The performance, safety, and longevity of an electric vehicle (EV) are fundamentally tied to the effective thermal management of its core energy storage component—the EV battery pack. Maintaining all cells and modules within a narrow optimal temperature range is critical. While controlling peak temperature is a common focus, the uniformity of temperature distribution within the pack is equally vital. Significant temperature gradients can lead to accelerated localized aging, capacity mismatch between cells, and ultimately, safety risks like thermal runaway. Therefore, ensuring excellent thermal uniformity is a paramount design objective for any EV battery pack thermal management system (BTMS).
In the design of passenger EV battery packs, the arrangement of battery modules within the available underbody space follows several paradigms. The three primary arrangements are single-layer, full duplex (or double-layer), and local duplex layouts. The local duplex arrangement, which strategically places modules in a double-layer stack in certain areas (e.g., under rear seats) while keeping others in a single layer, has gained widespread adoption. This configuration offers a compelling compromise: it increases the overall energy density of the EV battery pack without excessively raising the vehicle’s floor height, thereby preserving valuable cabin space. However, empirical data from field failure analyses and fire incident databases suggests a concerning trend: EV battery packs employing local duplex module arrangements are disproportionately represented in cases of spontaneous combustion or thermal runaway events. This correlation strongly indicates that this otherwise space-efficient layout may introduce inherent challenges to achieving adequate thermal uniformity, potentially creating hotspots and accelerating degradation in specific modules. This paper delves into a detailed numerical investigation of the thermal behavior of a local duplex configured EV battery pack. We aim to characterize its temperature distribution under various operational loads, identify the root causes of thermal non-uniformity, and evaluate the effectiveness of key liquid cooling system parameters in mitigating these disparities.
1. Numerical Modeling of the EV Battery Pack
To accurately simulate the complex thermal-fluid dynamics within an operating EV battery pack, a high-fidelity three-dimensional computational model is essential. Many existing studies simplify the system to a “modules + cold plates” setup. While useful, such models often neglect the influence of the pack enclosure, internal air cavities, external environmental interactions, and auxiliary components, which can significantly impact the overall thermal and humidity environment. To capture a more realistic scenario, we developed a comprehensive 3D CFD model of a commercially representative EV battery pack with a local duplex layout using COMSOL Multiphysics software.
1.1. Geometry and Configuration of the Target EV Battery Pack
The subject of this study is a ternary lithium-ion EV battery pack with a nominal voltage of 346.56 V and a capacity of 177 Ah. Its internal architecture features a local duplex arrangement. The 24 modules are logically divided into four groups for clarity in analysis:
- Front Modules: 8 modules in a single-layer arrangement.
- Central Modules: 6 modules in a single-layer arrangement.
- Rear Upper Modules: 5 modules forming the top layer of the duplex section.
- Rear Lower Modules: 5 modules forming the bottom layer of the duplex section.
Each module is underlaid with an aluminum cooling plate in tight thermal contact. The liquid cooling circuit employs a two-stage parallel flow design to reduce system pressure drop. A main coolant pipe branches into four primary channels supplying the front, central, and rear (both upper and lower) sections. Each primary channel then feeds multiple parallel cooling plate channels. The pack enclosure is equipped with pressure-equalizing vents (simplified as moisture-permeable boundaries in the model), which are the sole mass transfer paths between the internal environment of the EV battery pack and the outside atmosphere.
The geometric model was simplified by removing fine fillets, small fixtures, and wiring harnesses to reduce computational cost without compromising the primary thermal-fluid pathways. Key components like the thin enclosure walls and cooling plates were modeled using the “Thermal Thin Layer” feature, which is appropriate for layers with high in-plane thermal conductivity. The modules themselves were homogenized as solid blocks with uniform volumetric heat generation, a valid simplification for studying module-level thermal uniformity. A free tetrahedral mesh was generated, with particular refinement in the module and complex cooling plate regions to ensure accuracy. The final mesh consisted of approximately 1.21 million elements.
1.2. Governing Equations and Numerical Methods
The model couples conductive heat transfer in solids, convective heat transfer in the coolant and internal air, and fluid flow. The coolant flow within the plates is laminar, as confirmed by a Reynolds number calculation. The overall heat transfer is governed by the energy conservation equation. The primary heat source is the Joule heating (irreversible heat) generated by the battery modules during operation. The total heat generation rate $$Q_{\text{gen}}$$ for a battery can be expressed as:
$$Q_{\text{gen}} = Q_{\text{jou}} + Q_{\text{re}} = I(U_{\text{ocv}} – U_t) – IT_{\text{abs}}\frac{\partial U_{\text{ocv}}}{\partial T}$$
where $$I$$ is the current, $$U_{\text{ocv}}$$ is the open-circuit voltage, $$U_t$$ is the terminal voltage, and $$T_{\text{abs}}$$ is the absolute temperature. The second term, the reversible reaction heat $$Q_{\text{re}}$$, is often negligible compared to the irreversible Joule heating $$Q_{\text{jou}}$$ for the purposes of thermal management analysis. Therefore, the heat generation is approximated as $$Q_{\text{gen}} \approx I(U_{\text{ocv}} – U_t)$$. Based on cell-level testing, the average volumetric heat generation for modules under different discharge rates (C-rates) was calculated and applied uniformly, as summarized in Table 1.
