Data center fire protection and life-safety field guide

Data center fire and life safety, and the core challenge

Data center fire and life safety is the integrated set of detection, suppression, alarm, and egress systems that keeps a fire from reaching the IT load while making sure the protection itself never takes that load down. A data hall is an odd fire problem. The space is energized around the clock, it is packed with high-value electronics that water ruins as thoroughly as flame does, and it is occupied mostly by equipment rather than people, with crews in and out. So the fire strategy has three jobs at once: see a fire while it is still a warm bearing or a smoldering connector, put it out without destroying the gear you are trying to save, and never dump on a false alarm.

The tension between those jobs is the whole design. Detect too late and a smoldering power supply seeds a fire the suppression has to fight instead of snuff. Suppress with the wrong tool and the cure costs more than the fire. A wet pipe sprinkler line sitting full of water directly over live racks is the thing every operator wants to avoid, because one cracked head or one bumped fitting soaks a row of servers that were never on fire. Nuisance-dump a clean agent and you have spent a six-figure cylinder bank, lost the room to downtime, and trained everyone to distrust the alarm.

None of this is one system. It is detection talking to the fire alarm, the alarm driving a cause-and-effect sequence, that sequence stopping fans and closing dampers and releasing agent, the room built tight enough to hold the agent, the barriers built to keep a fire in one cell, and the egress built to get the few people out. Get one piece right and the others wrong and the chain still breaks. The standards below each govern a link in that chain, and the AHJ stitches them together.

The standards landscape

No single document governs a data center fire scheme. A stack of NFPA standards each owns a piece, and the building code plus the AHJ decide which apply and to what edition. Naming the right one for the right question is half of getting through plan review.

Two standards are written for these occupancies specifically. NFPA 75, the Standard for the Fire Protection of Information Technology Equipment, covers the IT room itself, its construction, and the detection and suppression expectations for the space. NFPA 76, the Standard for the Fire Protection of Telecommunications Facilities, covers carrier and central-office type spaces. Which one applies depends on whether the facility is an IT data center or a telecom plant, and large campuses can touch both.

Treat every number and clause as edition-dependent. NFPA revises on a three to five year cycle, the jurisdiction adopts a specific edition that is often a few cycles behind, and amends it locally. Confirm the adopted edition and the local amendments before you cite a clause on a submittal, and let the AHJ settle any conflict between documents. The insurer adds its own layer through the FM Global property-protection data sheets, which are frequently stricter than code and govern by contract where the carrier requires them.

Why won't an ordinary smoke detector work in a data hall?

An ordinary spot smoke detector struggles in a data hall because the cooling airflow dilutes and sweeps the smoke before enough of it reaches the detector to alarm. A computer room moves a lot of air. The same fans that keep the racks cool pull the early products of combustion away from a smoldering component and mix them into the volume, so a spot detector on the ceiling sees a thin, diluted sample long after a person standing at the rack would smell it.

The answer is aspirating smoke detection, ASD, widely known by the VESDA brand of very-early-warning detectors. An ASD unit runs a network of small-bore sampling pipe through the space and actively draws air past a sensitive laser detector, continuously, instead of waiting for smoke to drift to a passive head. Because it pulls its own sample, high airflow hurts it far less, and its sensitivity is set fine enough to catch combustion at the off-gassing stage, before there is visible smoke. That early warning is the time you need to investigate and pull a circuit before anything has to discharge.

Where you sample matters as much as the technology. Good designs sample at the ceiling, in the return air path where the room collects its air, and often at the rack or in the supply, because in a contained hot-aisle or cold-aisle layout the smoke travels with the airflow, not straight up. NFPA 75 expects detection suited to the airflow of the space. The rookie error is a grid of spot detectors on a 30 ft ceiling over a room turning its air over many times an hour, and then surprise when the alarm lands minutes after the smell.

What fire suppression do data centers use?

Data centers use one of three approaches over the IT space, or a layered combination: a clean agent gaseous system, a pre-action sprinkler system, or in some designs water mist, with the choice driven by what the owner is willing to risk and what the code and the insurer require. The one thing nearly everyone avoids is a conventional wet pipe sprinkler sitting charged with water directly above energized racks.

