Datacenter
Fire watch and impairment program field guide for data centers
When a fire watch is required, how the impairment program runs, the hot work watch, and the patrol log that proves the watch actually happened.
Direct answer
A fire watch is a trained person who patrols an area and watches for fire when the normal fire protection is impaired, during hot work, or when the authority having jurisdiction requires it, with the authority to sound the alarm and call the fire department. The AHJ and the insurer set the trigger and the duration.
Key takeaways
- A fire watch is a trained person who patrols an area, watches for fire when protection is impaired, and has authority to alarm and call the fire department.
- NFPA 25 commonly triggers a compensating measure when a water-based system is out of service more than 10 hours in a 24-hour period, counted cumulatively.
- NFPA 51B requires combustibles within a 35 ft radius of hot work to be removed, or covered with fire-resistant material where they cannot be moved.
- Post-work hot work watch is 30 minutes under older NFPA 51B and OSHA, extended to 1 hour in 2019 and later editions, plus up to 3 hours monitoring.
- End the watch only after the system is confirmed restored, proven, and rearmed, the tag is removed, and restoration notifications are made.
The fire watch, and what it actually stands in for
A fire watch is a trained person, or a team of them, who patrols an area and watches for fire when the normal fire protection cannot be relied on, with the authority to raise the alarm and call the fire department. It is a compensating measure. The building lost some of its automatic protection, or it never had it yet because it is still being built, so a human takes over the one job the missing system was supposed to do: notice a fire early and get help moving.
That is the whole idea, and it is narrower than people assume. A fire watch is not a guard checking badges, not a safety walk, and not the same person who happens to be in the room. It is a dedicated assignment with a defined area, a defined route, a defined frequency, and the means to act. The watcher does not fight a developed fire. The watcher catches the incipient one, sounds the alarm, gets people out, and calls the fire department, in that order.
On a data center this comes up constantly, because the fire protection is down or incomplete for most of a build, and even after turnover a clean agent room gets disabled for service and a pre-action zone gets drained for a trip test. Every one of those is a window where the automatic protection is not protecting, and the watch is what fills the window. The failure is almost never the concept. It is that the watch was understaffed, untrained, unequipped, or unlogged, so when something happened there was no one watching the right place and no record that anyone ever was.
When is a fire watch required?
A fire watch is required in three situations: when a required fire protection system is impaired beyond the limit the code allows, during and after hot work, and any time the authority having jurisdiction orders one. Those are the triggers a foreman actually meets, and they come from different standards.
The impairment trigger is the common one on an operating facility. When a water-based fire protection system is out of service past a set duration, the impairment program requires a compensating measure, and a fire watch is the usual one. NFPA 25 ties this to time out of service, commonly cited at more than 10 hours in a 24-hour period, but confirm the adopted edition and let the AHJ and insurer settle the number, because some insurers and jurisdictions are stricter.
Hot work is its own trigger. Welding, cutting, grinding, and brazing demand a fire watch during the work and for a period after it, under NFPA 51B and OSHA. Construction is a third: on a building under construction or major renovation, the systems are not commissioned, and NFPA 241 treats that as the high-risk condition a fire watch and a fire prevention program cover. And above all of these sits the AHJ, who can require a watch for an occupancy, an event, or a condition whenever they judge the risk warrants it. When the AHJ says watch, you watch.
| Trigger | Governing standard | What starts the watch |
|---|---|---|
| Impaired water-based system | NFPA 25 (after turnover) | System out of service beyond the allowed duration |
| Hot work | NFPA 51B / OSHA | Welding, cutting, grinding within range of combustibles |
| Construction / renovation | NFPA 241 | Fire protection incomplete or not yet commissioned |
| AHJ order | NFPA 1 / IFC | Any condition the AHJ judges to need a watch |
The impairment program
An impairment is any condition where a fire protection system, or part of one, is out of service. The impairment program is the discipline that manages it: who authorizes it, who gets told, how it is tagged, what compensating measures cover the gap, and how it gets closed. NFPA 25 lays out the program for water-based systems in its impairment chapter, and it splits impairments into two kinds that get handled differently.
