What a Supervisory Signal Is and Why It Is Not an Alarm

Why this matters

Fire protection systems spend their whole service life doing nothing visible, which means their real failure mode is not breakage, it is unavailability that nobody notices. A closed valve does not leak, does not make noise and does not fail an inspection anyone happens to run that day. The supervisory signal exists to make that condition visible, and it is the only mechanism in the building that does. Treating it as a nuisance to acknowledge and clear is treating the impairment detection system as noise. This card covers the three signal classes, what physically produces a supervisory, which standard owns which impairment, and what a filled in impairment record looks like when the process is actually followed.

Three classes, and what each one is a statement about

NFPA 72, the National Fire Alarm and Signaling Code, defines three signal classes, in the edition your authority having jurisdiction has adopted and amended. The AHJ is a named role with authority to approve, commonly the fire marshal or the building official, and not a synonym for whoever inspects.

The supervisory and trouble split is the one that gets blurred, and it is clean once you name it: supervisory is about the protection, trouble is about the reporting. A closed sprinkler control valve is supervisory, because the water is gone and the panel is fine. A ground fault on an initiating device circuit is trouble, because the water is fine and the panel can no longer tell you about a fire on that circuit. Both leave you unprotected, in different ways, and they route to different people.

Priority runs alarm over supervisory over trouble, and it is a real ordering rather than a display convention: a panel in trouble that receives an alarm reports the alarm.

What actually generates a supervisory signal

Each is a physical device with a physical trip point, and the trip point tells you what the signal does and does not say.

• Control valve tamper switch. On an outside screw and yoke or butterfly valve, it trips when the stem moves off fully open, typically within the first two turns. It tells you the valve moved, not how far or whether any water still gets through, which is why a tamper signal can never be treated as a partial impairment.
• Low and high air pressure on a dry pipe or preaction system. The air holds the clapper closed. Losing air trips the valve and floods the pipe; too much air can prevent it tripping when it should.
• Low water level in a storage tank, and low tank or room temperature. A wet system in a space that goes below freezing is an impairment waiting for a cold night.
• Fire pump conditions. Pump running, phase reversal, loss of a phase, controller switched out of automatic. A controller left in manual after a test is one of the quietest impairments in the trade.
• Low pressure on a suppression system cylinder, whether a clean agent bank or a kitchen hood cartridge.

Nothing about the alarm wiring is on that list. Those are trouble signals, and the sibling card on recurring troubles owns them.

Which standard owns which impairment

Water based systems - sprinklers, standpipes, fire pumps, water storage - are inspected, tested and maintained under NFPA 25, in the edition your AHJ has adopted and amended. Its impairments chapter holds the coordinator role, the tagging requirement and the compensating measure rules. It binds the building owner and reaches a service contractor through the service agreement and the AHJ's acceptance of the reports.

Fire alarm systems are inspected, tested and maintained under NFPA 72 in the adopted edition, so an alarm system out of service is an NFPA 72 impairment, not an NFPA 25 one. Filing it under NFPA 25 puts the wrong procedure on the wrong system, and it is a common paperwork error.

The federal floor is separate, narrower, and runs to your employees rather than the building. For an automatic sprinkler system like the one in the record below, the federal cite is 29 CFR 1910.159; 29 CFR 1910.160 covers other fixed extinguishing systems and expressly excludes automatic sprinklers, requiring the employer to notify employees and take temporary precautions when such a system is out of service; 29 CFR 1910.164 covers fire detection systems and requires prompt restoration after each test or alarm.

An impairment is a state, not an event

Start time is when the system went off normal, not when somebody noticed. These are usually different, and the gap is diagnostic: a large one says the annunciator is not being watched.

The compensating measure is what stands in for the system while it is down. For a sprinkler impairment that is normally a fire watch: a person with no other duties, walking a defined route on a defined interval, with a means to raise the alarm, and a written log. Where the adopted edition of NFPA 25 applies, an impairment to a water based system exceeding 10 hours within any 24 hour period in an occupied building requires either evacuation of the affected portion or an approved fire watch. Below that threshold the clause does not compel a watch, and the owner, the insurance carrier and the AHJ frequently do, so the operative question is who is requiring it here.

