What a Preaction System Is Protecting Against

Why this matters

A preaction system is the only water-based system in the building that is not primarily protecting the contents from fire. It is protecting them from the sprinkler system. Somebody looked at a room full of records, servers, artwork or frozen product, decided an accidental discharge would be a loss event in its own right, and bought an arrangement requiring two independent things to happen before water can enter the pipe. Often the correct purchase. What gets missed is the invoice: the detection system is now part of the suppression system, so a disabled detection zone is a total suppression outage, and nobody files it as one because nothing was done to the sprinkler side.

The three arrangements, and what each one requires

All three keep the piping dry under supervisory air or nitrogen, with a preaction valve holding water back at the riser, and all three use closed heads with thermal elements exactly like any other sprinkler system. What differs is what has to happen before the valve opens.

On every one of the three, a head must still open before water leaves the pipe. That is the sentence most people get wrong. On a single interlock system, detection filling the pipe does not discharge water; the interlock is a decision about whether to fill the pipe, not about whether to wet the room. Preaction does not reduce the water delivered during a real fire and does not delay it once both conditions are met and the pipe is charged. It only adds conditions before water can be in the pipe at all.

What it is actually protecting against

Three specific events, worth naming because they justify the cost. Mechanical damage to the piping, from a forklift mast, a scissor lift or a pallet swung wide: on a wet system that is an immediate discharge at supply pressure into whatever is below, and on a double interlock system it is a low-air supervisory signal and nothing else, because losing air alone does not open the valve. A head knocked off or damaged, same event and result. A fitting or head failing on its own, which is rare, and rare is not never, and the reason a room is on preaction is that the consequence is disproportionate.

All three are failures of the sprinkler system rather than fires. Preaction is a system bought to defend against its own failure modes.

The price, and it is paid in availability

Requiring two independent events before water flows is the same as requiring two systems to both be working. Reliability arranged that way multiplies rather than adds, and it moves in the unhelpful direction.

Take an illustrative 2 percent unavailability on each side, a figure chosen to show the arithmetic rather than measured from anything, and assume the two are independent. Both-required availability is 0.98 times 0.98, which is 0.9604, so unavailability rises from 2 percent to about 3.96 percent, close to double. Run the same two numbers through an either-suffices arrangement and unavailability is 0.02 times 0.02, or about 0.04 percent. Same components, same figures, two orders of magnitude between the outcomes, purely because of how they combine.

Two qualifiers ride with that arithmetic and neither is optional. The independence assumption is the weak part: a construction project that dusts out the detectors and gets a zone disabled, a contractor who shuts things down for a weekend, an owner who defers both inspections in the same budget cycle are all common-cause events that take both sides out together, and the multiplication does not protect you at all when they happen. The numbers are illustrative, and real availability comes from the inspection, testing and maintenance record for the specific systems.

There is a second, smaller price in the design. The adopted edition of NFPA 13 applies the same area of operation increase to double interlock preaction systems that it applies to dry pipe systems, commonly 30 percent at the same density, for the same reason: water takes time to arrive after the valve opens. The sibling card on wet versus dry systems owns that rule and its arithmetic.

The case: one head open, no water, no fire alarm history

The signal. A hospital's offsite records storage room, roughly 2,200 sq ft, protected by a double interlock preaction system with spot smoke detection. A fire started in a shredder waste bin against the back wall and one sprinkler head above the bin operated. No water came out of it. Staff put the fire out with a portable extinguisher from the corridor, leaving a single open head, a dry pipe, and a lot of questions.

First candidate: the water supply. Eliminated in ten minutes. The city supply and the riser control valve upstream of the preaction valve were open, supervised and reading normal pressure. The valve had not been asked to open.

Second candidate: the preaction valve mechanically stuck. Eliminated on the record rather than by testing. The valve had been trip tested at its last scheduled interval under NFPA 25, in the edition the authority having jurisdiction had adopted and amended, with a successful trip recorded and nothing done to it since.

Third candidate: the head never opened. Eliminated by looking at it. The element was released and the deflector clear.

What was true. The detection zone serving that room had been disabled at the fire alarm control unit three weeks earlier. A corridor renovation on the other side of the wall was generating dust, the zone had nuisance-alarmed twice in one night, and someone had disabled it to stop the horns. The disable was logged at the panel as a trouble condition and acknowledged every shift thereafter.

On a double interlock system both conditions are required. The head opening satisfied the air-loss condition. The detection condition was never going to be satisfied, because the zone was off, so the valve stayed shut and the pipe stayed dry.