A preplanned impairment is the one you schedule. You are draining a pre-action zone for a trip test, or valving off a sprinkler main to tie in a new branch. The program requires that someone authorize it before it starts, that the extent and expected duration are known, that the affected area is inspected for increased risk, and that the people who need to know are notified ahead of time. An emergency impairment is the one that finds you: a broken main, a failed valve, a leak that drops the system. The difference is that you cannot plan it, so the response is reactive, but the same notifications and compensating measures still apply once it is discovered.
The program names an impairment coordinator, the person responsible for running this. The property owner assigns one, and absent a named designee, the owner is the coordinator by default. The coordinator authorizes the planned impairment, makes the notifications, confirms the compensating measures are in place, and signs off the restoration. On a construction site the role often lands on the general contractor or the fire prevention program manager, but it has to land on someone, by name, with the authority to stop the work if the protection is not covered.
When does an impairment require a fire watch?
An impairment requires a fire watch when a water-based fire protection system is out of service beyond the duration the code allows and no other compensating measure is in place. NFPA 25 commonly cites the trigger at more than 10 hours of impairment in a 24-hour period, measured as cumulative time out of service, not necessarily consecutive. Confirm the figure against the adopted edition, because the number has moved between cycles and the insurer or AHJ can set a stricter one.
When the impairment passes that threshold, the coordinator has to arrange one of a short list of compensating measures, not the fire watch alone. The options usually run: evacuate the affected area, establish a fire watch, provide a temporary water supply to the impaired system, or run an approved program to remove ignition sources and limit the fuel available. On a working data center you cannot evacuate the load and you cannot easily re-water a drained system mid-test, so the fire watch is the practical answer most of the time.
Be careful with the clock. The 10-hour figure is a common reference, but a tighter reading and a stricter insurer can require the watch much sooner, and some properties require a watch the moment the system goes down, with no grace window at all. The blunt version: do not let the crew treat the threshold as free time before anyone has to watch. Set up the watch when the system goes out, and let the paperwork catch up, not the other way around.
Hot work, the permit, and the 35 ft rule
Hot work is any operation that makes a flame, a spark, or enough heat to ignite: welding, cutting, brazing, soldering, grinding, and torch work. It is one of the most common ignition sources on a construction site, and the controls around it come from NFPA 51B and the OSHA hot work rules. The center of the control is the permit, and the permit is worthless if it is signed and filed instead of worked.
The hot work permit is a checklist the permit authorizing individual runs before the work starts and keeps active while it runs. It confirms the area is safe to burn in: combustibles moved or covered, floor and wall openings protected, the right extinguisher staged, the detection or sprinkler status known, and a fire watch assigned. The clearance number to carry in your head is 35 ft. NFPA 51B calls for combustible and flammable material within a 35 ft radius of the work to be removed, and where it cannot be moved, covered or shielded with fire-resistant material. Where openings, cracks, or conveyors can carry sparks past that radius, the protection has to follow the spark path, not just the circle on the floor.
The watcher is required during the hot work, not just after. While the torch is running, the fire watch stays in position with eyes on the work and the surrounding area, an extinguisher in hand, and the means to alarm. The most common failure is treating the watch as a post-work task and leaving the work unwatched while it is hottest. The sparks that start fires fall during the cut, and they travel.
How long do you watch after hot work?
The fire watch continues after the hot work stops, because the dangerous period is not the cut, it is the slow smolder a hidden ember starts behind a wall or in a pile of debris. NFPA 51B and OSHA long called for a 30-minute watch after the work ended, and OSHA's hot-work rules still set that 30-minute minimum. The 2019 and later editions of NFPA 51B extended the post-work fire watch to a full hour, so the safe planning number on a current project is 60 minutes, with a 30-minute floor under older references.
