The Detector Types and What Each One Responds To

Why this matters

No detector detects fire. Each detects a single physical signature that fires sometimes produce, and it detects that signature whether a fire produced it or a forklift did. Selection is a matching problem with two sides: the fire signature you expect from what is actually stored and done in that space, and the nuisance signature the space generates on an ordinary Tuesday. Get the first right and the second wrong and the device gets disabled by the people who work there, which is a slower and more permanent failure than picking the wrong sensor.

The third variable is transport. Both signatures reach the sensor through moving air, so where the device sits in the airflow is not an installation detail on top of the type. It is part of the type.

One signature each

What each one is blind to

The more useful list, and the one that gets skipped.

• Photoelectric is weakest on the very small particles from clean, hot, flaming combustion. A fuel that burns with little visible smoke is the case it is slowest on.
• Ionization is weakest on the large particles from smouldering. A cigarette in upholstery, an overheated cable jacket, a slow-heating bearing: the fires it responds to late.
• Heat detection of both kinds is blind to smoke entirely, responding only when enough energy has arrived at the ceiling, which by definition means a larger fire. Not a defect: it is why heat detection is chosen where smoke detection is not viable. The failure card for smoke versus heat covers what that trade costs.
• Rate of rise is blind to a fire growing slowly enough to stay under its rate threshold, which is why it is almost always paired with a fixed-temperature element in the same device.
• Projected beam is blind to anything that does not cross its path, including smoke that stratifies below the beam.
• Aspirating systems are blind to whatever their sampling points do not sample. Sensitivity is high; coverage is exactly the pipe network.
• Radiant energy detectors are blind to smouldering, because a fire with no flame emits no flame radiation, and blind to anything out of line of sight.
• Carbon monoxide detection is blind to a fire producing little CO, and responds to any CO source in the space, including a vehicle or an unvented appliance.

The pattern worth carrying: each is blind to something another on the list is good at, which is the argument for multi-criteria devices and for mixing types rather than standardising on one.

Nuisance is a signature too

A device's nuisance profile is as much a selection criterion as its sensitivity, because a device that alarms on ordinary operations gets bypassed. Common sources:

• Dust, from woodworking, milling, grain handling and forklift traffic, scatters light and reads as particles.
• Steam and high humidity condense in a chamber and read as particles.
• Cooking aerosols read strongly on ionization devices, which makes placement near kitchens a recurring conflict.
• Insects are a real and underrated cause; screening is part of the device rather than an accessory.
• Arc welding, sunlight, hot process equipment and some lighting emit in the bands radiant energy detectors watch.
• Vehicle exhaust produces both particles and carbon monoxide.

Where the nuisance signature and the expected fire signature are the same physical quantity, no sensitivity adjustment fixes it. The answer is a different type, or a multi-criteria device requiring two signatures to agree.

Position is part of the type

The commonly adopted editions of NFPA 72 set the placement rules, and it binds only in the edition the authority having jurisdiction has adopted and amended; the AHJ is a named role with legal authority over that adoption, not a synonym for the inspector.

Three placement facts do most of the work:

Distance from a supply. Spot detectors are kept a minimum distance from an air supply outlet, commonly 3 feet in the adopted editions. A detector in a diffuser's discharge sits in conditioned, already-filtered air and never samples the room it protects. This is the most common way a correctly selected detector is functionally the wrong device.

Spacing on a smooth ceiling. Spot smoke detectors carry a nominal spacing, commonly 30 feet on a smooth flat ceiling, applied so no point on the ceiling is more than 0.7 times the spacing from a detector. That coefficient is not arbitrary: on a square grid at spacing S, the farthest point from any detector is the grid diagonal midpoint at S divided by the square root of 2, which is 0.707S. The rule and the square grid are the same statement. It holds at constant ceiling geometry, smooth and flat; beam pockets, joists, slopes and high ceilings each change it by their own rules.

Air movement. Where the air change rate is high, the adopted edition reduces the spacing, because moving air dilutes the signature and carries it past the device. High ceilings bring the opposite problem, stratification, where a rising plume cools to the temperature of the layer beneath a hot roof and stops rising before it reaches ceiling-mounted devices.

Worked selection: a 90 by 60 warehouse bay

A single bay, 90 feet by 60 feet, 5,400 square feet, smooth flat roof deck at 32 feet, propane forklift traffic, dusty slab, palletised mixed commodity. The question is whether spot smoke detection is right here.

Spot layout at nominal spacing. At a 30 foot nominal spacing on a smooth ceiling, the grid is 3 columns by 2 rows, detectors 15 feet from each wall and 30 feet apart: 6 devices. Check against the coverage rule above: a room corner is 15 feet in each direction from the nearest device, a straight-line distance of 21.2 feet, against a limit of 0.7 x 30 = 21.0 feet. The full-spacing square grid is the limiting case by construction and lands marginally over, so the real layout pulls the perimeter devices in.

