What’s the Story: The Cooling Method Is a Strategy for Meeting Safety Standards
For an ESS (energy storage system), choosing a cooling method is, before it is a performance question, a design decision about which layer will pass the thermal-runaway propagation test. No matter how well a battery cell is made, the moment thousands of them are packed into a container, half the problem becomes thermal design. This article compares the three paths — air, liquid, and immersion cooling — from the standpoint of design, standards, and patents.
Starting with the industry context: as ESS demand grows rapidly on the back of AI data centers, the cooling method has become the front line of ESS system design. An ESS is a structure densely packed with thousands to tens of thousands of cells, and in a fire the worst-case scenario is thermal-runaway propagation cascading to neighboring cells. Korean industry is pushing into the market with liquid-cooled ESS emphasizing fire safety, and even oil refiners have jumped into developing immersion cooling fluids, preparing for the next stage.
The tension is that safety requirements are intensifying as fast as the market is growing. ESS systems follow standards such as IEC 62619, UL 9540A, and NFPA 855, and UL 9540A’s thermal-runaway propagation requirements are often assessed as stricter than EV-related tests. So the question this article answers is: which cooling architecture satisfies safety standards, lifespan, and cost all at once, and what patent risks come with that choice?
Basic Principle: Where, and Through What, the Heat Leaves
The difference among the three methods lies in which medium the heat from a cell passes through on its way out. By analogy, air cooling is fanning, liquid cooling is sitting on a cold stone slab, and immersion cooling is jumping into cold water. Fanning is cheap and simple but struggles to cool people far away; the slab cools only the surface in contact; and immersion cools the whole body evenly, but filling and maintaining the water costs money.
Source: concept diagram reconstructed by the editorial team from the original article
Air cooling circulates air with fans. It is simple and cheap in structure, but air has low heat capacity, so temperature variation between cells grows large in a high-density pack. Temperature variation is directly degradation variation, and the hottest cell determines the system’s lifespan.
Liquid cooling places cells and modules on a cold plate and circulates a water-glycol mixture. The cold plate is a component that keeps temperature uniform by spreading heat flux (the amount of heat flowing per unit area) across a wide, flat surface, and its design varies with cell chemistry, module voltage structure, maximum heat-generation rate, and temperature-uniformity requirements. It is more uniform and powerful than air cooling, but it comes with piping, pumps, and a heat-exchange loop, and managing insulation between the coolant and electrical components becomes a design burden.
Immersion cooling submerges entire cells directly in an insulating, dielectric (non-conducting) coolant. The coolant makes direct contact with the cell surface to absorb heat quickly, and by blocking air intake itself, it structurally suppresses fire spread. It is an approach that targets both cooling performance and thermal-runaway-propagation blocking at once, but fluid weight and cost are the penalty. Compromise designs such as spray-type immersion cooling, which sprays coolant onto cells to cut fluid usage by 85%, have reportedly emerged at the research stage.
Comparison: Three Paths from a Design Standpoint
| Item | Air Cooling | Liquid Cooling (Cold Plate) | Immersion Cooling |
|---|---|---|---|
| Cooling capacity / temperature uniformity | Low — large inter-cell variation | High | Very high (direct contact) |
| Thermal-runaway propagation suppression | Weak (air supplies oxidizer) | Indirect contribution | Structural blocking (excludes air, doubles as suppression) |
| System complexity | Fans, ducts | Piping, pumps, heat-exchange loop | Adds tank sealing, fluid management |
| Weight / cost | Lowest | Medium | High (coolant volume) |
| Primary use | Low-density, small-scale | Current large-scale ESS mainstream | Next-generation, high-density, high-safety requirements |
Source: compiled by the editorial team from the original comparison table (qualitative assessment)
What this table says: The cost of passing a thermal-runaway propagation test is paid through passive measures such as firewalls and spacing for air cooling, through thermal management for liquid cooling, and through the architecture itself for immersion cooling. It is a choice of which layer to solve the same test at.
So what exactly is a “layer”? UL 9540A’s formal name is the “Test Method for Evaluating Thermal Runaway Fire Propagation in Battery Energy Storage Systems,” and according to UL Solutions it evaluates at four levels: cell, module, unit, and installation. It is also described as the only consensus standard explicitly cited in NFPA 855 for large-scale fire testing.
| UL 9540A test level | Main evaluation items | How cooling design comes into play (editorial interpretation) |
|---|---|---|
| Cell | Cell thermal-runaway characteristics, gas composition and flammability | Independent of cooling; determined by cell chemistry |
| Module | Tendency of cell-to-cell propagation, heat and gas release rates | Immersion aims to block propagation at this level |
| Unit | Likelihood of fire spread between modules, heat and gas release rates | Liquid cooling responds with thermal management and barrier design |
| Installation | Effectiveness of the fire protection system, heat and gas release rates | Air cooling relies heavily on spacing, firewalls, and suppression systems |
Source: compiled by the editorial team from the UL Solutions UL 9540A test method page (third column is editorial interpretation)
What this table says: The test climbs through four levels from cell to installation, and the earlier propagation is stopped, the lighter the burden at the levels above. Immersion cooling’s cost penalty can be read as the price of cutting propagation at the module level to reduce measures at the unit and installation levels.
