Liquid Cooling vs. Air Cooling vs. PCM: Battery Thermal Management Selection Guide for ESS Procurement

TL;DR #

In controlled thermal evaluation across five coolant channel architectures, liquid cooling outperformed air and phase-change alternatives on every metric that matters at scale — temperature uniformity, response speed, and long-term operational cost. For procurement teams specifying battery thermal management for large-format ESS, air cooling is not a viable primary strategy above low-power applications, and PCM-only approaches have not achieved industrial-scale deployment. Qualify your module supplier’s liquid cooling channel geometry before committing to a pack design — channel spacing, flow path shape, and coolant selection together determine whether your temperature delta stays under 5K in sustained discharge.

---

Overview #

Battery thermal management has quietly become one of the most consequential specification decisions in large-format lithium-ion procurement — and most buyers treat it as an afterthought. The cooling architecture embedded in a battery module determines cycle life, safety margin under fast discharge, and total cost of ownership far more than the cell chemistry alone. Research conducted at a Chinese marine propulsion engineering institute — drawing on multi-model numerical simulation and experimental validation across cylindrical, prismatic, and pouch cell configurations — provides a systematic comparative dataset that procurement engineers can use to set minimum performance thresholds before they even request samples.

The study evaluated three primary thermal management categories (air, liquid, and phase-change) plus two composite architectures, using temperature uniformity (expressed as maximum temperature differential across the module), peak surface temperature, pressure drop across the cooling circuit, and coolant flow rate as the core performance metrics. This is more comprehensive than most supplier datasheets, which typically report only maximum operating temperature rather than the temperature delta that actually drives cell degradation.

Understanding how cell format — cylindrical vs. pouch vs. prismatic — interacts with cooling channel design is essential context here. If you are evaluating Cell Formats & Form Factors in parallel with thermal management suppliers, the cooling architecture decisions cannot be made independently. A cylindrical cell pack running at high C-rate has fundamentally different thermal geometry than a pouch-based module.

---

Liquid Cooling vs. Air Cooling vs. PCM: What the Data Actually Shows #

This is where most procurement teams make their first costly mistake. They see air cooling as a cost reduction lever. It is — at commissioning. It is not over a three-year operating horizon.

The comparative data across five thermal management approaches — natural convection, forced-fan air cooling, air conditioning, tube-type liquid cooling, and cold-plate liquid cooling — reveals performance gaps that are difficult to argue away:

The PCM service life figure of 3–5 years is the one that should immediately flag concerns for any system with a 10+ year design life. Cold-plate liquid cooling carries higher upfront cost but matches or exceeds forced-air service life while delivering temperature uniformity no air system can approach.

The air cooling numbers deserve scrutiny beyond the table. Forced-fan systems are classified as open systems — they introduce ambient air, which means humidity, dust, and particulate contamination are ongoing operational risks for sealed battery modules. In marine, industrial, and outdoor ESS applications, this is not a theoretical concern. Contamination-driven contact resistance and moisture-induced short circuit risk are real failure modes that don’t show up in a 30-day factory acceptance test.

Liquid cooling operates as a closed circuit. The coolant — whether water-glycol, mineral oil, or fluorinated fluid — does not interact with the cell surface or interconnect hardware unless there is a leak. This is a fundamentally more controllable thermal environment.

Compliance with IEC 62619:2022 Safety requirements for secondary lithium cells and batteries does not specify cooling architecture, but the thermal abuse test conditions it mandates implicitly require that your module’s cooling system can contain temperature excursions within safe bounds. Air-cooled modules at high discharge rates routinely have narrower safety margins against those test conditions.

---

Liquid Cooling Channel Geometry: The Variables That Determine System Performance #

Here is where the engineering gets specific, and where supplier technical depth becomes directly visible during qualification.

For cylindrical cells, wrapping an aluminum block around the cell with an embedded flow channel has been shown to deliver superior temperature uniformity versus direct-contact cooling — because aluminum’s high thermal conductivity distributes heat across the entire interface before it reaches the coolant. The contact area between the aluminum block and the cell surface directly governs performance; variable-contact designs outperform fixed-contact designs on uniformity metrics. This is a design detail most suppliers will not volunteer.