The standard goes further than the hour. Recent NFPA 51B editions add a fire monitoring period of up to an additional 3 hours after the 1-hour watch, where the permit authorizing individual judges the hazard warrants it, because some assemblies hold heat and a smolder can take that long to show. Monitoring is a lighter touch than the continuous watch, but it is still someone responsible for checking the area, not nobody. Confirm both the post-work duration and any extended monitoring against the adopted edition and the permit.
After the watch and any monitoring, there is a final check, and it is the step that closes the permit. Before the watcher leaves, the area gets a deliberate sweep: feel for heat, look behind and under anything the sparks could have reached, confirm there is no hidden ignition, and only then sign the permit closed. A permit closed at the moment the torch went out, with the area never re-checked, is the one that turns into a 2 a.m. callback.
Construction fire safety under NFPA 241
On a building under construction, alteration, or demolition, the fire protection that will eventually run the place is not there yet, and NFPA 241, the Standard for Safeguarding Construction, Alteration, and Demolition Operations, governs the fire safety of that gap. A data center build is a near-perfect example of why the standard exists: incomplete sprinkler and detection, heavy combustible loading from packaging and temporary materials, constant hot work, temporary power and heat, and a large transient workforce that turns over daily.
NFPA 241 requires a written fire prevention program and a named fire prevention program manager, the FPPM, who owns it. The FPPM is meant to be a boots-on-the-ground person with the authority and the budget to actually run the program, not a title on an org chart. The program covers site fire prevention, the hot work program, housekeeping and combustible control, site access for responding apparatus, the temporary standpipe and water supply as floors go up, and the fire watch and impairment procedures for the systems that are not commissioned.
The construction watch is the version of the fire watch that covers the whole exposed period, not a single hot work job. Until the permanent detection and suppression are commissioned and accepted, large portions of the building have no automatic protection, so the program leans on patrols, temporary detection where it makes sense, strict hot work control, and a watch sized to the risk. The standard is the framework. The FPPM and the AHJ decide what the watch looks like for the specific site and phase, and that decision should be written down, not improvised by whoever is on shift.
The fire watch duties
The watcher has a defined set of duties, and all of them reduce to one priority: detect a fire early and get help moving before it grows. Everything else supports that. The watcher patrols an assigned route on a set frequency, looks for the early signs of fire, keeps the means to alarm and the means to fight an incipient fire within reach, and logs each round.
The patrol route is the heart of it. It is laid out to cover the whole impaired area, including the places fire hides, ceiling spaces, electrical rooms, storage, the work area itself, and it is walked on a frequency the AHJ or the program sets. A common frequency is rounds every 30 to 60 minutes, sometimes tighter for high-hazard areas, but the interval is set by the program and the AHJ, not by habit. What the watcher looks for is concrete: smoke, the smell of overheated insulation or burning, heat, sparks, smoldering debris, a hot work area that was left, anything off-normal in the impaired space.
Then there is the authority, and this is what separates a fire watch from a person walking around. The watcher must be able to raise the alarm and must know how, whether that is a manual pull, a horn, a radio call to a manned post, or a direct call to the fire department. The watcher must have standing authority to call the fire department directly without waiting for permission, and to order an evacuation. A watch with no means to alarm and no authority to call is not a fire watch. It is a witness.
What does a fire watch log need?
A fire watch log needs each round recorded with the time, the area patrolled, the watcher's name, and what was found, captured at the moment the round was made. That last part is the whole point. A log that proves the watch is a log written in real time, round by round, not a sheet filled in at the end of the shift to look complete.
The minimum useful entry per round is the time the round started, the area or route covered, the conditions observed, the watcher's name or initials, and any action taken. Across the shift the log also carries the impairment reference it is covering, the start and intended end of the watch, the means of communication available, and the handoff when one watcher relieves another. The AHJ may specify a format, and many will accept a written log or an electronic one as long as the timestamps are real.
