Two Suppression-Disabled Fire Test Disclosures, Six Days Apart, In Formats That Do Not Compare

APsystems announced on September 3 that its APstorage 261L liquid-cooled cabinet completed UL 9540A fire propagation testing witnessed by TÜV Rheinland, with the unit held at 100 percent state of charge and its own fire suppression system intentionally switched off.

The published results: heat flux of 0 kW/m² near the affected cabinet, against what the company describes as the UL 9540A threshold of 1.3 kW/m²; maximum module surface temperature of 83.9 degrees Celsius on both the initiating and the neighbouring modules; and a maximum adjacent wall temperature of 30.6 degrees, which the release puts at 3.1 degrees above ambient. Thermal runaway remained confined to a single enclosure, with no propagation and no loss of structural integrity.

Six days earlier, on August 28, a suppression-disabled test of Canadian Solar’s 940 kWh commercial and industrial cabinet was reported. Two suppression-off cabinet burns were therefore disclosed inside a fortnight by two manufacturers competing for the same commercial and industrial specification.

The disclosures. Both companies made the same choice: remove the mitigation, run the worst case, publish the outcome. Neither was obliged to do any of it. Certification requires a pass. It does not require a press release, and it does not require handing the market a figure a competitor can be measured against.

What each one reports. The two results are not expressed in the same units of evidence. APsystems published three specific figures, heat flux, module surface temperature and adjacent wall temperature, and named neither the standard edition nor the test level that produced them. Its announcement states the system was evaluated at the cell, module, unit and installation levels, without attributing the reported results to any one of those. The Canadian Solar disclosure, as carried in trade coverage, is identified by capacity and by the suppression-off condition, with no comparable temperature figures in circulation.

Per Mayfield Renewables, the sixth edition of UL 9540A replaced the former installation-level test with a fourth-level large-scale fire test that intentionally ignites off-gases, where earlier editions assessed thermal runaway without a deliberate ignition source. A wall temperature recorded under a propagation test and a wall temperature recorded under an off-gas ignition test are not observations of the same event.

Neither disclosure states the separation distance between cabinets, or between cabinet and wall. That is the variable that gives an adjacent-surface temperature its meaning. Neither states whether the tested arrangement represents an indoor installation.

Both results, as described, are good results. The difficulty is placing them in the same table.

Surprise’s vote. The Surprise City Council voted 7-0 to reject its own staff’s recommendation of a 1,500-foot separation between battery storage facilities and existing homes, adopting instead a variable band of 100 to 500 feet, with city officials determining the appropriate distance based on the specifics of each proposal.

Assistant City Manager Lloyd Abrams told ABC15 that the rejected figure came from the standard evacuation radius for a hazardous-materials incident. The 1,500 feet was a hazmat evacuation radius, borrowed intact. It was not an engineering finding about a battery.

The vote carries local weight because the 2019 APS McMicken explosion happened in Surprise. Council members who supported the flexible band cited safety improvements in battery technology since that incident.

Discretionary siting. Where a jurisdiction writes a fixed distance into an ordinance, test data changes nothing about whether a project is permittable. Where it writes a band and reserves discretion, the applicant’s submittal becomes the entire argument, and large-scale fire test results are the only category of evidence that speaks directly to separation.

Surprise published no criteria for placing a given project at 100 feet rather than 500. On the available reporting, the record in each case will be built from whatever the applicant brings.

What the applicant will bring, increasingly, is a vendor test summary of the kind issued in the past fortnight. Those summaries do not share a format, a standard edition, a test level, or a set of reported quantities. A reviewer attempting a side-by-side reading has a temperature without a separation distance in one file and a suppression-off condition without published temperatures in the other.

What would force a common format. Two mechanisms are plausible. A large buyer writes the disclosure format into a procurement specification, requiring edition, test level, separation geometry and adjacent-surface temperature rise as a matched set. Or an authority having jurisdiction rejects a submittal as non-comparable and the rejection circulates. Neither has been reported. Until one occurs, publishing a suppression-off result remains a marketing decision made one vendor at a time, and the metrics will be selected accordingly.

The stronger read of the past fortnight is that the competitive claim in commercial storage has moved from holding a certificate to publishing a specific burn artifact. That shift favours manufacturers with test data they are willing to attach a figure to, which is a narrower group than the set of manufacturers holding a listing.

Surprise’s band gets tested the first time a developer asks for 100 feet and a neighbourhood asks why. Whatever goes into the record that night will set the operative precedent, not the ordinance text, and on current evidence the segment’s two most recent fire test disclosures could not be entered into the same table.

Sources