| Discharge Rate (C) | Heat Generation per Module (W) |
|---|---|
| 1/3 | 288.00 |
| 1 | 1003.80 |
| 2 | 2463.84 |
Boundary conditions included natural convection on the outer surfaces of the EV battery pack enclosure, a moisture flux boundary for the vents, and defined inlet temperature/flow rate for the coolant. The initial temperature for the entire pack was set equal to the ambient temperature. Material properties for all components (modules, aluminum casing/cold plates, coolant, plastics) were assigned based on standard values at 30°C. The coupled equations were solved using a finite element method with an implicit time-stepping scheme for transient analyses.
1.3. Model Validation
The accuracy of the numerical model was validated against experimental data. The physical EV battery pack was instrumented with thermocouples at key locations (S-box, BMU, central module surface, rear module surface) and placed in a natural outdoor environment under a sun shield for 24 hours. The ambient temperature profile was recorded and used as the environmental input for a corresponding 24-hour simulation with the pack in a non-operating (no heat generation) state. The comparison between simulation results and experimental measurements at the four probe locations showed excellent agreement in temporal trends. The quantitative error was minimal, with the maximum absolute error at any point being 1.55°C and the average error across all probes being less than 0.8°C. This close correlation confirmed that the model reliably captures the heat transfer dynamics of the complete EV battery pack system, establishing a solid foundation for subsequent parametric and operational studies.
2. Thermal Uniformity Under Fast-Charging and Discharge Scenarios
Using the validated model, we investigated the thermal behavior of the local duplex EV battery pack under high-stress operational conditions, starting with fast charging, which is a common trigger point for thermal incidents in real-world data.
2.1. Temperature Distribution During Fast Charging (1C Rate)
Simulating a 1C fast-charge process revealed distinct and persistent temperature distribution patterns. The coolant flow direction (right inlet to left outlet) created a clear thermal gradient. High-temperature zones (above 33°C) consistently developed on the left and upper sections of modules, particularly in the central and rear upper modules. Conversely, low-temperature zones (below 27°C) were found on the right and lower sections, with the most extensive cool area in the rear lower modules.
As the charging progressed from 20 to 60 minutes, the high-temperature regions expanded across most modules. Critically, the thermal disparity within the duplex rear section intensified. The low-temperature zone in the rear lower modules grew larger, while the temperature of the rear upper modules continued to rise more sharply than others. This is a direct consequence of the shared cooling resource. The rear upper modules are cooled only from below by a cold plate that also serves to cool the rear lower modules from above. This configuration leads to a competition for cooling capacity. The lower modules benefit from an additional cooling plate on their underside, resulting in stronger heat dissipation. Consequently, a significant vertical temperature gradient develops within the duplex stack, with the upper layer becoming a persistent hotspot. This phenomenon highlights a fundamental thermal challenge in local duplex designs for EV battery packs: unequal cooling intensity between vertically stacked modules.
2.2. Transient Thermal Behavior Under Various Discharge Rates
We next analyzed the pack’s thermal response during discharge at 1/3C, 1C, and 2C rates. The cooling system was controlled by a simulated BMS logic: coolant flow (10 L/min, 25°C inlet) initiated when the maximum module temperature reached 35°C and stopped when it fell below 30°C. The initial state-of-charge (SOC) was 100% for all modules.
The results, plotted as the average temperature for each module group over time, revealed consistent ranking and dynamics. Upon activation of cooling, the module temperatures re-ordered predictably: Rear Lower (coolest) < Front < Central < Rear Upper (hottest). The inflow of coolant effectively lowered the overall pack temperature but simultaneously increased the instantaneous temperature difference (ΔT) between the hottest (rear upper) and coolest (rear lower) modules. This ΔT spike upon cooling activation is a critical transient phase for thermal uniformity.
The relationship between discharge rate and cooling system operation was also clear. Higher discharge rates (e.g., 2C) caused the coolant pump to activate earlier and, despite the shorter discharge period, to run for a longer total duration to bring the pack temperature back down. This implies that for an EV battery pack with a given energy content, operating at higher power demands increases the total energy consumption of the cooling system’s pumps and chillers, reducing overall vehicle efficiency. After cooling stopped, the temperatures of all modules gradually converged as heat redistributed within the pack via conduction and internal convection.
