharge, I recommend the 63-channel layout, a coolant velocity of 0.1 m/s, a composite phase change material with 5% graphite by mass, and a counter-current flow arrangement. This configuration keeps the maximum temperature near 39.8°C and the maximum temperature difference near 2.8°C. For low-temperature preheating at -20°C ambient, I recommend the 111-channel layout, a preheating fluid velocity of 0.25 m/s, 5% graphite by mass, and a counter-current flow arrangement. This configuration reaches the 20°C target in approximately 194 s with a temperature difference of about 3.25°C. If the ambient is more severe, the preheating fluid temperature can be raised, but the designer must accept a larger temperature difference or adopt a staged preheating strategy.
| Operating mode | Recommended layout | Recommended flow | Recommended phase change material | Expected thermal result |
| High-temperature cooling | 63 microchannels | 0.10 m/s, counter-current | 5% graphite composite | Maximum temperature about 39.8°C; temperature difference about 2.8°C |
| Low-temperature preheating | 111 microchannels | 0.25 m/s, counter-current | 5% graphite composite | Preheating to 20°C in about 194 s; temperature difference about 3.25°C |
| Cold parking insulation | Existing coupled structure | No active flow | 5% graphite composite | Longer residence above 0°C than a system without phase change material |
The physical interpretation of these results is consistent. The microchannels provide a high heat transfer coefficient and allow active control of the electric vehicle battery pack temperature. The composite phase change material provides thermal inertia, reduces transient temperature spikes, and improves spatial uniformity when its thermal conductivity is sufficiently enhanced. The aluminum frame distributes heat laterally and connects the cells to both the phase change material and the microchannel fluid. The counter-current arrangement reduces axial thermal gradients because it avoids a monotonic temperature rise from one side of the electric vehicle battery pack to the other. The graphite fraction has a clear optimum because adding graphite improves conduction but reduces latent heat and increases electrical conductivity. The flow velocity also has an optimum because higher velocity improves convection but increases pumping power and can worsen temperature uniformity. The number of microchannels must be matched to the total flow rate; simply adding more channels does not guarantee better performance.
From a design perspective, the most important contribution of my work is the integration of cooling, insulation, and preheating into one architecture for an electric vehicle battery pack. Many conventional systems are designed for only one thermal objective. A cooling plate may perform well in summer but provide no useful preheating in winter. A positive temperature coefficient heater may preheat the electric vehicle battery pack but adds mass and complexity and does not help with high-temperature cooling. A phase-change-only system may buffer temperature during discharge but can become saturated during repeated high-rate operation. By coupling microchannels with phase change material, I obtained an active–passive system that can switch functions by changing the fluid temperature and flow direction. This functional flexibility is valuable for electric vehicle battery pack thermal management because real vehicles operate across a wide range of climates and duty cycles.
I also note several limitations of the present study. The discharge current was assumed constant, whereas a real electric vehicle battery pack experiences dynamic current profiles during acceleration, regenerative braking, and fast charging. This means that the heat generation rate in real operation is transient and spatially non-uniform. Future work should incorporate dynamic load profiles and electrochemical–thermal coupling to capture the effects of rapid current changes. In addition, I used a single-phase water coolant. Future studies could examine nanofluids, dielectric fluids, or refrigerant-based cooling for more aggressive fast-charging scenarios. The effect of contact resistance between the cells, the phase change material, and the aluminum frame should also be quantified experimentally. Manufacturing tolerances and aging of the phase change material may alter long-term performance. Finally, I optimized the parameters one at a time. More advanced multi-objective optimization methods, including surrogate models and evolutionary algorithms, could identify better trade-offs among maximum temperature, temperature uniformity, pumping power, mass, and cost for an electric vehicle battery pack.
In conclusion, I designed and numerically evaluated a microchannel–phase change material coupled thermal management system for an electric vehicle battery pack. The system uses an aluminum frame with embedded microchannels, a composite paraffin–expanded graphite phase change material, and a 4×4 arrangement of 18650 lithium-ion cells. I established a validated heat generation model, solved the coupled continuity, momentum, energy, and phase change equations, and performed mesh and time-step independence studies. I compared phase-change-only cooling with coupled microchannel–phase-change cooling and found that the coupled system reduced the maximum temperature from 41.98°C to 39.83°C and the maximum temperature difference from 3.50°C to 2.95°C. I identified 5% graphite by mass as the best composite formulation, 0.1 m/s as the balanced cooling velocity, and counter-current flow as the best arrangement for temperature uniformity. For cold conditions, I showed that the same electric vehicle battery pack architecture can preheat effectively. With 5% graphite, a 111-channel layout, a preheating velocity of 0.25 m/s, and counter-current flow, the electric vehicle battery pack reaches 20°C in about 194 s with a temperature difference of about 3.25°C. The phase change material also provides useful insulation during cold parking. These results demonstrate that a properly optimized microchannel–phase change material system can serve as a multifunctional thermal management solution for an electric vehicle battery pack, improving safety, uniformity, and cold-weather usability while retaining the potential for further optimization under dynamic real-world conditions.