The rapid global transition towards sustainable transportation has made electric vehicles (EVs) a cornerstone of modern energy and environmental policy. At the heart of every EV lies its battery pack, a complex assembly whose performance, safety, and longevity are paramount. Among various battery chemistries, lithium-ion batteries are the dominant choice for automotive applications due to their high energy density, long cycle life, and relatively good stability. However, their operational characteristics are highly sensitive to temperature. Efficient thermal management is therefore not a luxury but a critical necessity. A well-designed battery management system (BMS) must incorporate an effective thermal management strategy to maintain optimal operating conditions, prevent premature aging, and mitigate catastrophic risks like thermal runaway.
During operation, lithium-ion batteries generate significant heat due to irreversible (ohmic) and reversible (entropic) processes. The heat generation rate ($$ \dot{Q}_{gen} $$) can be expressed as a function of current (I), internal resistance (R), and entropic coefficient ($$ \frac{\partial U_{ocv}}{\partial T} $$):
$$ \dot{Q}_{gen} = I^2 R + I T \left( \frac{\partial U_{ocv}}{\partial T} \right) $$
Under dynamic driving conditions—characterized by acceleration, deceleration, and varying loads—the current (I) fluctuates widely. This leads to a highly variable and often intense heat generation profile, posing a substantial challenge for any thermal battery management system. Prolonged exposure to temperatures above 50°C accelerates degradation mechanisms, while large temperature gradients (ΔT) within a cell or across a pack lead to state-of-charge (SOC) imbalance, uneven aging, and reduced usable capacity. The consensus in the industry is to maintain the pack temperature between 20°C and 45°C and to keep the maximum temperature difference, ΔTpack, below 5°C for optimal performance and lifespan.
Various thermal management solutions have been explored, including air cooling, liquid cooling, phase change materials (PCM), and heat pipes. Recently, direct refrigerant cooling (or direct cooling) has gained considerable attention. This approach integrates the battery cooling system directly with the vehicle’s air conditioning (AC) refrigerant loop. A cold plate, through which refrigerant evaporates, is placed in direct or indirect contact with the battery pack. The latent heat absorbed during refrigerant evaporation provides extremely efficient cooling with low thermal resistance, compact structure, and lightweight design—all crucial advantages for EVs. This makes the direct cooling BMS particularly attractive for handling high thermal loads.
However, much of the existing literature on direct cooling battery management systems focuses on performance under static conditions: fixed ambient temperatures (e.g., 25°C, 30°C) and constant discharge/charge rates (e.g., 1C, 2C). While these studies provide valuable foundational data, they do not fully capture the complex, transient thermal behavior of batteries in real-world EV applications. The dynamic nature of urban and highway driving, with its frequent power surges and regenerative braking events, creates a thermal management scenario that is far more demanding. Evaluating a BMS under such realistic, dynamic operating profiles is essential for validating its design and ensuring real-world reliability.
This work presents an experimental investigation into the thermal performance of a direct refrigerant cooling battery management system for a 52 Ah lithium-ion battery pack under three representative dynamic driving conditions. We move beyond the standard metrics of maximum temperature (Tmax) and overall pack temperature difference (ΔTpack) to introduce a more granular analysis of temperature uniformity. Specifically, we examine longitudinal temperature differences within individual cells (ΔTcell) and transverse temperature gradients across the pack at different heights (ΔTh). This multi-dimensional analysis provides deeper insight into the thermal homogeneity challenges inherent in single-sided cooling configurations. Furthermore, when the baseline system struggles under the most aggressive dynamic profile, we propose and experimentally validate a simple yet effective enhancement—the integration of aluminum fins—to significantly improve both cooling efficiency and temperature uniformity. The findings offer practical guidance for the design and optimization of high-performance direct cooling battery thermal management systems.
Experimental Setup and Methodology
A dedicated experimental platform was constructed to emulate a vehicle’s direct refrigerant cooling circuit and evaluate its interaction with a battery pack under controlled dynamic loads. The core of the battery management system test setup was housed within a walk-in environmental chamber capable of maintaining precise ambient conditions (temperature range: -40°C to +50°C, ±0.5°C; humidity range: 40-90% RH, ±5%). For this study, ambient temperatures (Tamb) of 25°C, 30°C, and 35°C were selected to represent mild, warm, and hot climate scenarios, respectively.