What generalises. A single interlock system in the same room would have behaved identically, because on single interlock the valve opens only on detection and detection was disabled. Only a non-interlock arrangement would have flowed water. So the finding is not that double interlock was the wrong choice. It is that on every preaction arrangement except non-interlock, the detection system is a component of the suppression system, and a disabled detection zone is a suppression outage.

Why nobody caught it. The disable was filed, correctly, as a fire alarm impairment under NFPA 72, and the sibling card on detection and suppression owns that split. Both files were accurate. Neither said the words "the sprinkler system in the records room cannot deliver water." The compensating measure attached to the alarm impairment was hourly rounds for occupant notification, which is right for a notification outage and wrong for a suppression outage.

Corrections, printed.

• Raw reading: the sprinkler system failed. Correction applied: the sprinkler side performed exactly as arranged, and the arrangement requires two conditions. Corrected reading: one required condition was unavailable, so the system was unavailable.
• Raw impairment scope: an alarm zone was disabled. Correction applied: on a preaction system the detection zone is a component of the water-based system, so the impairment is re-based to cover both systems rather than filed twice against one. Corrected scope: alarm notification for the zone plus suppression for the entire preaction area.
• Raw compensating measure: hourly rounds. Correction applied: the measure has to match the outage, and a suppression outage needs a fire watch with extinguishing capability in the area and a stop on hot work and ignition sources there. Corrected measure: continuous fire watch for the preaction area.
• Raw availability assumption: two systems, so twice as safe. Correction applied: two systems in series multiply, so unavailability roughly doubled on the illustrative 2 percent figures, to about 3.96 percent. Corrected reading: the arrangement deliberately trades availability for protection against inadvertent discharge.

The fix. The nuisance source was addressed rather than the detector: the corridor work got dust containment and the zone was returned to service, the head was replaced under a tagged impairment on the preaction system, the pipe restored to supervisory air pressure, and the valve reset and confirmed in the set position with both gauges read and recorded. The impairment procedure was rewritten so that disabling any detection zone that releases a suppression system generates a suppression impairment as well, with the fire watch as the standing compensating measure.

Sibling-rule check on this case. The head was treated as a thermal device responding to gas temperature at its element, consistent with the head card. The replacement matched type, K-factor, temperature band, response class and trim together, consistent with the head types card. The 30 percent area increase was not applied anywhere here because no demand was recalculated, and the wet versus dry card owns it where it applies. Alarm impairment was filed under NFPA 72 and water-based impairment under NFPA 25, consistent with the detection card. No valve was left closed and no zone left disabled at close-out.

When preaction is the wrong answer

• When the objection is water damage in general. Ordinary offices, retail and light manufacturing get wet systems. Few heads operate in a real fire, and the sprinkler discharge is a fraction of what a fire service hose stream puts into the same space.
• When the room already has a gaseous agent system. A clean-agent or carbon dioxide system plus preaction sprinklers is defensible and it doubles the coordination burden. Before anyone tests detectors in that room, lock out the agent release circuit and clear and post the space, because carbon dioxide discharge can drive a room below a life-supporting oxygen concentration within seconds.
• When nobody will maintain the air supply. A chronically low-air preaction system is trending toward an unintended trip, and its supervisory signals have already been trained out of the staff.
• When the real risk is a freezer and only a few heads are involved. Dry pendent heads off a heated wet system are simpler and have no interlock to fail, an option the wet versus dry card owns.

How to verify you got this right

1. Read the valve trim, not the label on the door. Establish from the valve arrangement and the system record whether the system is non-interlock, single interlock or double interlock. The three behave differently in exactly the situation you are being asked about.
2. Trace which detection zone releases the valve and check its status at the panel. If that zone is disabled, the suppression system is unavailable right now, whatever the sprinkler tags say.
3. Read the compensating measure on any open impairment and ask what outage it matches. A measure written for notification does not cover suppression.
4. Never enter standing water in a room with energized equipment. After a discharge, the supply to equipment in the affected area is opened and verified dead at the point of work before anyone walks in, because water plus energized equipment is a shock path and a records room floor is not where you find out.

References

• NFPA 13, in the edition adopted and amended by the authority having jurisdiction, for preaction system types and the area of operation increase applied to double interlock systems
• NFPA 25, in the adopted edition, for preaction valve trip testing, air pressure supervision and the water-based impairment program
• NFPA 72, in the adopted edition, for the releasing service detection that forms one half of the interlock and for alarm impairment
• See related: What a Wet Pipe and a Dry Pipe System Each Solve; Why Detection and Suppression Are Different Problems