3. Parametric Influence of the Liquid Cooling System
Given the observed thermal non-uniformity, we systematically evaluated the influence of two key liquid cooling system parameters—coolant inlet temperature and flow rate—on the thermal performance of the local duplex EV battery pack. The goal was to assess whether standard cooling system adjustments could effectively mitigate the inherent thermal gradients.
3.1. Effect of Coolant Inlet Temperature
The coolant inlet temperature was varied from 15°C to 25°C in 2°C increments, with a fixed flow rate of 10 L/min under a 1C discharge. The results are summarized in Table 2 and shown graphically.
| Coolant Inlet Temp. (°C) | Pack T_max (°C) | Module ΔT_max (°C) | Average Module Temp. Drop (°C) |
|---|---|---|---|
| 25 | 45.6 | 8.7 | Base |
| 23 | 43.8 | 8.6 | ~1.8 |
| 21 | 41.9 | 8.5 | ~3.7 |
| 19 | 40.1 | 8.4 | ~5.5 |
| 17 | 38.2 | 8.4 | ~7.4 |
| 15 | 36.4 | 8.4 | ~9.3 |
The analysis yielded two key findings. First, reducing the inlet temperature is highly effective at lowering the absolute peak temperature (T_max) of the EV battery pack. A 10°C reduction in coolant temperature (from 25°C to 15°C) led to an approximately 9.2°C drop in T_max. This strong linear correlation offers a direct lever for controlling worst-case temperatures during high-load events.
Second, and more importantly for this study, the coolant inlet temperature has a minimal impact on improving thermal uniformity. The maximum temperature difference (ΔT_max) between modules decreased by only about 0.3°C over the same 10°C coolant temperature range. Lowering the coolant temperature simply shifts the entire temperature field of the EV battery pack downward without significantly altering the relative gradients between differently cooled modules (like the rear upper vs. rear lower). Therefore, while valuable for peak temperature suppression, adjusting the chiller setpoint is not a viable strategy for solving the fundamental thermal uniformity issue in a local duplex EV battery pack.
3.2. Effect of Coolant Inlet Flow Rate
The coolant flow rate was varied from 5 L/min to 20 L/min, and its effect was evaluated across three different discharge rates (1/3C, 1C, 2C). The results, focusing on T_max and ΔT_max, are presented in Table 3.
| Discharge Rate | Flow Rate (L/min) | Pack T_max (°C) | Module ΔT_max (°C) | Observation |
|---|---|---|---|---|
| 2C | 5 | 48.9 | 10.1 | Strong initial benefit, diminishing returns. Improves both T_max and ΔT. |
| 10 | 42.9 (-6.0) | 8.7 (-1.4) | ||
| 15 | 40.8 (-2.1) | 8.1 (-0.6) | ||
| 20 | 39.7 (-1.1) | 7.9 (-0.2) | ||
| 1C | 5 | 46.1 | 8.9 | Moderate benefit, faster saturation. ΔT improvement is limited. |
| 10 | 45.6 (-0.5) | 8.7 (-0.2) | ||
| 15 | 45.3 (-0.3) | 8.6 (-0.1) | ||
| 20 | 45.1 (-0.2) | 8.6 (~0.0) | ||
| 1/3C | 5 | 35.8 | 5.2 | Negligible effect. Thermal load is easily managed. |
| 10 | 35.7 (~0.0) | 5.2 (~0.0) | ||
| 15 | 35.7 (~0.0) | 5.2 (~0.0) | ||
| 20 | 35.7 (~0.0) | 5.2 (~0.0) |
The influence of flow rate is tightly coupled to the thermal load (discharge rate) of the EV battery pack:
- At High Discharge Rates (2C): Increasing flow rate provides significant benefits. Both T_max and ΔT_max decrease substantially when moving from 5 to 15 L/min. This is because the higher convective heat transfer coefficient associated with increased flow rate more effectively removes concentrated heat. However, the improvement shows clear diminishing returns; the gain from 15 to 20 L/min is much smaller than from 5 to 10 L/min. The system approaches a performance saturation point where further increases in pump power (and associated parasitic loss) yield minimal thermal benefit.
- At Medium Discharge Rates (1C): The benefit of increasing flow rate is less pronounced and saturates more quickly. The reduction in ΔT_max is particularly marginal, indicating that higher flow does little to balance the cooling intensity between modules.
- At Low Discharge Rates (1/3C): The thermal load is low enough that even a modest flow rate (5 L/min) is sufficient to maintain temperatures. Increasing the flow rate has virtually no effect on either T_max or ΔT_max. This highlights that operating the cooling pump at high speeds during mild conditions is energetically wasteful for the EV battery pack thermal management system.