The refrigerant circuit comprised a variable-speed compressor, a condenser, an electronic expansion valve (EEV), and the battery cold plate. A key feature was the ability to isolate the battery cooling loop from the cabin AC loop using solenoid valves, allowing independent operation focused solely on BMS performance. The cooling strategy employed a direct expansion approach: liquid refrigerant expanded through the EEV and entered the cold plate, where it evaporated by absorbing heat from the battery pack. The EEV opening was dynamically controlled by a Proportional-Integral-Derivative (PID) controller to maintain a constant superheat of 6°C at the cold plate outlet, ensuring efficient evaporator operation and protecting the compressor from liquid slugging.
The battery pack consisted of twelve prismatic 52 Ah lithium-ion cells connected in series (12S configuration), forming a nominal 38.4 V pack. The cells were arranged in a 2×6 matrix and mounted onto a single, flat aluminum cold plate. The cold plate (450 mm × 295 mm × 23 mm) featured an embedded serpentine copper tube (10 mm outer diameter) with 9 flow passes. To minimize thermal contact resistance and provide electrical insulation, a 2 mm layer of thermally conductive, electrically insulating silicone grease was applied between the cell bases and the cold plate surface. The entire assembly was enclosed in an insulated box to eliminate the effects of stray air currents from the environmental chamber, ensuring that heat removal occurred predominantly via the direct cooling battery management system.
To accurately capture the thermal response, seven K-type thermocouples were strategically attached to the battery pack and cold plate. Their locations were chosen to diagnose both global and local thermal behaviors, especially potential non-uniformities arising from the one-sided cooling and the refrigerant’s sensible heating along the flow path. The thermocouple positions are described in Table 1.
| Thermocouple ID | Location Description | Primary Measurement Purpose |
|---|---|---|
| T1, T4, T6 | On cell surface, 1 cm from bottom, along central axis of different cells in the flow direction. | Capture temperature at the strongest cooling region; monitor transverse gradient along flow. |
| T2, T5 | On cell surface, 9 cm from bottom (near top). | Capture longitudinal (vertical) temperature gradient when combined with T1/T4/T6. |
| T3 | On central cell surface, at geometric center. | Monitor temperature in a potentially thermally disadvantaged region. |
| T7 | On cold plate surface, 1 cm from refrigerant outlet. | Monitor cold plate temperature, detect local overheating or poor heat transfer. |
The battery pack was cycled using a programmable battery tester (Neware). Data for voltage, current, capacity, and all temperature channels were recorded at a frequency of 1 Hz. Prior to each dynamic discharge test, the pack was charged to 100% State of Charge (SOC) using a standard constant-current constant-voltage (CC-CV) protocol.
Definition of Dynamic Operating Conditions and Performance Metrics
To simulate real-world driving, three distinct dynamic operating conditions were defined, moving beyond simple constant-current tests. These profiles represent different segments of a driving cycle, as summarized in Table 2.
| Operating Condition | Simulated Driving Scenario | Discharge Profile | Total Time |
|---|---|---|---|
| Steady Operation | Highway cruising at constant speed. | Constant 0.5C (26 A) discharge. | 2400 s |
| Alternating Load Operation | Urban driving with frequent stops and starts. | Cycle: 1C (52 A) for 60 s, then 0.5C (26 A) for 540 s. Repeated 6 times. | 3600 s |
| Progressive Acceleration | Aggressive driving or highway on-ramp acceleration. | Sequential: 0.5C (26 A) for 300 s, then 1C (52 A) for 300 s, then 1.5C (78 A) for 300 s. | 900 s |
The thermal performance of the direct cooling battery management system was evaluated using a comprehensive set of four key metrics. These metrics provide a holistic view, extending from overall pack limits to internal uniformity:
- Maximum Pack Temperature (Tmax): The highest temperature recorded among all battery surface thermocouples (T1 to T6) during the test. This is the primary safety and degradation indicator.
$$ T_{max} = \max(T_1, T_2, T_3, T_4, T_5, T_6) $$ - Overall Pack Temperature Difference (ΔTpack): The difference between the highest and lowest temperatures on the battery pack surface. It indicates the global thermal gradient.
$$ \Delta T_{pack} = T_{h} – T_{l} $$
where $$ T_h = \max(T_{1-6}) $$ and $$ T_l = \min(T_{1-6}) $$. - Cell Longitudinal Temperature Difference (ΔTcell): The vertical temperature gradient within a single cell. For a cell instrumented with a bottom (Tb) and a top (Tt) thermocouple, it is calculated as:
$$ \Delta T_{cell} = T_t – T_b $$
This metric highlights the effectiveness of heat conduction away from the cell’s core to the cooled base. - Pack Transverse Temperature Difference at Height h (ΔTh): The temperature variation across the pack at a specific height from the cold plate. For example, at the 1 cm height, it would be:
$$ \Delta T_{1cm} = \max(T_1, T_4, T_6) – \min(T_1, T_4, T_6) $$
This metric reveals uneven cooling along the refrigerant flow path or due to pack geometry.
