Norwegian Field Audit of Seven Batteries in Occupied Buildings Finds None Met NFPA 855 Spacing

Seven operating battery installations in Oslo, Bergen and Trondheim, ranging from 41 kWh to 450 kWh and sited in kindergartens and office buildings, were inspected by SINTEF under the Research Council of Norway’s SafeBESS project. None of the seven met the 0.9-meter separation between units that NFPA 855 requires.

One site had gas detection. Two had explosion control. Six relied on smoke detectors alone, which do not register the off-gassing that precedes thermal runaway.

Four of the seven sites ran repurposed Nissan Leaf NMC packs, with no documented cascading-failure resistance on file. That is a chemistry the stationary storage market has largely moved away from in favour of LFP, installed in buildings occupied by children and office workers.

The documentation gap. Three of the seven sites had no cell-, module- or unit-level test data on file despite claiming UL 9540A certification.

UL Solutions describes UL 9540A as the Standard Test Method for Evaluating Thermal Runaway Fire Propagation in Battery Energy Storage Systems, and states that it is a test method rather than a certification standard. What it produces is a set of test reports, run in sequence at cell, module and unit level. A site claiming UL 9540A certification without those reports has never had its fire-propagation behaviour characterised, while holding a document whose name implies that it has.

What the reports are for. NFPA 855 uses UL 9540A results as inputs to the design of an installation. Separation distances between units, and the question of whether a fire in one system spreads to the one next to it, trace back to what the large-scale test showed. An authority having jurisdiction reviewing an indoor installation is reading those numbers. Absent the reports, the review has no object. (Readers should confirm the specific 2026-cycle NFPA 855 and UL 9540A Edition 6 provisions against the published standards; the summaries circulating in trade coverage vary.)

Where the finding travels. NFPA 855 has no legal force in Norway. It was the yardstick the researchers chose, not the rule the installers were breaking, and a US authority having jurisdiction operating under an adopted fire code has enforcement tools that a Norwegian performance-based regime does not. The audit measures how a set of installations compares against an international benchmark, not a record of violations.

The result still travels further than its sample size suggests, because the failure mode it describes is procedural rather than national. A claimed certification that no one asked to see is available in any jurisdiction where the reviewing authority does not request the underlying reports. The seven Norwegian sites are the first published measurement of that gap in the sub-500 kWh class inside occupied buildings.

ASHRAE’s parallel move. In the same week, ASHRAE moved Addenda M and N, the greenhouse-gas emissions reduction requirements in ANSI/ASHRAE/IES Standard 90.1, out of the main body of the standard and into optional status for the 2028 and 2031 editions. Committee co-vice-chair Tom Culp and committee chair Richard Lord of Carrier both spoke to the change. ASHRAE said the standard will “maintain its primary focus on energy efficiency and energy cost justified requirements.”

Standard 90.1 governs building energy use and has nothing to do with battery safety. The mechanism is what the two developments share. In each case an obligation written into a standard’s body is relocated to a party downstream: the jurisdiction that chooses to adopt the optional addenda, or the engineer who authors the documentation an authority may or may not read. Jurisdictions typically adopt a new 90.1 edition on a lag of one to eight years, so relocation of this kind is slow and uneven by default.

The installed base. The audited systems were permitted, energized and are presumably still operating. They predate the documentation practices now being written into the current code cycle, and code editions generally apply to new installations rather than existing ones. As jurisdictions adopt newer editions, the distance between what a new installation must prove and what an existing one ever proved widens rather than closes.

That distance is the practical finding. Four of seven sites in this sample were running second-life EV packs of a chemistry the stationary market has largely left behind, monitored by smoke detectors that register the fire after the gas. Nothing in the audit suggests those installations were unusual for their vintage. They were assembled under a performance-based regime, by parties who supplied a certification claim that satisfied whoever asked.

What is not known. No comparable field audit of the sub-500 kWh in-building class appears to have been published in the United States, though absence of a published study is weaker evidence than a study finding nothing. The US installed base of this size class is larger than Norway’s and was built under a similar mix of prescriptive tables and professional judgement.

The paper is likely to be read by fire marshals and property insurers over the coming quarters for what it says about seven buildings in three cities. The more consequential question is what an equivalent inspection would find elsewhere, and how few jurisdictions currently have a basis for answering it.

Sources