A missed round, or a round back-filled after the fact, is the failure, not a paperwork foot-fault. If a fire starts in the hour nobody patrolled, the log shows the gap, and a log full of identical entries written in one hand at one time tells the investigator the watch was not really run. This is exactly the record the firewatch-patrol-log tool is built to keep honest, with each scan or round timestamped where it happened so a round cannot be quietly invented later, and the open impairment behind it is the kind of item tradeos exists to track to closure. Treat the log as evidence, because after a loss that is precisely what it becomes.
Coverage and staffing
Coverage is the question of how much area one watcher can actually cover, and it is where well-meaning programs quietly fail. One person cannot watch a 200,000 sq ft floor, and one person cannot be in the electrical room and the data hall at the same time. The area per watcher and the number of watchers come from the size and layout of the impaired space, the hazard, and what the AHJ accepts, not from what staffing the schedule wanted to spend.
There are two basic patterns. A continuous, fixed watch keeps a watcher stationed at a single high-hazard point, line of sight on the work or the hazard, common during active hot work where the watcher stands right at the operation. A roving patrol walks a route covering a larger impaired area on a set frequency, common for a system impairment that affects a whole floor. Big sites use both: a roving patrol for the floor plus a fixed watch at each active hot work job. The two are not interchangeable. A roving patrol does not cover a live torch, and a single fixed post does not cover a floor.
Line of sight and travel time set the limits. If the route is so long that 30 minutes pass between visits to a high-hazard area, that area is effectively unwatched for half an hour at a time, and a fire has that long to grow. The honest test is whether one watcher can reach and see every part of the assignment often enough to matter. If they cannot, the answer is another watcher or a shorter route, not a hope that nothing happens between rounds. Staff the watch to the area, and write the staffing basis into the plan so it can be defended.
The watcher's equipment
A fire watch is only as good as what the watcher can do when they find fire, and that comes down to four things in hand: a way to raise the alarm, a way to fight an incipient fire, a way to communicate, and the PPE to do it safely. Missing any one of them turns the watch into a spectator.
The means to alarm is first. The watcher needs a reliable way to alert the building and summon the fire department, sized to the site: a charged manual pull station on the route, an air horn or whistle where there is no alarm system yet, a radio to a manned fire command post, and a working phone to call 911 directly. On a construction site where the permanent alarm is not commissioned, this is often a temporary system, and it has to be tested, not assumed. The means to fight a small fire is the staged extinguisher, the right type and rating for the hazard, charged and within reach of the work, with the watcher trained to use it on an incipient fire only.
Communications and PPE round it out. The watcher needs a charged radio or phone with known coverage in the space, because a watcher who finds a fire in a basement dead zone and cannot call out is no help. PPE matches the environment, which on a construction data center means hard hat, eye protection, hi-vis, and the protection the area demands, plus a flashlight for the dark and unfinished spaces where the watch matters most. None of this is exotic. It is the gear that has to be present and working before the watch starts, checked at the start of the shift, not discovered missing during a fire.
Impairment in an operating data center
In a live data center the impairment problem is sharper than on a construction site, because you cannot evacuate the load and you cannot tolerate the protection being down for long. The systems get impaired anyway, routinely, for testing and service, and each one opens a window the watch has to cover while the room stays full of running equipment.
The pre-action sprinkler out for a trip test is the classic case. To trip-test the valve you take the system out of its normal armed state, and for the duration of the test and the reset the room's water-based protection is impaired. That is a planned impairment: tag it, notify the AHJ and insurer per the program, and post a watch on the affected area until the system is proven back and rearmed. The trip test itself, its method and its acceptance criteria, is covered in the dry-pipe and pre-action trip test guide. The point here is that the test is also an impairment, and the watch is part of running it correctly.
