In the context of China’s carbon peak and carbon neutrality strategic goals, renewable energy sources such as wind and solar power are gradually replacing non-renewable energy. The instability, volatility, and intermittency of these sources necessitate effective energy storage solutions. Among emerging electrochemical storage technologies, lithium battery energy storage systems (energy storage system) have gained widespread attention due to their high energy density, long cycle life, high energy efficiency, and decreasing costs. However, during charging and discharging, especially under high-rate conditions, lithium batteries generate significant heat due to electrochemical reactions and internal ohmic resistance. The optimal operating temperature for lithium batteries is between 30 °C and 45 °C to ensure safety and longevity. This poses a critical challenge for thermal management, particularly as the energy density of energy storage system continues to increase.
Prior to 2022, air cooling was the dominant thermal management method for energy storage system due to its simplicity and low cost. However, the low specific heat capacity of air limited its heat dissipation capability. Subsequently, indirect liquid cooling using bottom cold plates became widely adopted because of the higher specific heat and thermal conductivity of liquid coolants. These approaches typically suit moderate power scenarios (0.25 P to 0.5 P). For grid frequency regulation, energy storage system often requires 1 P or higher power, increasing heat generation by two to three times. Under such conditions, bottom liquid cooling alone becomes inadequate.
To address high-power thermal management, two main research directions have emerged: direct liquid cooling (immersion cooling) and advanced indirect liquid cooling. Direct immersion cooling using dielectric fluids or phase change materials (PCMs) offers increased heat transfer area and eliminates contact resistance, but poses structural sealing challenges and limited heat removal for non-flowing configurations. Indirect methods, such as dual-side liquid cooling or large-area liquid cooling (e.g., CATL’s Qilin battery), face issues like condensation on top cold plates and structural reliability problems due to cell swelling.
In this study, we propose a novel gas-liquid coupled dual-phase thermal management design for energy storage system. The concept retains the proven bottom liquid cooling plate while introducing forced air cooling on the top of the cells. This approach avoids condensation issues associated with top liquid cooling and increases the effective heat transfer area by utilizing the entire top surface of the cells instead of only the pole tabs. Figure below illustrates the concept: gaseous working medium (forced air) flows across the top of the cells, absorbs heat, then passes through the bottom liquid cooling plate equipped with fins to exchange heat with the coolant, and finally returns as cooled air to the top.
We built a simulation model in Flotherm-XT for a 1P48S battery pack (48 cells in series) with dimensions 900 mm × 670 mm × 225 mm. The pack was subjected to a 3C charge/discharge cycle. Key parameters used in the simulation are summarized in Table 1.
| Component | Density (kg/m³) | Specific heat (J/(kg·K)) | Thermal conductivity (W/(m·K)) | Dynamic viscosity (Pa·s) |
|---|---|---|---|---|
| Cell | 2205.5 | 1008.7 | X:17.8, Y:4.2, Z:2.3 | / |
| Coolant | 1077 | 3473 | 0.4 | 0.0046 |
| Cold plate (Al) | 2700 | 900 | 177 | / |
| Thermal interface material | 2000 | 1200 | 1.5 | / |
| Busbar (Al) | 2702 | 903 | 237 | / |
| Enclosure (steel) | 7850 | 470 | 52 | / |
| End plate | 2760 | 963 | 96.2 | / |
| Aerogel insulation | 320 | 1000 | 0.025 | / |
We used a structured mesh with adaptive refinement near fins and channels. Grid independence was verified: when the mesh count reached 2 million, the maximum cell temperature change was less than 1%. The cell heat generation power as a function of time during the 3C cycle was derived from experimental data and used as input.
First, we compared the conventional bottom liquid cooling pack with our gas-liquid coupled pack (without any fin optimization). In the conventional pack, the maximum cell temperature reached 54 °C with a top-bottom temperature difference of 20 °C per cell and a maximum pack temperature difference of 15 °C. In the gas-liquid coupled pack, the maximum temperature dropped to 46.5 °C, and the pack temperature difference reduced to 14 °C. The air flow velocity at the top was 5.7 m/s. The improvement is attributed to the increased heat transfer area on the cell top surface (more than twice that of only pole contact in top liquid cooling). However, a temperature gradient along the flow direction still existed because the air heated up as it moved from front to back without intermediate cooling.
To overcome this, we introduced an “alternating gas channel” design. The channel is divided into two layers: in the first half of the pack width, cold air contacts the cells; after warming up, it returns to the upper layer, while fresh cold air from the upper layer descends to cool the second half. This design maintains a more uniform cooling effect. With the alternating channel, the maximum temperature decreased by about 1 °C and the temperature difference by 0.8 °C compared to the single-flow design.
Next, we optimized the fin spacing in the top alternating channels. The heat transfer performance was analyzed using the NTU (Number of Transfer Units) method. The total heat transfer rate Q is given by:
$$ Q = \varepsilon \, m C_p (T_f – T_w) $$
where ϵ is the heat exchanger effectiveness, m is mass flow rate, Cp is specific heat of air, Tf is inlet air temperature, and Tw is cold plate wall temperature. The effectiveness for a single-pass crossflow is:
$$ \varepsilon = 1 – \exp(-\text{NTU}) $$
with:
$$ \text{NTU} = \frac{h A}{m C_p} $$
The convection coefficient h depends on flow regime. For laminar flow, h ∝ u0.5 d-0.5, where u is velocity and d is fin spacing. Assuming constant fan power, we derived that u ∝ d and m ∝ d2/(t+d). Substituting into NTU yields:
$$ \text{NTU} \propto u^{-0.5} d^{-1.5} (2w + d) $$
where w is fin height. The total heat transfer Q is proportional to ϵ×m. At very small d, ϵ ≈ 1, Q ∝ m which decreases with d; at very large d, ϵ ≈ NTU, Q ∝ (2w+d)/(t+d) which decreases with d. Thus, an optimal d exists. Simulations for d from 3 mm to 18 mm showed that the optimal spacing for top straight fins was 3 mm, yielding the lowest cell top-bottom temperature difference (13.5 °C) and the best overall pack temperature uniformity. The results are summarized in Table 2.