For flow rate optimization in a dual-inlet/dual-outlet liquid cooling system serving a 2-parallel, 12-series module, experimental data identified an optimal coolant flow rate of 450 L/h. Performance increases with flow rate up to this threshold, then degrades — a non-linear relationship that contradicts the intuitive assumption that more coolant always means better cooling. Exceeding 450 L/h introduced diminishing returns and increased parasitic energy consumption. Critically, setting coolant temperature significantly below ambient temperature did not reliably improve cooling performance — the relationship between coolant temperature and thermal outcome is not linear, and over-cooling the inlet can create condensation and differential thermal stress across the module.

For pouch cell architectures, simulation across curved, open, and rectangular channel geometries produced a clear ranking:

• Curved channels: best temperature distribution uniformity, lowest pressure drop across all conditions tested
• Rectangular channels with water coolant: higher heat absorption capacity but significantly higher pressure drop
• Open channels: worst cooling performance in all tested configurations

The Stanton number (a dimensionless measure of convective heat transfer efficiency relative to fluid thermal capacity) increases with flow velocity but with diminishing marginal returns — which reinforces the 450 L/h finding from the module-level experiments. You can optimize flow geometry far more cost-effectively than you can compensate with brute-force flow rate increases.

For cold-plate liquid cooling, a dual-layer I-shaped channel design demonstrated that optimal geometry parameters are a length ratio of 0.70, width ratio of 0.85, and channel spacing of 2.0 mm. Under these parameters, maximum pressure drop was only 26.64% of what equivalent serpentine channel designs produced — a substantial pumping energy reduction with no temperature uniformity penalty. Length ratio had the largest influence on thermal performance of the three variables.

Buyers evaluating Cell Selection & Sourcing for high-C-rate applications should be requesting cold-plate geometry specifications — not just coolant inlet/outlet temperature specs — from thermal management module suppliers. The geometry is where the performance lives.

---

Dielectric Coolant Selection and Composite Cooling Systems #

The choice of coolant matters as much as the channel geometry in immersion cooling systems. Five fluid categories are used industrially: fluorinated electronic fluids, hydrocarbons, esters, silicone oils, and water-based fluids. Each has trade-offs across six performance indicators: dielectric constant, thermal conductivity, specific heat capacity, flash point (flammability), freezing point, and material compatibility with battery system components.

No single fluid dominates all six criteria. Water-based fluids have excellent specific heat capacity and thermal conductivity but poor dielectric properties — unsuitable for direct cell contact. Fluorinated fluids score well on dielectric constant and non-flammability but have lower thermal conductivity and higher cost. Ester fluids balance moderate performance across most indicators with reasonable cost. Silicone oils are chemically inert but thermally mediocre.

The selection decision is application-driven: immersion cooling for large-format stationary ESS favors fluorinated or ester fluids; indirect cold-plate systems most commonly use water-glycol, which is cost-effective, well-understood, and broadly compatible with aluminum cold-plate hardware.

On composite cooling systems: the research evaluated two composite approaches. The first combined forced-air with microchip liquid cooling for cylindrical cells. The hybrid achieved a peak module temperature of 304.98 K (31.83°C) with maximum temperature differential held below 5 K — better than either system alone. However, the operational reality undermines the engineering advantage. Forced-fan systems as open circuits accumulate contamination over time. Fan motor efficiency declines. Energy consumption increases. Maintenance intervals compound. The air subsystem that provided initial benefit gradually degrades and imposes costs on the liquid subsystem it was supposed to supplement.

The second composite approach integrated PCM with liquid cooling. Under simulated thermal runaway conditions, the hybrid system achieved a maximum cell temperature of 93.3°C — a reduction of 63.5°C compared to liquid cooling alone. This is the strongest argument for PCM integration: thermal runaway buffering capacity, not steady-state cooling efficiency. PCM acts as a thermal shock absorber; the liquid circuit handles baseline heat removal. For high-capacity, high-discharge-rate applications, standalone PCM cooling fails. The hybrid gives you both.