The clean agent system disabled for service is the other common one. When a gaseous system is taken offline for maintenance, a cylinder swap, or a panel update, the early gear-saving protection is gone, and the room is back to whatever sprinkler or manual response remains. A clean agent room with its system disabled and no compensating watch is a room betting that nothing fails during the service window. Tag the disabled system, cover the room with a watch, and confirm both the agent system and any cross-connected detection are fully restored and armed before the watch ends.
What does the AHJ or insurer ask for after an incident?
After a fire or a near miss during an impairment, the first thing the AHJ and the insurer ask for is the paper: the impairment record and the fire watch log for the window the incident fell in. They are reconstructing what protection was in place, what was tagged out, and whether the compensating measures the program promised were actually running. If the watch was real and logged, the record protects everyone. If it was not, the record is the finding.
Expect to be asked for the impairment authorization and tag, showing what system was out and when, who authorized it, and who was notified. Expect to be asked for the fire watch log covering the incident, with the rounds, the times, and the conditions, and for proof the watcher was trained and equipped. On a construction project, expect the NFPA 241 fire prevention program, the hot work permits for any burning near the origin, and the FPPM's records. The insurer in particular, often FM Global on these assets, may have policy conditions that require the watch and the notifications, and a coverage question can turn on whether they were met.
The blunt reality is that the documentation is not bureaucracy, it is the defense. A facility that ran a real watch and kept a real log walks into that conversation able to show it did the right thing. A facility that ran the watch but never logged it is in nearly the same position as one that ran no watch at all, because there is nothing to show. Build the record as the work happens, because you cannot build it after the fire when it is the one thing everyone wants to see.
Common impairment scenarios on a data center build
Three impairment scenarios show up on nearly every data center build, and each one is a known window the watch and the program have to cover. Naming them ahead of time is how the schedule stops treating them as surprises.
Detection not yet commissioned is the long one. For most of the construction period the permanent fire alarm and aspirating detection are installed but not accepted, so the building has no automatic way to notice a fire. That whole period is an impairment under NFPA 241, covered by the construction fire watch, the hot work program, and combustible control, until the detection is commissioned and turned over. The sprinkler not charged is the parallel structural gap: the sprinkler piping is up but the system is not yet in service, or a floor is valved off while it is being installed, so the water-based protection is impaired across that area and a watch covers it.
The cutover is the sharp one. When the temporary or partial protection is switched to the permanent commissioned system, or when a zone is taken from one state to another, there is a moment when neither is fully in service. That transition has to be planned as an impairment, with the watch covering the gap and the restoration confirmed before the tag comes off. The cutover is where people assume the new system is live because it was energized, and energized is not commissioned, and commissioned is not armed. Treat the cutover as the highest-risk window of the three, because it is the one where everyone thinks they are protected and the system has not actually been proven.
Restoration: confirming the system is back and ending the watch
Restoration is the close of the impairment, and it is more than turning a valve and removing a tag. The coordinator confirms the system is actually back in service and working, makes the notifications that it is restored, removes the impairment tag, documents the restoration, and only then ends the watch. Each of those steps has to happen, in that order, and the order matters because the watch is the last thing to go, not the first.
Confirming the system is back means proving it, not assuming it. A sprinkler system valved back on gets the control valve confirmed open and locked or supervised, and a main drain test confirms water is actually there. A pre-action system gets confirmed rearmed and in its normal supervised state, not just refilled. A clean agent system gets confirmed back online with its detection and release path armed. NFPA 25 then expects the coordinator to notify that protection is restored: the fire department, the insurance carrier, the alarm company, the owner, and any other AHJ that was told it was down. The tag comes off as part of that close, so a removed tag means a restored system, every time.
The common mistake, and it is a serious one, is ending the watch early, before the system is confirmed back. The crew finishes the test, the valve gets opened, the watcher gets sent home, and then the rearm has a problem, or the valve was left half-open, or the system trips back out, and now the room is impaired with nobody watching and no one realizing the protection never actually came back. Keep the watch running until restoration is confirmed and documented. The watch ends when the system is proven back, not when the work feels done.