Patent Perspective: Immersion ESS Boxes Are Already Being Patented
Because the original article did not address patents, we checked published patents for a case claiming an immersion-type ESS battery box structure. The case in point is the “immersed liquid cooling energy storage battery box” filed on the same day, December 31, 2021, by three companies in China’s NR Electric group. The utility model CN216750062U was published on June 14, 2022, and is listed on Google Patents as Active (anticipated expiry December 31, 2031), while the invention patent application of the same title, CN114335806A, was published on April 12, 2022, and is listed as Pending.
Claim summary (restated): Claim 1 of both documents contains nearly the same configuration.
- A box body with a detachable cover plate and coolant inlet and outlet on the front wall
- Two or more battery modules arranged side by side inside the box
- Flow channels formed between adjacent modules and between the modules and the side walls
- An S-shaped circulation path formed by the channels connected in sequence
- A flow in which coolant entering the inlet passes along the S-shaped path, contacts the top and bottom surfaces of the modules, and exits through the outlet
Source: reconstructed by the editorial team from the claims of CN216750062U and CN114335806A
Insight: In immersion cooling, the scope of rights tends to hinge on “channel layout” rather than “coolant.” The concept of submerging cells in an insulating fluid is itself an old idea and hard to claim broadly. So actual rights tend to narrow, as in this case, to structural limitations such as channel shape, inlet and outlet positions, and module arrangement. This suggests that FTO (freedom-to-operate) review for immersion ESS may need to center on the internal flow-design drawings of the box rather than on coolant composition.
Avoidance points: Under the all-elements rule, literal infringement requires both the S-shaped circulation path and simultaneous contact with the top and bottom of the modules. Parallel branching channels, one-directional bottom-to-top flow, and side-inlet structures are ways to keep distance from the literal scope.
Design-around scenario (rebuttal): That said, merely changing the channel to a “zigzag” or “repeated U” shape could be seen as achieving substantially the same function (sequential cooling of all module surfaces) in the same way, making it a potential target of an equivalents infringement analysis. Conversely, from the patentee’s side, utility models in China are registered without substantive examination, so whether the right survives an invalidation challenge depends on the comparison against prior art. Companies planning to sell immersion ESS in China should track changes in the legal status of both documents together.
There is also a material-property trap that is easy to miss on the design side. Dielectric fluids are cited as having 20–40% lower specific heat and 30–50% lower thermal conductivity than a water-glycol mixture. This means it is not simply “submerged in liquid, therefore automatically cooler”; the gain from direct contact must be weighed against the loss from fluid properties. Pump sizing is the same: because the fluid’s viscosity spikes at low temperature, failing to verify against cold-start conditions leads to an undersized design.
Immersion is not always the right answer. In installations with low heat density, air-conditioned indoor sites, or cost-driven bidding markets, proven liquid cooling — even air cooling — remains reasonable. Immersion cooling’s long-term reliability as a consumable fluid (material compatibility, corrosion, degradation) is still building a track record, and conservative buyers are likely not to move until that verification data is sufficient.
Limitations and Outlook
The optimistic scenario is that as safety regulation strengthens, immersion cooling’s “safety solved through architecture” earns a premium, and the coolant market grows together across data centers, ESS, and EVs. There is an opposing scenario as well. If liquid-cooled pack designs and suppression systems mature to a level that sufficiently meets standard requirements, immersion cooling could remain confined to specialty high-density applications because of its weight and cost penalty. There are two forks in the road: how wide the gap between methods actually turns out to be in real data from UL 9540A-class propagation tests, and when long-term reliability data for coolants becomes sufficient.
A strategic alternative is not to lock in a single cooling method but to validate liquid and immersion cooling in parallel, using module-level propagation-blocking performance as the criterion. Quantifying how much module-test results reduce the cost of unit- and installation-level measures makes it possible to narrow down the specific installation conditions under which immersion cooling’s cost penalty is justified.
Practical action: Professionals planning immersion or liquid-cooled ESS should prepare a table comparing their box’s internal flow-channel drawings against the channel configuration in claim 1 of CN216750062U and CN114335806A (S-shaped circulation, contact with module tops and bottoms), and confirm whether CN114335806A has been granted before launching in China.
Items requiring verification
- The primary measurement source and target fluids for the specific heat (20–40% lower) and thermal conductivity (30–50% lower) comparison of dielectric fluids
- The original paper and test conditions behind the 85% coolant reduction figure for spray-type immersion
- The specific comparative basis for the assessment that UL 9540A propagation requirements are stricter than EV-related tests
- The status of ESS field demonstrations of Korean refiners’ immersion cooling fluids
- Whether CN114335806A is ultimately granted, and the scope of its granted claims
Sources
- UL Solutions, UL 9540A Test Method for Battery Energy Storage Systems (BESS)
- IEC, IEC 62619:2022 Safety requirements for secondary lithium cells and batteries, for use in industrial applications (Edition 2.0, 2022-05-24)
- Google Patents, CN216750062U – Immersed liquid cooling energy storage battery box
- Google Patents, CN114335806A – Immersed liquid cooling energy storage battery box