Results and Discussion: Baseline System Performance
1. Steady Operation Condition
Under the steady 0.5C discharge, representing a mild thermal load, the direct cooling BMS demonstrated effective and stable performance across all three ambient temperatures. The results are consolidated in Table 3.
| Ambient Temp. (Tamb) | Peak Tmax (°C) | Final Tmax (°C) | Final ΔTpack (°C) | Final ΔTcell (Typical, °C) | ΔTh (1cm & 9cm, °C) |
|---|---|---|---|---|---|
| 25°C | 26.7 (at 584 s) | 25.2 | 2.8 | < 2.0 | < 1.0 |
| 30°C | 30.6 (at 344 s) | 27.9 | 3.0 | < 2.5 | < 1.0 |
| 35°C | 35.9 (at 209 s) | 33.4 | 3.5 | < 3.0 | < 1.0 |
The system dynamics were clear: after an initial temperature rise, the cooling power of the evaporating refrigerant balanced the battery’s heat generation, causing temperatures to peak and then gradually decrease. The final Tmax values were only slightly above ambient, and the final ΔTpack was well within the 5°C target. Both longitudinal (ΔTcell) and transverse (ΔTh) differences were minimal, confirming that for low, constant loads, this single-sided direct cooling battery management system provides adequate and uniform cooling.
2. Alternating Load Operation Condition
This urban-driving profile presented a more challenging, pulsating thermal load. The BMS successfully managed the recurring 1C load spikes, as shown by the characteristic sawtooth pattern in Tmax. Key results are in Table 4.
| Ambient Temp. (Tamb) | Peak Tmax after 2nd 1C pulse (°C) | Final Tmax (°C) | Final ΔTpack (°C) | Final ΔTcell (°C) | ΔTh (1cm & 9cm, °C) |
|---|---|---|---|---|---|
| 25°C | 26.6 | 23.8 | 6.0 | < 2.0 | ~1.0 |
| 30°C | 31.0 | 27.5 | 6.7 | < 2.0 | ~1.0 |
| 35°C | 35.7 | 31.0 | 6.6 | < 2.0 | ~1.0 |
The system demonstrated good thermal recovery during the low-load periods, bringing temperatures down significantly. While the instantaneous ΔTpack during the high-load pulses was higher than in steady operation, it showed a stabilizing trend. Crucially, the internal uniformity metrics (ΔTcell and ΔTh) remained excellent, indicating that the thermal mass of the system and the cooling response were sufficient to prevent large internal gradients from developing, even under this dynamic profile. The direct cooling BMS proved capable of handling urban driving cycles.
3. Progressive Acceleration Condition
This aggressive profile exposed the primary limitation of the baseline design. The stepwise increase in discharge rate created a steep, cumulative thermal load that overwhelmed the cooling capacity, especially at higher ambient temperatures. The results, detailed in Table 5, reveal a critical failure to meet thermal management targets.
| Ambient Temp. (Tamb) | Final Tmax (°C) | Final ΔTpack (°C) | Final ΔTcell (Cell-10) (°C) | Final ΔTh at 1cm (°C) | Status |
|---|---|---|---|---|---|
| 25°C | 34.1 | 7.8 | 6.3 | 3.5 | Marginal |
| 30°C | 42.1 | 13.6 | 10.5 | 4.8 | Unsafe |
| 35°C | 49.8 | 16.5 | 11.2 | 5.9 | Critical Failure |
At 35°C ambient, the final Tmax of 49.8°C exceeds the safe upper limit, posing a direct risk of accelerated aging. More critically, the ΔTpack of 16.5°C and the individual cell longitudinal difference (ΔTcell) of over 11°C are unacceptable. These large gradients indicate severe thermal inhomogeneity. The root cause is the poor longitudinal (vertical) heat conduction within the cells coupled with the single-sided cooling. Heat generated in the upper portions of the cells cannot be transferred quickly enough down to the cold plate. Additionally, the transverse gradient (ΔTh) increased, likely due to refrigerant warming along the serpentine path, leaving downstream cells slightly warmer. This condition clearly demonstrates that the baseline direct cooling battery management system, while effective for moderate loads, requires enhancement for high-stress, dynamic operations typical of performance driving or hot climates.