End time is when the system is verified restored, not when the valve was turned. An outside screw and yoke valve opened most of the way looks open from across the room, and the tamper switch resets long before it is fully open. What proves the supply is back is a main drain test.

Preplanned versus emergency is decided entirely by the record. Closing a riser valve to cut in a branch line is a preplanned impairment when the coordinator ran the notifications and set the watch first, and an emergency impairment when nobody did. The pipe does not know the difference; everyone downstream of the paperwork does.

The record, filled in

A tenant fit-out on floor 3 of an occupied four story building, wet pipe sprinkler system, one riser per floor.

The main drain test, with its arithmetic. Static pressure at the floor 3 riser gauge read 78 psi. With the 2 in main drain fully open, residual read 62 psi. The last annual report recorded static 76 psi and residual 61 psi on the same riser.

Under NFPA 25 in the adopted edition, a reduction of about 10 percent in the full flow drain pressure compared with the previous test requires investigation. Ten percent of the prior 61 psi residual is 6.1 psi, so anything below about 54.9 psi this year would have sent somebody looking for a partly closed valve or an obstruction. The reading was 62 psi, one above the prior year, so the supply is where it was and the valve is genuinely open.

The gauge error, with its basis and its character, before it enters the comparison. These are 300 psi dial gauges with accuracy specified as a percentage of full scale, not of reading. Take a 3 percent of full scale class: that bound is 9 psi anywhere on the dial, which at a 62 psi reading is 14.5 percent of the reading. Its character is a fixed systematic offset for a given gauge, so:

• Within today's test, static and residual are read on the same gauge, so the offset cancels in the difference and the 16 psi drop carries only the small percentage error on the difference itself.
• Across years, if the gauge was replaced between tests, two independent worst case bounds apply and worst case bounds add linearly. The comparison then carries an error not greater than 18 psi, written with one inequality sign and never as a plus or minus interval. That bound is nearly three times the 6.1 psi investigation trigger, so the 10 percent year over year comparison is only meaningful when the gauge is the same instrument or has been calibrated. Record the gauge's identity and last calibration on the report, or the comparison cannot carry the weight you are putting on it.

Hazards the work created, each with its own action. Operating the main drain discharges water at supply pressure: route the discharge to the designated drain or exterior point, confirm nobody is standing at the outlet before you crack the valve, and open it slowly, because a fast open produces a surge at the outlet and a hammer in the riser. Silencing and resetting at the fire alarm control unit put hands near the 120 V primary supply, so that work ran with the branch circuit opened, locked and tagged under 29 CFR 1910.333(b)(2), absence of voltage verified on a meter proved live, dead, live on a known source per NFPA 70E-2021, 120.5, and the sealed battery set treated as still energized regardless of the panel's AC state.

Sibling rule check, printed. Against the notification appliance card: no appliance operated on this supervisory point, which is correct behaviour and not a defect. Against the duct detector card: the building's rooftop duct detectors annunciate as supervisory at the same panel, and the same tag and record process applies when one is pulled for service. Against the recurring trouble card: a ground fault two weeks earlier was a trouble, recorded separately, and created no sprinkler impairment, so the two do not belong on one record.

How to verify you got this right

Pull the last twelve months of supervisory signals from the panel history and set them beside the impairment records. Every supervisory should have a record, every record should carry both timestamps and a stated compensating measure, and every restoration on a water based system should carry a drain test result with the gauge identified. Signals with no matching record are the finding, and in most buildings they are the majority.

References

• NFPA 72, National Fire Alarm and Signaling Code, in the edition your authority having jurisdiction has adopted and amended, which defines the alarm, supervisory and trouble classes and owns fire alarm system impairment
• NFPA 25, Standard for the Inspection, Testing, and Maintenance of Water-Based Fire Protection Systems, in the adopted edition, whose impairments chapter owns the coordinator role, the tag, the compensating measure and the main drain comparison
• 29 CFR 1910.159 (automatic sprinkler systems), 29 CFR 1910.160 (other fixed extinguishing systems, automatic sprinklers expressly excluded) and 29 CFR 1910.164 (fire detection systems), the federal duties running to your employees when a system is out of service
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5, for the electrical work inside a fire alarm control unit
• See related: What a Duct Detector Is For and What It Does Not Cover; How to Work Out Why a Panel Keeps Going Into Trouble