Clean agents put a fire out by gas and leave no residue, so the gear that was not burning survives untouched. Pre-action keeps the sprinkler pipe dry until a detection event opens a valve, so a single damaged head over a rack does not leak water. Water mist uses very fine droplets to cool and smother with a fraction of the water volume. None of these is free. Clean agent rooms have to be built tight and the cylinder bank is expensive to refill. Pre-action adds detection and valve complexity. The decision is a risk and cost tradeoff run room by room, not a default.

A common real-world layout pairs the two. A clean agent system handles the early event in the room, and a pre-action sprinkler sits behind it as the structural backstop the building code still requires, since a gas system protects against a fire it can reach but not against one that is fully developed. The clean agent buys time and saves the gear on the small event. The sprinkler keeps the building from burning down on the large one.

Clean agent types and the HFC phase-down

Clean agents split into two families: halocarbons that interrupt the fire chemically, and inert gases that put it out by lowering oxygen to a level that will not sustain flame but a person can still breathe briefly. Which family you specify is now as much a regulatory question as a fire one, because the most common halocarbon agent is being phased down.

FM-200, the halocarbon HFC-227ea, was the workhorse data center agent for years. It works and it is compact, but it is a hydrofluorocarbon with a high global warming potential, around 3,200, and HFC production is being cut under the AIM Act, which steps domestic HFC production down by roughly 85 percent by 2036. That does not make an installed FM-200 system illegal tomorrow, but it does mean the agent gets scarcer and pricier to recharge, which matters a lot when your protection model assumes you can refill after a discharge. The low-GWP halocarbon alternative is FK-5-1-12, the fluoroketone formerly sold as Novec 1230, with a global warming potential near 1.

Inert gas systems sidestep the phase-down entirely with zero ozone impact and a global warming potential of effectively zero. The common blends are IG-541 (nitrogen, argon, and carbon dioxide, sold as Inergen), IG-55 (argon and nitrogen), and IG-100 (straight nitrogen). The cost is space and discharge dynamics. Inert gas is stored as high-pressure cylinders at roughly 200 to 300 bar and takes far more cylinder room than a halocarbon, and the discharge is loud enough that nozzle noise has been shown to disturb and even damage spinning hard drives, so listed acoustic nozzles and placement matter where rotating media is present. Pick the agent for the room, the recharge model, and the regulatory horizon, not for what the last job used.

What is a clean agent room integrity (door fan) test?

A room integrity test, also called a door fan test, measures how leaky a protected enclosure is, because a clean agent only works if the room holds the gas at extinguishing concentration long enough to put the fire out. A gas does not stay put the way water does. If the room leaks, the agent drains out through unsealed cable penetrations, gaps above the suspended ceiling, poor door seals, and open dampers, and the concentration falls below what is needed before the fire is dead.

The test puts a calibrated fan in a doorway, pressurizes and depressurizes the enclosure, and from the measured leakage calculates the retention time, the predicted minutes the room will hold agent. NFPA 2001 sets the performance target the test confirms: hold a minimum of 85 percent of the adjusted minimum design concentration at the highest level of combustibles for at least 10 minutes. Fall short and the room fails until the leaks are found and sealed. The test is run at commissioning and repeated periodically, commonly annually, because rooms do not stay sealed. Every cable pull and every contractor who props a ceiling tile changes the leakage.

This is the step that gets skipped, and skipping it is how a system that discharges perfectly still fails to extinguish. The cylinders fire, the gas floods, and the concentration is gone in three minutes through a cable trench nobody sealed. A clean agent with no integrity test is a very expensive noise. If the test was not done, or the seal package was never finished, the room is not protected, no matter what the nameplate on the cylinder bank says.

Pre-action sprinklers and the interlock logic

A pre-action sprinkler system keeps the sprinkler piping dry, filled with supervised air instead of water, and admits water only after a separate fire-detection event opens the pre-action valve. That dry pipe is the whole point in a data hall. A sprinkler head that is bumped, corroded, or cracked over a rack leaks air, not water, and the supervision alarms instead of soaking the row.

The interlock logic comes in two forms, and the difference is what it takes to fill the pipe. A single interlock system admits water when the detection system operates. A double interlock system admits water only when both the detection operates and a sprinkler head opens, so a false detection alone, or a single head failure alone, will not fill the pipe. Double interlock sounds safer and is sometimes specified for the driest tolerance, but it is slower to deliver water, and NFPA 13 penalizes the design with a larger required area of operation, since dry-pipe and double-interlock systems get a 30 percent increase in the design area. For computer rooms that otherwise require wet pipe protection, guidance commonly steers toward single interlock pre-action rather than double for that reason.