What to document
The impairment and fire watch record is what the AHJ and insurer reconstruct the event from and what defends the project after a loss, so it has to be built as the work happens, not assembled afterward. Capture the impairment itself, every round of the watch, and the restoration, tied together by a single impairment reference so the log and the tag and the close all point to the same event.
At a minimum, record the date and the affected area, the impairment reference and what system was out, the watcher on each round with the round times, the conditions and findings on each round, the means of alarm and communication available, and the restoration with who confirmed the system back and who was notified. If hot work was the trigger, the permit number and the post-work watch and final check belong in the same record. The table below is the spine of it.
| Field to record | Why it matters |
|---|---|
| Date and affected area | Defines where and when protection was reduced |
| Impairment reference and system out | Ties the watch to the specific impairment it covers |
| Round times | Proves the patrol frequency was actually met |
| Watcher name per round | Names who was responsible and confirms a real person walked it |
| Findings and action taken | Shows conditions were observed, not just a box ticked |
| Means of alarm and communication | Proves the watcher could act, not just observe |
| Restoration and notifications | Shows the system was confirmed back and the right parties told |
Common mistakes
- Back-filling the log at the end of the shift instead of recording each round when it is walked.
- Ending the watch when the work finishes, before the system is confirmed restored and rearmed.
- Giving the watcher no means to alarm and no authority to call the fire department directly.
- Assigning a watcher who is untrained, or who is also doing another job and not actually watching.
- Running an impairment with no tag, so nobody downstream knows the system is out of service.
- Treating the 10-hour figure as free time before the watch has to start, instead of staffing the watch when the system goes down.
- Leaving an active hot work job unwatched and only posting a watch after the torch goes out.
- Closing the hot work permit at the moment work stops, with no post-work watch and no final heat check.
- Covering a whole floor with one roving patrol while a live torch burns in one corner unwatched.
- Notifying nobody, so the AHJ and insurer first learn the system was down when they investigate the fire.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
The impairment program for water-based systems lives in NFPA 25, the Standard for the Inspection, Testing, and Maintenance of Water-Based Fire Protection Systems. Its impairment chapter sets the impairment coordinator role, the preplanned and emergency impairment procedures, the tag, the notifications, and the compensating measures, including the fire watch triggered when a system is out of service beyond the allowed duration. The duration is commonly cited at more than 10 hours in a 24-hour period, but it has changed between editions, so confirm the figure against the adopted edition and any stricter insurer or AHJ requirement.
Hot work is governed by NFPA 51B, the Standard for Fire Prevention During Welding, Cutting, and Other Hot Work, together with the OSHA hot work requirements. NFPA 51B sets the permit, the 35 ft clearance, the fire watch during the work, and the post-work watch, which earlier editions held at 30 minutes and the 2019 and later editions extended to 1 hour, with monitoring of up to an additional 3 hours where the permit authorizing individual judges it warranted. Construction, alteration, and demolition fire safety comes from NFPA 241, the Standard for Safeguarding Construction, Alteration, and Demolition Operations, which requires the written fire prevention program and the fire prevention program manager.
The fire code sits over all of it. NFPA 1, the Fire Code, and the International Fire Code give the AHJ the authority to require a fire watch for an occupancy or condition, and the AHJ adopts the editions and settles conflicts. The property insurer, frequently FM Global on data center assets, imposes its own impairment and hot work conditions through its data sheets and policy, and those can be stricter than code and govern by contract. Section and clause numbers shift between cycles, so confirm them against the adopted edition and the local amendments before citing them on a submittal, and let the AHJ and the insurer govern where the documents disagree.
Units, terms, and acronyms
The impairment and fire watch world carries a specific vocabulary, and the same terms travel across the NFPA standards, the insurer's data sheets, and the permit. The ones below are the terms that decide who does what.