The key takeaway is that while increasing coolant flow can help manage temperatures and slightly improve uniformity under extreme loads, it does not address the root cause of the non-uniformity in a local duplex layout and becomes ineffective at lower loads.
4. Discussion on Improving Thermal Uniformity in Local Duplex EV Battery Packs
The numerical analysis conclusively demonstrates that the local duplex module arrangement creates inherent thermal management challenges that cannot be sufficiently resolved by simply tuning traditional cooling system parameters like inlet temperature and flow rate. The core issues are fixed flow direction leading to longitudinal gradients and asymmetric cooling intensity in duplex stacks leading to vertical gradients. To meaningfully improve the thermal uniformity and, consequently, the safety and longevity of such an EV battery pack, design modifications at the module arrangement and cooling architecture level are necessary. Below are proposed directions based on the findings.
4.1. Mitigating Longitudinal Temperature Gradients
The “left-high, right-low” pattern stems from coolant warming along its fixed path. Countermeasures must disrupt this unidirectional heating effect.
- Dual Inlet/Dual Outlet or Z-flow Designs: Redesigning the cooling plate manifold to have coolant inlets on both ends or implementing a Z-shaped flow path can drastically reduce the overall temperature rise of the coolant across the plate, leading to a more uniform base temperature for all modules.
- Active Flow Reversal: Incorporating electronically controlled valves to periodically reverse the direction of coolant flow could help average out the longitudinal gradient over time. While adding complexity, this would ensure that no module is permanently located in the disadvantaged “hot outlet” region.
4.2. Mitigating Vertical Temperature Gradients in Duplex Stacks
The “upper-hot, lower-cool” disparity in the duplex section is due to the shared and unequal cooling resource. Solutions must aim to balance the cooling intensity.
- Dedicated Cooling Layers: The most effective solution is to provide independent cooling plates for the upper and lower layers of the duplex stack, each with its own controlled flow path. This decouples their thermal management and allows for balanced cooling.
- Insulation and Flow Control: If a shared interlayer cooling plate must be used, two supplemental measures can help:
- Apply a thermal insulation layer on the top surface of the lower module. This forces more of the heat from the lower module to be dissipated through its dedicated bottom cooling plate, reducing the thermal “stealing” from the upper module’s cooling resource.
- Implement flow control valves or design channel restrictions to actively manage and balance the proportion of total coolant flow directed to the upper-layer cooling channels versus the lower-layer ones, ensuring equitable cooling power distribution.
4.3. System-Level Optimization Strategy
The final design of an EV battery pack must balance thermal performance with cost, weight, and complexity. A holistic optimization strategy is recommended:
- Use advanced CFD models (like the one demonstrated here) in the early design phase to map thermal gradients for different module layouts and cooling circuit architectures.
- Formulate a multi-objective optimization problem minimizing maximum temperature (T_max), maximum temperature difference (ΔT_max), and pumping power. Design variables can include flow channel geometry, inlet/outlet positions, and the placement of any insulating layers.
- Implement an intelligent, predictive BMS controller. Instead of simple on/off thresholds, the controller could use models and sensor data to pre-emptively adjust cooling based on driving style, requested power, and ambient conditions, optimizing for both uniformity and energy efficiency.
5. Conclusion
This comprehensive numerical study has provided critical insights into the thermal uniformity challenges associated with locally duplexed module arrangements in electric vehicle battery packs. The key conclusions are as follows:
- The local duplex configuration inherently creates non-uniform temperature distributions. During fast-charging and high-rate discharge, significant gradients develop both longitudinally (due to coolant flow direction) and, more critically, vertically within the duplex stack. The upper-layer modules in the duplex section consistently become hotspots due to inferior cooling intensity compared to the lower-layer modules.
- Standard thermal management control parameters have limited effectiveness in resolving these inherent gradients. Lowering the coolant inlet temperature is effective for reducing the absolute peak temperature of the EV battery pack but has a negligible impact on improving module-to-module temperature uniformity. Increasing the coolant flow rate improves both peak temperature and uniformity under high thermal loads (e.g., 2C discharge) but exhibits strong diminishing returns and is ineffective under low to moderate loads.
- The thermal non-uniformity is a structural issue rooted in the layout and cooling architecture. Therefore, achieving a substantial improvement requires design interventions at the component and system level, not just control parameter adjustments. Proposed solutions include redesigning coolant flow paths (e.g., dual inlets, Z-flow), implementing active flow control or reversal, and, for the duplex section, employing dedicated cooling layers or strategic insulation to balance cooling intensity.
This analysis underscores the importance of integrating detailed thermal modeling early in the design process of an EV battery pack, especially when employing space-efficient but thermally challenging layouts like the local duplex. By identifying and addressing these thermal uniformity pitfalls, designers can develop more robust, safe, and durable battery systems, ultimately enhancing the reliability and consumer confidence in electric vehicles.