UL Solutions Opens a UL 9540 Route for DC Battery Systems Paired With Separately Evaluated Subsystems

UL Solutions has launched a certification pathway that lets a direct current energy storage system demonstrate UL 9540 compliance when it is paired with an approved, separately evaluated subsystem such as a power conversion system. The pathway covers configurations that include components carrying no UL certification of their own. Energy-Storage.News carried the launch on October 22; wire coverage dates the announcement to October 20.

What the evaluation covers. UL assesses electrical, communication and interoperability performance between the DC ESS and each evaluated subsystem. The subject of the work is the interface, which is where a battery cabinet from one vendor and a power conversion system from another have historically failed to combine into a single listed product.

Where the record lives. Compatible sub-assembly models are published in UL’s Product iQ database under the DC ESS certification category, so an authority having jurisdiction can verify a given pairing independently. A plan reviewer who can confirm a configuration from a searchable database is working from a different evidentiary basis than one working from a vendor letter.

UL’s stated rationale. The company frames the pathway as an answer to modular and matched-assembly designs, which fit poorly into a traditional listing built around a single manufacturing program. That framing describes the commercial problem the service addresses: systems increasingly arrive as combinations of separately sourced parts, while the listing regime was built for products that arrive whole.

The analytical read. One interpretation, not a sourced industry finding, is that system-level UL 9540 listing has functioned as a de facto screen in US commercial projects, keeping out generic imported cabinets assembled from cells and inverters that each hold their own paperwork and no shared one. On that reading, a route that assembles a UL 9540 claim from separately evaluated parts lowers the screen for anyone able to buy qualified components and document the interfaces between them. The offsetting consideration is that Product iQ publication gives a code official a verifiable record of which pairings were actually evaluated, which raises the quality of the documentation available at plan review rather than lowering it.

Sungrow’s large-scale fire test. On October 22, Sungrow disclosed that a 20-foot PowerTitan 3.0 configuration completed a UL Solutions large-scale fire test in China, conducted to UL 9540A:2026 and CSA/ANSI C800-2025, witnessed by the Fire Risk Alliance and observed by Jensen Hughes. The setup placed 28.56 MWh of liquid-cooled units at 100% state of charge alongside a 7.4 MW medium-voltage substation, with all suppression and thermal management disabled and rooftop pressure-relief panels fully opened. Peak temperature inside the initiating container reached 1,259.8°C over 7.5 hours. Containers B through E and the substation remained structurally intact, with no propagation recorded.

The 2026 NFPA 855 requirement. Aaroh Kharaya, director of engineering at Primergy, set out the requirement in an Energy-Storage.News commentary circulating this month: the 2026 edition requires large-scale fire testing alongside UL 9540A data, with the test proving that a fire in one ESS unit will stay contained and spread no further. Kharaya’s conclusion is that building-based BESS has become untenable to build and operate, and that modular outdoor containers with verified non-propagation behaviour now carry the cleaner permitting story. The commentary has been syndicated to secondary outlets, putting the argument in front of utility-scale developers and their counsel.

Two documents, two cost curves. The listing is becoming composable. A manufacturer holding its own DC ESS evaluation and buying an evaluated power conversion system can reach a UL 9540 claim through an interface assessment, with the pairing published in a database an official can search. The fire test is a capital program. Burning 28.56 MWh under third-party witness for 7.5 hours, with thermal management deliberately switched off, is a line item only a well-capitalised integrator carries, and the result is specific to the geometry that was burned.

The asymmetry runs in one direction. Certification has been the gate that decides which products reach a US commercial site. As the paper becomes more composable and the empirical evidence stays expensive, the scarcer document is the test report, and test reports do not transfer between configurations.

Outdoor geometry. The Sungrow test, like the outdoor container programmes it belongs to, measures spread across a yard. Relief panels vented upward. Container spacing was horizontal, with neighbours on a pad. The quantity being proved is container-to-container propagation in a configuration where the pressure wave has somewhere to go.

An indoor fire area is a different geometry: a ceiling, a rated assembly between the battery and the occupancy above it, and a maximum allowable quantity written per fire area. The 2026 edition’s large-scale testing requirement applies to that installation as well. Published results from tests run inside an enclosed room, against an occupied space, are not evident in the current public record. This is interpretation rather than a reading of the standard, but it follows that a vendor publishing a credible interior test result would hold a permitting position competitors cannot replicate by combining certified parts, because the subject of the test is the room.

Kharaya’s argument and UL’s announcement approach the same underlying development from different sides. Compliance for a battery system is decomposing into a stack of separately verifiable claims, and those claims differ widely in what they cost to produce. An interface evaluation is an engineering review. A propagation test is a controlled burn with witnesses.

A fire marshal reviewing a commercial project in a jurisdiction that has adopted the 2026 editions can look up a subsystem pairing in Product iQ in about a minute. The large-scale test that the same code requires took Sungrow seven and a half hours to run, and it was run outdoors.

Sources