Whichever you use, the interlock is only real if it is proven. A trip test confirms the valve actually opens on the right inputs and water reaches the inspector's test connection in the time the standard expects, on the order of 60 seconds to the test connection for the dry-side designs. An interlock that was wired but never trip-tested is an assumption, not a system. For the full trip-test method and acceptance criteria, see the dry-pipe and pre-action trip test guide.

Clean agent vs pre-action: which does a data hall need?

Clean agent and pre-action are not competitors so much as different jobs, and most serious data halls end up with both. Clean agent is the early, gear-saving response. It floods the room and snuffs a small fire, a smoldering power supply or a cable fault, while it is still small, and it leaves the surviving equipment dry and clean. Pre-action is the building's structural fire protection, the system the code and the insurer require to keep the structure from being lost in a fully developed fire.

The trap is treating one as a substitute for the other. A clean agent system has one shot. It discharges, and unless and until it is recharged the room is unprotected, and it cannot fight a large, ventilation-fed fire the way water can. A pre-action system protects the structure but wets the gear when it actually flows. So the clean agent handles the event you can catch early, and the pre-action is there for the event you could not. Removing the sprinkler because there is a gas system is how a contained incident becomes a lost building.

Where budget forces a single system, the building code and the AHJ usually settle it, and the answer is most often the sprinkler, because life safety and structural protection outrank equipment protection. The clean agent is the upgrade that protects the asset. The sprinkler is the baseline that protects the building and the people. Decide it on the risk model and the adopted code, not on which discharge looks more impressive in the brochure.

What is the fire alarm cause-and-effect matrix?

The cause-and-effect matrix is the table that defines exactly what every fire input makes the building do, input by input and output by output, and it is the single most important document in a data center fire scheme. NFPA 72, the National Fire Alarm and Signaling Code, governs the detection, the notification, and the control functions, and the matrix is how the abstract sequence of operations becomes a row-by-row specification a technician can test against.

A typical sequence runs in stages. A first detection, often called Fire 1 from a single ASD zone or detector, raises a supervisory or first-stage alarm and tells someone to investigate. A confirmed second detection, Fire 2, escalates: it shuts down the air handlers so the suppression is not blown out of the room, closes the smoke and fire dampers so the fire and its products stay in the cell, releases the door holds, signals the suppression to begin its discharge time delay, and notifies the monitoring station. Two-stage cross-zoned detection before any agent release is what keeps a single faulty detector from dumping the room.

The matrix is only worth what the integrated test proves. Every cause has to be initiated for real and every effect watched and recorded: trip the detector, confirm the fan stops, confirm the damper drives closed, confirm the abort and the discharge delay behave, confirm the signal reaches the panel and the monitoring station. A matrix that was drawn and signed but never demonstrated end to end is the most common place the scheme fails, because the wiring that proves out one device at a time can still have the cross-zone logic backwards. Test the sequence, not just the devices.

HVAC shutdown, dampers, and smoke control

The cooling system and the fire system are wired together because air movement decides whether suppression works. On a confirmed alarm the air handlers serving the room shut down, so the clean agent is not diluted and swept out the moment it discharges and the sprinkler is not fighting a fan-fed fire. The same airflow the cooling design fights to deliver is the airflow that defeats both detection and suppression if it keeps running through the event.

Dampers do the containment. Smoke and fire dampers in the duct close on the alarm to keep smoke from being distributed through the building and to keep the protected room sealed enough for the agent to hold. NFPA 72 ties the duct smoke detectors into the alarm and requires the damper to be actuated and proven as part of the detector testing, and the detectors sit close to the dampers they drive. A damper that is wired to a detector but rusted open is a leak path the integrity test will find and the fire will use.

Because cooling and fire are coupled this tightly, the two systems get commissioned against each other, not in isolation. The fan-shutdown and damper-closure steps live in both the fire cause-and-effect matrix and the mechanical sequence of operations, and they have to agree. The cooling design and its airflow strategy are covered in the data center cooling and airflow overview. The point here is that you cannot prove the fire scheme without driving the mechanical system through its fire response.

Do you need a fire watch when the system is down?

Yes. When a required fire protection system is out of service, the standard answer is a fire watch plus a managed impairment until the system is back, and on a data center under construction the systems are down or incomplete for most of the build. A fire watch is a trained person, or people, patrolling the area with the authority to raise the alarm and the means to fight an incipient fire, standing in for the automatic protection that is not yet working.