System Enhancement: Integration of Thermal Fins
To address the identified weakness—poor longitudinal heat spreading—a passive enhancement was introduced: aluminum fins. A single, thin aluminum plate (360 mm × 179 mm × 2 mm, k ≈ 202.4 W/m·K) was attached to the top of the cold plate, sandwiched between the cold plate and the battery cells with thermal interface material. The fins serve as extended surfaces, effectively increasing the thermal contact area between the cold plate and the battery pack. Their primary function is to facilitate lateral (in-plane) and longitudinal heat conduction, helping to draw heat away from the upper regions of the cells and distribute it more evenly across the cooled surface.
The modified system was tested under the most challenging condition: Progressive Acceleration at 35°C ambient. The comparative results against the baseline are stark and presented in Table 6.
| Performance Metric | Baseline System | System with Aluminum Fins | Improvement |
|---|---|---|---|
| Final Tmax | 49.8 °C | 40.9 °C | ↓ 8.9 °C |
| Final ΔTpack | 16.5 °C | 5.0 °C | ↓ 11.5 °C |
| Final ΔTcell (Cell-10) | 11.2 °C | 4.6 °C | ↓ 6.6 °C |
| Final ΔTh at 1cm | 5.9 °C | 1.2 °C | ↓ 4.7 °C |
The enhancement delivered by the fins is transformative across all metrics. The reduction in Tmax by nearly 9°C brings the pack temperature safely back within the optimal range. The most dramatic improvement is in temperature uniformity: ΔTpack is reduced from a dangerous 16.5°C to an excellent 5.0°C, meeting the critical design target. This is a direct consequence of the significantly improved longitudinal conduction, as evidenced by ΔTcell dropping from 11.2°C to 4.6°C. The fins effectively created a more thermally “connected” system, allowing heat from the top of the cells to be conducted sideways and downwards more efficiently.
Interestingly, the transverse uniformity also improved substantially (ΔTh from 5.9°C to 1.2°C). The fins likely helped to homogenize the base temperature distribution, mitigating the “inlet-to-outlet” temperature rise effect of the refrigerant. This shows that the fins addressed not only the primary longitudinal issue but also secondary transverse non-uniformities, leading to a more globally isothermal pack. This simple modification proves that the performance envelope of a direct refrigerant cooling battery management system can be significantly expanded with well-designed passive enhancements, making it robust enough to handle extreme dynamic loads.
Conclusion
This experimental study provides a comprehensive evaluation of a direct refrigerant cooling battery management system under realistic, dynamic operating conditions for electric vehicles. By moving beyond static tests, we captured the transient thermal challenges that truly define BMS performance in the real world. The key findings are:
- The baseline direct cooling BMS is highly effective for mild to moderately dynamic loads. It successfully maintained safe temperatures and good uniformity during steady cruising (0.5C) and urban-style alternating load conditions, even in a warm (35°C) environment.
- Aggressive, sustained high-power dynamics, simulated by the Progressive Acceleration profile, revealed a critical design limitation. Under a 35°C ambient condition, the baseline system failed, with the battery pack reaching a maximum temperature of 49.8°C and developing a catastrophic temperature difference of 16.5°C, primarily due to insufficient longitudinal heat conduction from the cells to the single-sided cold plate.
- The integration of simple aluminum fins as a passive thermal enhancement proved to be a highly effective solution. For the same severe operating condition, the fins-enhanced system reduced the maximum temperature to 40.9°C and, more importantly, slashed the overall pack temperature difference to 5.0°C—meeting the critical industry benchmark. This was achieved by drastically improving longitudinal heat spreading (ΔTcell reduced from 11.2°C to 4.6°C) and also improving transverse temperature uniformity.
The research underscores that while direct refrigerant cooling is a potent technology for EV battery thermal management, its design must be validated against high-stress dynamic profiles, not just steady-state benchmarks. The multi-dimensional temperature analysis (Tmax, ΔTpack, ΔTcell, ΔTh) provides a far richer understanding of thermal homogeneity than traditional metrics alone. The successful application of fins demonstrates that hybrid approaches, combining the high cooling power of direct refrigerant systems with passive thermal spreading elements, offer a promising path toward developing robust, reliable, and safe thermal battery management systems capable of supporting the next generation of high-performance electric vehicles.