Applying the air movement reduction. This bay has high air change from unit heaters and dock doors. Take, illustratively, a reduction to two thirds of nominal spacing, with the actual reduction routed to the adopted edition's air movement table rather than assumed. Device count scales with the inverse square of spacing at constant room shape and a smooth flat ceiling, so two thirds of the spacing predicts (3/2) squared = 2.25 times the devices. The layout resolves to a 5 by 3 grid, 15 devices, 2.5 times the original 6. The realised multiple exceeds the predicted 2.25 because counts round up to whole devices and the perimeter takes a half spacing, a gap that grows as rooms get smaller relative to the spacing.

The diffuser conflict. The unit heater discharges and make-up air diffusers sit on roughly a 20 foot grid. With detectors now on an 18 to 20 foot grid, several land inside the 3 foot minimum from a supply outlet stated above, so those positions shift. Every shift is checked back against the 0.7 times spacing rule rather than moved by eye, since a device relocated for one rule commonly breaks the other.

The signature match, which overturns the layout. Three things are true of this bay at once. The nuisance signature is dust and propane exhaust, which a scattering chamber reads as particles. The ceiling is 32 feet, high enough that a plume can stratify and stall before reaching a roof-deck device. And the fire signature from a palletised mixed commodity begins as smouldering, the case radiant energy detection is blind to.

So the arithmetic was worth doing for what it ruled out. Fifteen spot devices at 32 feet, in dust, above a stratification layer, is a high device count buying low reliability and a high nuisance rate. The realistic candidates become projected beam below the expected stratification height, or an aspirating system with sampling points placed to suit, and the choice between them, with the actual spacing reduction and any listing-specific criteria, belongs to the designer and the AHJ rather than to a field decision.

Sibling-rule check, run against this selection before it ships:

• The smoke versus heat failure card's direction: smoke detection responds earlier than heat detection to a smouldering fire, and heat detection requires a larger fire. Used unchanged, and it is why heat detection is not offered as the cheap substitute in a bay where early warning is the objective.

• The panel card's rule that supervision is what makes a device meaningful: applied. Whichever type is selected, its circuit is supervised by the panel, and an aspirating unit adds airflow supervision that reports as a trouble.

• The obstruction card's principle that geometry is part of a listing: applied to placement, where the 0.7 coefficient is stated with the smooth flat ceiling it was derived under.

• The impairment card's rule that removing protection is a state: applied to the testing note below.

• Multi-criteria devices are blind to whatever every one of their constituent sensors is blind to, and are additionally slowed wherever the algorithm requires two signatures to agree. Combining sensors narrows the nuisance response; it does not add a signature none of them measures.

• Count check, last mention first: nine detector types in the table and nine in the blindness list; three placement facts announced and three given; three conditions named as overturning the spot layout and three enumerated.

• No rounding runs in the flattering direction: 21.2 feet against a 21.0 foot limit is reported as over, not as effectively equal.

Hazard control for the work this example requires. Installation and testing at a 32 foot deck is done from a properly set aerial lift, not from racking or a stacked platform; fall protection selection belongs to the ladder and lift cards. Before testing any device, notify the supervising station and open an impairment, because a zone placed in test is protection removed. Use only the manufacturer's listed test method: canned test aerosols carry a propellant and an aerosol you breathe at head height on a lift, so read the safety data sheet, ventilate during use, and follow the respiratory protection the sheet calls for under a program meeting 29 CFR 1910.134 if it calls for any. Never test a detector with smoke from an open flame or smouldering material, which introduces an ignition source into the space the detector protects and can contaminate the chamber permanently.

How to verify you got this right

Stand under each device and find the nearest supply outlet before anything else. That check takes seconds and catches the most common functional failure above.

Confirm the type against what the space does now, not what the drawings said when it was built. Tenancy changes signature: a unit that became a woodshop or a charging room has changed both sides of the matching problem.

For beam and aspirating systems, verify the geometry rather than the device. A beam whose path now runs through new racking, or a sampling pipe with a capped port, is a correctly functioning device covering nothing.

Read the panel history for repeat alarms on one address. A device that has alarmed three times with no fire is telling you the nuisance signature won, and the next step is a type change or a multi-criteria device, not a sensitivity adjustment reversed the following year.

References

• NFPA 72, National Fire Alarm and Signaling Code, in the edition the authority having jurisdiction has adopted and amended, for detector selection, spacing, the coverage rule, minimum distance from air supplies and air movement reductions.
• 29 CFR 1910.134 for respiratory protection where a test aerosol's safety data sheet calls for it, plus the sheet itself for the specific product.
• Manufacturer listing documentation for the detector, which states sensitivity, the listed test method and any spacing limits particular to that model.
• See related: the smoke versus heat failure card and the fire alarm panel card in this category.