Two programs drive this. During construction and major renovation, NFPA 241, the construction fire safety standard, governs the site, and it treats the incomplete fire protection, the combustible loading, the active ignition sources, and the transient workforce as exactly the high-risk condition a fire watch and a fire-prevention program exist to cover. After turnover, NFPA 25 governs impairments of the water-based systems: when a sprinkler or standpipe is shut off or out of service, you tag it, notify the AHJ and the insurer, and put compensating measures in place, with a fire watch commonly triggered when the impairment runs long, often cited around 10 hours in a 24-hour period. The AHJ and the insurer set the trigger.

The discipline is the documentation, not just the patrol. A fire watch that is not logged, by patrol, by time, by area, did not happen as far as the AHJ or the insurer is concerned, and a missed patrol during an impairment is the gap that shows up after a loss. This is exactly the record the firewatch-patrol-log tool is built to keep, and the impairment paperwork is the kind of finding tradeos exists to track to closure. Treat the watch as a compensating measure that has to be provable, because when there is a fire during an impairment, the log is the first thing everyone asks for.

Fire pump and water supply

Where the building relies on sprinklers or standpipes, the fire pump and its water supply are part of the life-safety chain and get commissioned like any other critical system. The pump exists because the municipal supply often cannot make the pressure and flow the sprinkler design demands at the top and the far end of a large building, and a data center campus can be both tall and sprawling.

Acceptance is a flow test. The pump is run across its curve, at churn with no flow, at rated flow and pressure, and at 150 percent of rated flow, and the measured points are compared to the certified pump curve to prove it makes its numbers. The controller is exercised, the automatic start on a pressure drop is proven, and on an electric pump the transfer to standby or generator power is demonstrated, since a fire pump that cannot start when the utility drops is no pump at all. Diesel-driven pumps add their own fuel, battery, and run-test regime.

After turnover the pump falls under the NFPA 25 testing cadence, including a regular no-flow churn run and a periodic full flow test against the curve. The number that matters over the life of the building is the trend. A pump that is quietly losing capacity shows up as a flow test that drifts below the curve, and only the baseline from acceptance lets you see the drift. Skip the acceptance flow test and you have no reference to maintain against.

Compartmentation, rated barriers, and firestopping the cable penetrations

Fire-rated walls and floors divide a data center into compartments so a fire in one cell cannot spread to the next, and the rating is only as good as the penetrations through it. A two-hour wall with an unsealed cable opening is not a two-hour wall. It is a hole with a fire rating written on the drawing, and in a data center those walls are perforated by cable trays, conduit, and busway by the hundreds.

Firestopping is the engineered seal that restores the rating where something passes through. Each penetration gets a listed firestop system, a specific combination of sealant, wrap, collar, or pillow tested as an assembly for that wall type and that penetrant, not a generic smear of caulk. The detail matters because the listing is for the tested configuration. The wrong product or the wrong annular space voids it. On-site inspection of installed firestops is itself standardized, commonly to ASTM E2174 for through-penetrations and ASTM E2393 for joints, so the inspector works to a method rather than a guess.

This is the work that gets defeated continuously after turnover, not just during construction. Every cable add and every move pulls new conductors through rated walls, and the crew that opened the firestop to pull the cable is rarely the crew that reseals it correctly. The result is a building full of rated barriers slowly losing their rating one cable pull at a time. The same penetrations are the leak paths that fail the clean agent room integrity test, so firestopping and gas retention are the same fight. Seal the penetrations or lose both the barrier and the agent.

Means of egress and life safety

A data hall is mostly machines, but the few people in it have to be able to get out fast, and NFPA 101, the Life Safety Code, governs that egress regardless of how empty the room usually is. The code sets the exit count, the travel distance to an exit, the door swing and hardware, the marking and illumination of the path, and the emergency lighting that keeps the path visible when the normal lighting drops.

Two details bite in data centers specifically. Aisle containment and dense rack rows can create long dead-end travel and tight aisles that blow past the allowed travel distance if the layout was drawn for cooling and rack count without checking egress. And the security that data centers depend on, mantraps, badge-in doors, and locked aisles, has to release on a fire alarm so a locked door never traps a worker, which puts the access-control system squarely inside the fire cause-and-effect matrix. A door that stays locked on alarm is a life-safety failure, not a security feature.