| Fin spacing (mm) | Max cell temperature (°C) | Pack temperature difference (°C) | Cell top-bottom ΔT (°C) |
|---|---|---|---|
| 3 | 46.5 | 8.5 | 13.5 |
| 6 | 47.2 | 9.0 | 14.2 |
| 11 | 46.8 | 6.0 | 14.0 |
| 18 | 48.0 | 9.5 | 15.0 |
After optimizing the top channel, we focused on enhancing the heat exchange between the heated air returning to the bottom cold plate and the coolant. We replaced the straight fins on the bottom cold plate with wavy (corrugated) fins. The wavy profile forces flow separation and secondary flows, increasing turbulence and effective heat transfer area. We optimized both the fin spacing and the wave period (w) of the wavy fins. For straight fins, the best performance occurred at 3 mm spacing. For wavy fins, we varied spacing from 3 mm to 10 mm; the optimal was 6 mm. At this spacing, the maximum cell temperature was 46 °C and pack temperature difference was 8.5 °C — a reduction of 3 °C and 0.5 °C respectively compared to straight fins.
We then varied the wave period p of the wavy fins (from 50 mm to 130 mm) while keeping spacing at 6 mm. The results are shown in Table 3.
| Wave period (mm) | Max cell temperature (°C) | Pack temperature difference (°C) |
|---|---|---|
| 50 | 47.5 | 9.8 |
| 70 | 46.8 | 9.0 |
| 90 | 46.0 | 8.5 |
| 110 | 47.2 | 9.2 |
| 130 | 49.0 | 10.5 |
The optimal wave period was 90 mm, where the wavy fins increased the effective heat transfer area by 45% compared to straight fins (from 8500 mm² to 12350 mm² per fin). The enhanced turbulence and secondary flow reduced the average air temperature at the cold plate outlet by more than 3 °C. Overall, the heat transfer efficiency of the energy storage system pack improved from 61% (conventional bottom liquid cooling) to 75% (optimized gas-liquid coupled design). The efficiency was calculated as:
$$ \eta = \frac{\text{heat absorbed by coolant}}{\text{total cell heat generation}} $$
The total heat generated by 48 cells during the 3C cycle was 3,755,904 J. The heat absorbed by the coolant in the conventional pack was 1,464,802 J (η=61%), while the gas-liquid coupled pack absorbed 938,976 J of the total? Wait, that would give lower efficiency. Actually the heat absorbed should be higher. Let’s recalculate: From the paper, the heat absorbed by the coolant in the conventional pack was 1,464,802 J? No, the paper says: “Q2 for bottom liquid cooling is 1,464,802 J, Q3 for gas-liquid coupled is 938,976 J.” That seems lower for gas-liquid coupled. But that cannot be right because gas-liquid coupled has better cooling. Actually, Q2 and Q3 are the heat stored in the battery (i.e., temperature rise). The heat removed by the coolant is total heat minus heat stored. The efficiency formula in the paper is eta = (heat removed) / (total heat). They define Q as heat stored (C M ΔT). Then they compute eta = (Q_total – Q_stored)/Q_total? Actually the formula they gave: η = Δt / Q原 — this is ambiguous. But from context, they define heat removed as total minus stored. For conventional pack: stored Q2 = 1,464,802 J; removed = 3,755,904 – 1,464,802 = 2,291,102 J; efficiency = 2,291,102/3,755,904 = 61%. For gas-liquid coupled: stored Q3 = 938,976 J; removed = 3,755,904 – 938,976 = 2,816,928 J; efficiency = 75%. This matches. So the improved design reduces the heat stored in the battery, meaning more heat is removed by the coolant. Good.
The lower cell temperatures have a direct positive impact on cycle life. For lithium iron phosphate (LFP) cells, experimental data show that every 5 °C increase above 30 °C reduces cycle life by about 1000 cycles. Our optimized gas-liquid coupled design reduces the average cell temperature by 8 °C compared to conventional bottom liquid cooling, leading to an estimated life extension of more than 1400 cycles. Figure below shows the cycle life vs. temperature for LFP cells.
In conclusion, we have presented a gas-liquid coupled dual-phase thermal management design for lithium battery energy storage system. By combining forced air cooling on top of cells with bottom liquid cooling through optimized wavy fins and alternating flow channels, we achieved a maximum cell temperature of 46 °C and a pack temperature difference of 8.5 °C under 3C charge/discharge, compared to 54 °C and 15 °C for conventional bottom liquid cooling. The heat transfer efficiency increased from 61% to 75%. The design also eliminates condensation risks associated with top liquid cooling. These results provide a theoretical and engineering foundation for thermal management of high-power battery systems and can be extended to other high-power electronic and electrical products.
The key contributions of this work include:
- Introduction of alternating gas channels to overcome the short thermal “stroke” of air.
- Optimization of top fin spacing (3 mm) and bottom wavy fin parameters (spacing 6 mm, wave period 90 mm) to maximize heat transfer.
- Demonstration of a 30% increase in bottom cold plate heat transfer capacity through gas-liquid coupling.
- Quantification of cycle life improvement (over 1400 cycles) due to reduced temperature.
Future work will validate performance under extreme conditions such as low-temperature startup, high-temperature operation, and pulsed high-rate discharge, further expanding the applicability of this design in diverse energy storage system scenarios.