The UPS Battery Room: What NFPA 855 (2026) Changed at Procurement

A lithium-ion UPS battery room design that cleared permitting in 2023 will not necessarily clear it today. The equipment did not change. The building did not change. The standard governing the installation did, and it did so in ways that attach directly to what was bought rather than to how it was installed.

The 2026 edition of NFPA 855, the Standard for the Installation of Stationary Energy Storage Systems, makes Hazard Mitigation Analysis the default rather than the exception, tightens explosion control, and removes a venting strategy that a large number of existing battery room designs were built around. For data center operators, colocation providers and the contractors building for them, this is not a fire protection matter that arrives late in the project. It is a procurement matter, because the evidence the Authority Having Jurisdiction now expects is documentation that a supplier either possesses for the configuration being bought or does not.

What actually changed

Earlier editions worked on thresholds. If an installation stayed below a defined maximum stored energy limit, it could often proceed without formal hazard analysis. The 2026 edition removes that threshold-based approach. An HMA is now a standard part of the safety evaluation for most installations unless a technology-specific chapter explicitly exempts them. Under Section 4.4 it may cover the initiation and propagation of thermal runaway, the generation and dispersion of hazardous gases, the potential for deflagration or explosion, the performance of detection, suppression and ventilation systems, and defects in the separation, containment or compartmentation strategy.

The second change is sharper. Installations must now incorporate explosion control and prevention designed to NFPA 69, or a performance-based alternative supported by installation-level fire and explosion testing and engineering evaluation. NFPA 68 deflagration venting is no longer permitted as a primary explosion control strategy. That single sentence invalidates the governing assumption behind a great many battery room layouts drawn in the last code cycle. The direction of travel is from passive relief of an explosion toward active prevention of the gas accumulation that causes one.

Supporting that shift, the standard adds requirements for combustible concentration reduction systems. These must remain operational during failure scenarios, and their design has to address gas composition and volume data derived from testing such as UL 9540A, partial volume deflagration scenarios, and hazardous gas migration between interconnected units. Every one of those inputs is a property of the battery product and its arrangement, not of the room.

Where the procurement error occurs

The first and most expensive mistake is buying the cabinet and assuming the test data follows. It does not follow automatically. UL 9540A data is generated against a specific cell, module, unit and arrangement. If the room layout, rack density or interconnection pattern deviates from what the manufacturer actually tested, the AHJ is entitled to ask for evidence covering the arrangement being built. Where that evidence does not exist, the options are to redesign the room to match the tested configuration or to commission installation-level testing. Both are measured in months. Neither was in the schedule.

The second mistake is treating battery chemistry as a purely technical or commercial choice. Lithium-ion wins on energy density, footprint and increasingly on price, and in a facility where white space carries revenue that argument is strong. But the chemistry selection is simultaneously a code selection. It determines whether an HMA is straightforward or contested, whether explosion prevention is a modest engineering scope or a substantial one, what gas detection and ventilation the room requires, and how much of the surrounding building has to be reconsidered. Lead-acid and nickel-zinc installations generally carry a lighter fire protection and explosion control burden. That difference belongs in the evaluation at RFQ stage, priced, rather than surfacing as a variation order after award.

The third mistake is sequencing. The AHJ has historically been treated as a final check before commissioning. Under the current framework, coordination is expected during design and before responder training, and the emergency response plan is submitted to the fire department rather than filed internally. A procurement team that selects a supplier, places the order and only then discovers what the local fire marshal will accept has inverted the process. The permitting position should be understood before the purchase order, because it is the purchase order that fixes most of the variables the permit turns on.

The fourth is scope ambiguity between packages. Gas detection, ventilation, suppression, containment and the battery system itself are frequently procured under different packages, sometimes under different contracts. The HMA assesses them as one system. Where no single party owns the interfaces, the analysis has no coherent subject and the permit stalls while three suppliers explain that the gap belongs to somebody else.

How we qualify a battery system

D1R7K0N treats the documentation package as part of the goods. A quotation for a lithium-ion energy storage or UPS battery system is incomplete without the UL 9540A test report, an explicit statement of the tested configuration, the gas composition and volume data the explosion control design will be built on, and confirmation of what the manufacturer will and will not certify for the arrangement the project intends to install. If a supplier cannot produce that at quotation, the gap does not close later. It simply transfers to the buyer at the worst point in the schedule.

Practically, we qualify against three questions. Does the test data cover this arrangement, at this rack density, with this interconnection pattern, or does it cover something adjacent that was easier to test? Is the explosion control strategy compliant on its own terms, given that deflagration venting can no longer carry the primary role, and is the combustible concentration reduction system specified to remain operational during the failure it exists to manage? And is one party contractually responsible for the performance of the integrated system, so that the hazard analysis has a single accountable author rather than three partial ones?

We also establish the AHJ position early, because it varies. The standard sets the framework, but the jurisdiction interprets it, and the interpretation is a project input with real commercial weight. A twenty minute conversation before award is worth more than a strong argument after rejection.

The permit is bought, not obtained

Energy storage inside the data center has moved from a backup detail to a core part of the power architecture, and the code has moved with it. The practical consequence for buyers is narrow and worth stating plainly: in a battery room, the permitting outcome is largely determined by procurement decisions, and by the time the design is submitted the important choices have already been made.

For anyone with a battery package in specification or in tender now, the useful step is small. Ask each bidder for the UL 9540A report and the tested configuration, compare it against the arrangement actually being built, and price the difference. Most of the time it will match and the exercise costs an hour. When it does not match, finding out at bid stage is the difference between changing a supplier and changing a schedule.