The honest evaluation: PCM-liquid composite systems have not achieved industrial-scale deployment. Material performance degradation over multi-year operation and cost constraints remain unresolved. The research confirms that optimizing a single, well-engineered liquid cooling system outperforms poorly integrated composite architectures. Composite cooling is the future direction; cold-plate liquid cooling is the current industry standard.

Compliance documentation under UL 9540A Test Method for Evaluating Thermal Runaway Fire Propagation in Battery Energy Storage Systems specifically tests whether your thermal management architecture can contain thermal runaway propagation between adjacent cells. PCM-liquid hybrid results at 93.3°C maximum — versus 156.8°C for standalone liquid — show directly relevant safety margins for UL 9540A qualification. For buyers sourcing into North American markets, this distinction matters at the certification stage.

For stationary ESS sold into European markets, EU Battery Regulation 2023/1542 imposes thermal management performance disclosure requirements that will increasingly favor documented liquid cooling specifications over undocumented air-cooling assumptions.

---

Practical Guidance for Buyers #

If you are specifying a battery module for anything above 10 kWh continuous cycling — ESS, marine propulsion, industrial UPS — air cooling should not appear in your shortlist. The data is clear and the industry has moved on. The qualification question is no longer “liquid cooling vs. air cooling” but rather “what cold-plate geometry, what coolant, what flow rate, and what control strategy.”

Honestly, most buyers over-specify the coolant type and under-specify the channel geometry. A supplier quoting “water-glycol at 25°C inlet” has told you almost nothing. What matters is whether they can demonstrate temperature uniformity (ΔT across module) below 5 K at your rated continuous discharge current, validated by test rather than simulation alone.

At CompactBESS — a Guangzhou-based sourcing service connecting global OEM buyers with verified Chinese manufacturers of battery modules, thermal management systems, and ESS packs — we review thermal validation documentation during supplier qualification. If a supplier’s thermal characterization report shows only maximum cell temperature with no temperature differential data, that is a qualification flag. Buyers sourcing from Chinese manufacturers through our platform can request matched suppliers with documented cold-plate test data before initiating an RFQ.

Mass flow rate in PCM-liquid hybrid systems should be kept below 0.001 kg/s — higher flow rates do not meaningfully improve thermal performance in hybrid configurations and increase system energy consumption and component wear. This is a verifiable specification, and suppliers who cannot provide it have not done the characterization work.

Need help identifying qualified suppliers for liquid-cooled battery modules or thermal management systems? Talk to our sourcing team →

---

Supplier Qualification Questions #

1. What is the measured maximum temperature differential (ΔT) across your battery module at rated continuous discharge current, and at what coolant flow rate was this measured — specifically, can you demonstrate ΔT below 5 K at the test conditions you report?
2. For your cold-plate liquid cooling design, what are the channel length ratio, width ratio, and channel spacing, and can you show that your pressure drop is below 30% of an equivalent serpentine channel design under the same flow conditions?
3. What is your validated optimal coolant volumetric flow rate for your module cooling circuit, and can you provide the flow rate vs. temperature performance curve showing the non-linear response above and below the optimal point (reference: optimal observed at 450 L/h in dual-inlet/dual-outlet 12S2P configurations)?
4. If your design incorporates PCM-liquid composite cooling, what is the documented maximum cell surface temperature under simulated single-cell thermal runaway, and how does this compare against your standalone liquid cooling configuration — can you provide the ΔT reduction figure from runaway testing?
5. For immersion or direct-contact cooling configurations, what dielectric coolant fluid class do you specify, and can you provide compatibility test data showing no measurable corrosion or degradation of cell housing, busbar, and module structural materials after 1,000-hour immersion exposure?