Clean agent rooms add one more: the discharge itself. Inert-gas systems work by lowering oxygen, and the discharge is loud and disorienting, so the egress design assumes people leave promptly on the pre-discharge alarm. The pre-discharge warning, the abort station, and the time delay before discharge exist so the room can be cleared first. Egress and suppression are designed together, because the agent that saves the gear is not something you want to be standing in.

Commissioning and the integrated fire and life-safety test

Fire and life-safety systems get commissioned the same way the power and cooling do: witnessed, scripted, and proven end to end, not just inspected as installed. The integrated test is where it counts, because the fire scheme is a chain of handoffs between detection, alarm, suppression, HVAC, dampers, doors, and power, and the failures live in the handoffs, not the individual devices.

A real integrated test drives each initiating cause and confirms every effect in the cause-and-effect matrix actually happens, in order and in time. Trip a confirmed alarm and watch the air handlers stop, the dampers close, the door holds drop, the access-control doors release, the suppression count down through its delay and abort, the notification reach the panel and the monitoring station, and the emergency power pick up the life-safety loads. Each step is recorded against the matrix with a witness signature, the same discipline a power integrated systems test uses on the electrical plant.

The fire test does not stand alone, because the fire response reaches into the mechanical and electrical systems. The fan shutdown is a mechanical function, the damper closure is shared, and the life-safety power is electrical, so the fire integrated test is sequenced with the cooling and power commissioning rather than run in a vacuum. The data center electrical commissioning and power QA guide covers the power-side integrated systems test. The fire matrix has to be proven against the same plant, on the same gear, so the two sequences agree. A fire scheme proven on paper and never integrated-tested is the one that fails on the night it is needed.

Lithium-ion and BESS fire considerations

Lithium-ion batteries are showing up in data centers as UPS replacements and as larger battery energy storage systems, and they bring a fire problem that gas and sprinkler schemes were not originally designed for: thermal runaway. A failing lithium cell vents flammable gas, generates its own heat, and can propagate cell to cell, producing a fire that reignites and that a clean agent flood will knock down but not necessarily stop, because the chemistry keeps making heat and fuel from inside the pack.

NFPA 855, the Standard for the Installation of Stationary Energy Storage Systems, is the governing document, and it pulls in detection, suppression, gas detection, explosion control, spacing, and a hazard mitigation analysis that the 2026 edition makes the default approach. The standard applies above threshold quantities that vary by chemistry, on the order of 20 kilowatt-hours for lithium-ion in some occupancies, so a small UPS battery may fall below it while a real BESS does not. Confirm the threshold and the adopted edition, because this standard is moving fast.

The practical lesson is that a battery room is a different fire problem than a data hall and has to be treated as one. The suppression and detection sized for an electronics fire do not automatically cover a runaway pack, deflagration of vented battery gas is a real explosion hazard that needs ventilation and explosion control, and the protection and thermal interlocks are tested with the electrical performance, not after it. The BESS commissioning punch list is heavy enough to deserve its own workflow, and the data center electrical commissioning and power QA guide covers the battery side in depth.

What fire-system testing does the owner inherit?

The owner inherits a recurring inspection, testing, and maintenance cadence the day the building turns over, and most of it lives in NFPA 25 for the water-based systems plus the manufacturer and NFPA schedules for detection and clean agent. NFPA 25, the Standard for the Inspection, Testing, and Maintenance of Water-Based Fire Protection Systems, sets the intervals for the sprinklers, the valves, the fire pump, and the water supply, from weekly or monthly checks up through annual and multi-year tests.

The intervals run on a mix of clocks. Sprinkler and pre-action valves get periodic inspection, and the pre-action gets a trip test. The fire pump gets a no-flow churn run on a frequent cycle and a full flow test against its curve annually. Fire alarm and detection get tested under NFPA 72, the clean agent room integrity is re-verified periodically, commonly annually, and the clean agent cylinders get weighed or pressure-checked on schedule. The exact intervals come from the standards and the equipment listings, so build the calendar from the adopted editions, not from memory.

The reason the baseline from commissioning matters is that ITM is a trending exercise. A flow test, a battery impedance reading, a room integrity retention time, each only means something against the number the system made when it was new. The acceptance records are the reference. An owner who gets a building with no commissioning baseline is maintaining blind, comparing this year's reading to nothing. The handoff that actually protects the building is the complete record set, not just the working hardware.