- Fire watch
- A trained person assigned to patrol an area and watch for fire, with the authority and means to raise the alarm and call the fire department
- Impairment
- Any condition where a fire protection system, or part of one, is out of service; either preplanned or emergency
- Impairment coordinator
- The person who authorizes an impairment, makes the notifications, ensures compensating measures, and confirms restoration
- Impairment tag
- A tag posted at the system control valve and fire department connection showing the system is out of service
- Hot work permit
- The authorization and checklist that controls welding, cutting, and other spark- or flame-producing work
- Permit authorizing individual (PAI)
- The person under NFPA 51B who authorizes hot work, confirms the area is safe, and sets any extended monitoring
- Compensating measure
- A temporary protection put in place during an impairment, such as a fire watch, evacuation, or temporary water supply
- Fire prevention program manager (FPPM)
- The boots-on-the-ground person required by NFPA 241 to run the construction fire prevention program
- AHJ
- Authority having jurisdiction, the official who adopts the editions, can order a fire watch, and settles conflicts
FAQ
When is a fire watch required?
A fire watch is required when a fire protection system is impaired beyond the allowed duration, during and after hot work, on a construction site before systems are commissioned, and whenever the AHJ orders one. NFPA 25, NFPA 51B, and NFPA 241 cover the cases, and the AHJ and insurer set the trigger.
How long does an impairment have to last before a fire watch is required?
NFPA 25 commonly triggers a compensating measure when a water-based system is out of service more than 10 hours in a 24-hour period, counted cumulatively. A fire watch is one allowed measure. Confirm the figure against the adopted edition, because it has changed between cycles and many insurers require the watch sooner.
How long do you watch after hot work is finished?
Older NFPA 51B and OSHA references called for a 30-minute post-work fire watch, but the 2019 and later editions of NFPA 51B extended it to 1 hour, with monitoring of up to 3 additional hours where the permit authorizing individual judges it needed. Confirm the duration against the adopted edition and the permit.
What is an impairment in fire protection?
An impairment is any condition where a fire protection system, or part of one, is out of service, planned or emergency. Under NFPA 25 it has to be authorized, tagged at the control valve and fire department connection, notified to the AHJ and insurer, and covered by a compensating measure until the system is restored.
What does a fire watch log need to contain?
A fire watch log needs each round recorded with the time, the area patrolled, the watcher's name, and the conditions found, written when the round is walked. It should also carry the impairment reference, the means of alarm available, and the restoration. A back-filled log is the failure the AHJ looks for after a loss.
What is the 35 ft rule for hot work?
NFPA 51B requires combustible and flammable material within a 35 ft radius of hot work to be removed, and where it cannot be moved, covered or shielded with fire-resistant material. Where sparks can travel past that radius through openings or conveyors, the protection has to follow the spark path, not just the circle on the floor.
Who is responsible for the impairment program on a construction site?
On a construction site the responsibility usually falls on the general contractor or the fire prevention program manager required by NFPA 241, working as or with the impairment coordinator named under NFPA 25. Someone has to be named, with the authority to authorize impairments, make notifications, and stop work if the protection is not covered.
Can one fire watcher cover a whole data center floor?
Usually not. The area per watcher depends on the size, layout, and hazard of the impaired space and what the AHJ accepts, not on the staffing the schedule wanted. A roving patrol cannot cover a live hot work job, and a single post cannot cover a floor. Staff the watch so every high-hazard area is reached often enough to matter.
When can you end a fire watch during an impairment?
End the watch only after the system is confirmed restored, proven working, and rearmed, the tag is removed, and the restoration notifications are made. Ending it when the work feels done, before the system is verified back, is a common and serious mistake, because a failed rearm leaves the area impaired with nobody watching it.
What do the AHJ and insurer ask for after a fire during an impairment?
They ask for the impairment record and tag, the fire watch log covering the incident, proof the watcher was trained and equipped, and any hot work permits near the origin. On a data center the insurer, often FM Global, may tie coverage to whether the watch and notifications were done, so the documentation is the defense.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.