---

Sourcing Checklist #

• ☐ Supplier provides temperature uniformity test report showing ΔT ≤5 K across module at rated continuous discharge current under liquid cooling
• ☐ Cold-plate channel geometry is documented with length ratio, width ratio, and channel spacing — channel spacing confirmed at or near 2.0 mm optimum
• ☐ Pressure drop across cooling circuit is specified and verified as ≤30% of equivalent serpentine channel design (reference: dual-layer I-channel achieves 26.64% of serpentine pressure drop)
• ☐ Coolant flow rate specification is provided with a validated optimal operating point and non-linear performance curve (reference benchmark: 450 L/h for 12S2P module)
• ☐ For PCM-liquid hybrid systems: maximum cell temperature under thermal runaway simulation is documented and below 100°C (reference: 93.3°C achieved in hybrid vs. 156.8°C liquid-only baseline)
• ☐ Coolant fluid specification includes dielectric constant, thermal conductivity, flash point, and material compatibility data for the specific cell housing and aluminum cold-plate materials used
• ☐ Module-level thermal test report references IEC 62619:2022 or equivalent standard test conditions, not simulation-only data
• ☐ For composite air-liquid systems: supplier provides long-term degradation data on air subsystem performance over minimum 2-year operation period

---

Key Specifications Table #

Can’t find a supplier meeting these specs? Submit your requirements and we’ll match you within 48 hours.

---

References #

Data source: Comparative Analysis of Liquid Cooling Architectures and Thermal Management Strategies for Lithium-Ion Battery Modules in Energy Storage Applications, B. Tang et al., Journal of Energy Storage, 2025

---

Frequently Asked Questions #

Why does increasing coolant flow rate beyond a certain point reduce cooling performance?

Cooling performance follows a non-linear curve with respect to flow rate. Experimental data on a 12S2P module in a dual-inlet/dual-outlet circuit showed optimal performance at 450 L/h. Above this threshold, flow velocity increases turbulence and residence-time mismatch in the channel, reducing heat transfer efficiency while increasing parasitic pump energy consumption. The relationship between flow rate and thermal performance is application-specific, but the non-linearity is consistent across multiple module geometries.

Can air cooling ever be sufficient for battery modules above 10 kWh?

Technically yes, in low C-rate applications with generous thermal margins and controlled ambient conditions. Practically, no — at least not as a primary strategy. Air cooling as an open system accumulates contamination over time, and fan motor efficiency degrades. The initial cooling performance advantage over liquid cooling narrows and eventually reverses after sustained operation without maintenance. For anything above low-power, low-cycle-rate applications, the long-term total cost of ownership strongly favors liquid cooling.

What makes curved channel geometry superior to rectangular channels for pouch cells?

Curved channels distribute coolant residence time more evenly across the heat exchange surface, producing more uniform temperature distribution with lower pressure drop compared to rectangular channels. Rectangular channels absorb more total heat per unit time but impose significantly higher pressure drop, which means higher pump energy and mechanical stress on fittings. Curved channels achieve better thermal uniformity at lower operating cost — the primary qualification metric for stationary ESS applications.

Is PCM-liquid composite cooling worth the added complexity for commercial ESS?

Not yet, at scale. PCM-liquid hybrid systems show compelling thermal runaway performance — reducing maximum cell temperature by 63.5°C versus standalone liquid cooling in simulated runaway events — but PCM material performance degrades over multi-year operation cycles, and the technology has not been commercialized at industrial scale. The right application is high-consequence, compact installations where thermal runaway containment justifies the cost premium. For standard commercial ESS, a well-engineered cold-plate liquid system with proper channel geometry optimization outperforms a poorly integrated PCM hybrid.

What dielectric coolant type should I specify for immersion-cooled battery modules?

It depends on the application and your cost constraints. Fluorinated electronic fluids score highest on dielectric constant and non-flammability but are expensive and have lower thermal conductivity than water-based fluids. Ester fluids offer balanced performance across most parameters at moderate cost and are increasingly used in EV and stationary ESS immersion systems. Water-based fluids are thermally superior but unsuitable for direct cell contact due to poor dielectric properties. For indirect cold-plate systems, water-glycol remains the default — cost-effective, well-characterized, and compatible with standard aluminum cold-plate hardware.

Published by compactbess.com Technical Team | Request a sourcing quote

---