What to document

The fire and life-safety record is what the owner defends the building with and what the next contractor maintains against, so it has to outlive the crew that built it. Capture the design basis, every acceptance test, the cause-and-effect matrix as proven, the room integrity result, and the impairment and fire-watch logs from construction, because the question after any incident is what was working and what was tagged out.

Common mistakes

• Specifying a clean agent system and never running the room integrity (door fan) test, so the agent drains before it extinguishes.
• Detecting with spot detectors tuned for a still room instead of aspirating detection sized for the airflow, so the alarm lands minutes late.
• Wiring the pre-action interlock and never trip-testing it, so nobody has proven the valve fills the pipe on the right inputs.
• Drawing the cause-and-effect matrix and signing it without an end-to-end integrated test of every cause and effect.
• Skipping firestopping at cable penetrations, which defeats both the rated barrier and the clean agent room integrity.
• Running no fire watch, or an unlogged one, during a construction impairment, so there is no compensating measure on record.
• Treating the clean agent as a replacement for the sprinkler, leaving the structure unprotected on a developed fire.
• Putting a wet pipe sprinkler charged with water directly over live racks where a single damaged head soaks the row.
• Securing egress and access-control doors that never release on a fire alarm, trapping workers behind the security.
• Installing a lithium-ion battery room without NFPA 855 detection, explosion control, and thermal interlocks proven at commissioning.

Field checklist

Standards and references

The occupancy standards come first. NFPA 75, the Standard for the Fire Protection of Information Technology Equipment, covers the data center room and its protection, and NFPA 76, the Standard for the Fire Protection of Telecommunications Facilities, covers carrier and central-office spaces. Detection and notification follow NFPA 72, the National Fire Alarm and Signaling Code, which also governs the control sequence behind the cause-and-effect matrix.

Suppression splits by type. Clean agent systems follow NFPA 2001, the Standard on Clean Agent Fire Extinguishing Systems, which sets the agent concentration and the hold the room integrity test confirms. Water-based sprinkler systems, including pre-action, are installed to NFPA 13, the Standard for the Installation of Sprinkler Systems, while water mist falls under NFPA 750 and is not treated as a sprinkler system. After turnover, the water-based systems are inspected, tested, and maintained to NFPA 25.

The rest of the stack rounds out the life-safety picture. Construction fire safety, the fire watch, and the impairment program come from NFPA 241. Means of egress and occupant protection come from NFPA 101, the Life Safety Code. Stationary battery energy storage is governed by NFPA 855, and emergency and standby power for the life-safety loads by NFPA 110. The electrical work is built to NFPA 70, the NEC, with worker safety under NFPA 70E, and the property insurer's FM Global data sheets often impose criteria stricter than code. Edition numbers and clause references change between cycles, so confirm the adopted edition and any local amendments against the project documents, and let the AHJ govern where documents conflict.

Units, terms, and acronyms

Data center fire and life safety carries its own vocabulary, and the same idea reads differently across a fire alarm submittal, a clean agent cut sheet, and an insurer data sheet. The terms below are the ones that travel across the whole scheme.

ASD / VESDA

Aspirating smoke detection, which actively draws air through sampling pipe for very-early warning; VESDA is a common brand

Clean agent

A gaseous fire-extinguishing agent that leaves no residue, either a chemical halocarbon or an oxygen-lowering inert gas

Halocarbon vs inert gas

Halocarbons (FM-200, FK-5-1-12) interrupt the fire chemically; inert gases (IG-541, IG-55, IG-100) lower oxygen below combustion

Pre-action

A sprinkler system whose pipe stays dry until a detection event opens a valve, so a damaged head does not leak water

Single vs double interlock

Single fills the pipe on detection alone; double requires both detection and an open sprinkler head before water is admitted

Cause-and-effect matrix

The table mapping every fire input to every output, the testable form of the fire sequence of operations

Room integrity / door fan test

A pressurization test that measures enclosure leakage to confirm a clean agent room holds concentration for the required time

Impairment

Any condition where a fire protection system is out of service, requiring tagging, notification, and compensating measures

Thermal runaway

A self-sustaining lithium battery failure that makes its own heat and flammable gas and can propagate cell to cell

AHJ

Authority having jurisdiction, the official who adopts the editions, approves the design, and settles conflicts

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How long does data center electrical commissioning take?
• What is bypass airflow in a data center?
• What fire suppression do data centers use?
• What is the difference between a fire damper and a smoke damper?
• What is a door fan test?