3.1 Fire and Gas Detectors

Why Fire and Gas Detection Is a Measurement Problem

On the PE Control Systems exam, fire and gas (F&G) detection sits in Measurement spec 1.C because the wrong sensor is a physics error, not a wiring error. You must match wavelength, absorption, oxygen demand, and line-of-sight to the fuel and the plot plan, then defend coverage, voting, and false-alarm behavior. A missed jet fire, a spurious emergency shutdown (ESD), or a catalytic bead that reads “safe” because the atmosphere no longer contains oxygen all look like instrumentation questions.

Optical Flame Detection

A flame detector looks for electromagnetic signatures of combustion. It is not a thermocouple waiting for bulk gas to heat up.

Ultraviolet (UV) flame detectors respond to the solar-blind UV continuum of a flame, typically about 185–260 nm. They are fast (milliseconds) and they see hydrogen and other low-luminosity flames that barely radiate in the mid-infrared. The penalty is false-alarm sources: lightning, arc welding, industrial X-ray, and some UV lamps. Oil film, grease, ice, or salt on the lens obscures UV and can fail the detector blind while the logic still looks “healthy” if dirty-optics supervision is weak.

Infrared (IR) flame detectors look at hot-CO2 and hydrocarbon emission, classically near 4.3–4.4 μm, often with flicker discrimination so a steady hot flange is less likely to trip. Single-frequency IR is relatively inexpensive and works well on sooty hydrocarbon fires. Sunlight, hot process equipment, and modulated heat sources still cause spurious alarms when the algorithm is weak.

UV/IR detectors require coincident UV and IR before declaring fire. That AND logic cuts welding and lightning trips compared with UV-only, and cuts hot-object trips compared with IR-only. The exam trap is fuel mismatch: a hydrogen flame is UV-rich and IR-poor at the hydrocarbon CO2 band. A UV/IR unit whose IR channel is hydrocarbon-tuned can fail to confirm a hydrogen fire even though the UV channel is in alarm.

Triple infrared (IR3) uses three IR bands plus ratio and flicker logic. For outdoor hydrocarbon pool and rim-seal fires it is the usual high-immunity choice: strong rejection of sun, hot metal, and many welding arcs, and more tolerance of light rain or mist than a UV lens. IR3 is the wrong default for hydrogen, and it can be weak on other fuels that do not produce the expected CO2 or soot IR fingerprint.

Flame cameras and closed-circuit television (CCTV) analytics add spatial intelligence. A video or IR imager classifies flame-like pixels, can ignore a known welder location, and gives operators a picture for verification. Capital cost is high (imager, network, analytics host, lighting). Fog, steam plumes, and heavy rain reduce contrast. Cameras complement gas detection; they do not automatically retire point combustible heads.

Specify field of view, sensitivity against a defined pan fire (industry listings such as FM 3260-style distance ratings), material and ingress protection, and whether the output is 4–20 mA, relay contacts, or an addressable loop. Distance ratings collapse if the actual fuel, fire size, or obscuration differs from the listing fire.

Gas Detection Physics

Point infrared using nondispersive infrared (NDIR) measures absorption of an IR beam in a sample cell at a hydrocarbon or CO2 wavelength. It does not need oxygen, is not poisoned by silicones, and is more stable than a catalytic bead. It cannot see hydrogen, which has no useful IR absorption in those bands.

Open-path IR places a transmitter and receiver, or a retro-reflector, tens to hundreds of meters apart and reports a path-integrated concentration, typically in LEL·m (lower explosive limit times meters). A 10 m cloud at 50% LEL is 5 LEL·m. Open-path is excellent for tank-farm perimeters and long pipe racks. It is a line-of-sight instrument: misalignment, fog, steam, snow, birds, and dirty windows cause beam-block / obscuration. That condition should be a fault, not a gas alarm, if trip logic is specified correctly. Congested modules with vessels, piping, and scaffolding break the beam into lucky sightlines. Open-path does not replace a point-detector grid unless a coverage study shows the remaining volume is actually seen.

Laser open-path, typically tunable diode laser absorption spectroscopy (TDLAS), locks to a single absorption line (methane, hydrogen sulfide, ammonia, hydrogen fluoride). Selectivity and long-path sensitivity are excellent; capital cost and alignment skill are the highest in the gas family. You still have a beam to keep aligned, and you still cannot claim volume coverage you have not mapped.

Catalytic bead (pellistor) sensors oxidize combustible gas on a heated catalyst and infer concentration from the resulting resistance change. They are inexpensive and they do see hydrogen. They require oxygen (vendor-specific, often on the order of 10% O2 or more). In nitrogen-purged, inerted, or post-fire oxygen-depleted atmospheres they read low or zero—the classic PE trap. Poisoning by silicones, lead, phosphorus, and some sulfur compounds is permanent; halogenated hydrocarbons can inhibit reversibly. A “calibrated last month” bead in a silicone-caulked analyzer house is not a healthy detector.

Electrochemical cells are the workhorse for toxic gases and for oxygen deficiency (H2S, CO, Cl2, NH3, O2). They are consumable, temperature- and humidity-sensitive, and they have documented cross-sensitivities. They are not a combustible LEL detector unless the stem gives a specialized cell and a combustible-gas basis, which is unusual.

Metal-oxide semiconductor (MOS) sensors change resistance in the presence of reducing gases. They are cheap and broad-spectrum, which means they are also non-specific and they drift with humidity. They appear on light commercial or refrigerant applications more often than on safety-integrity-level hydrocarbon F&G.

Coverage, Voting, Environment, and Cost

Coverage mapping—the engineering practice behind detector-layout studies—asks two different questions. Geographic coverage is the fraction of the plot plan inside a detector’s credited view. Scenario coverage is whether a defined release or fire actually intersects a detector before escalation. Voting such as 2ooN (two out of N) reduces spurious ESD from a single dirty window or a passing truck. 1ooN is more sensitive and more nuisance-prone; toxic clouds are sometimes left 1ooN because a missed detection is worse than a trip. Do not vote two detectors that share the same obstruction or the same dirty-air path and then claim independence.

Cost, in round PE terms: MOS and catalytic point heads are cheapest; electrochemical is moderate; point IR is moderate-to-high; UV/IR and IR3 are high; open-path IR is high; laser open-path and camera systems are highest, plus towers, alignment kits, and the voting logic solver.

Worked Example: Hydrocarbon Tank Farm Versus Hydrogen Compressor Shed

A Gulf Coast tank farm stores gasoline and diesel in floating-roof tanks. The F&G engineer must detect pool and rim-seal fires outdoors among hot tank shells, sunlight, and occasional welding. IR3 is the rational primary flame choice: a hydrocarbon IR signature, high false-alarm immunity, and no dependence on a UV lens that dirties in marine air. Point IR or open-path IR handles combustible vapor on the bund perimeter. Catalytic beads are a poor primary there because of silicone-containing maintenance products and because oxygen is not the limiting issue—false alarms and poisoning are.

A hydrogen compressor shed is the opposite fuel. The flame is pale, UV-strong, and weak at 4.3 μm. IR3 can miss it. Specify UV or a hydrogen-rated UV/IR, plus a combustible sensor that can actually see H2: a catalytic bead if the shed is air-ventilated, thermal conductivity, or a hydrogen-specific analyzer—not hydrocarbon NDIR. If the compressor is nitrogen-purged, catalytic beads are invalid; you need a method that does not consume oxygen. Open-path hydrocarbon IR across the door is not a hydrogen detector.

Exam Traps

• Catalytic bead in an oxygen-deficient or inerted atmosphere.
• Using hydrocarbon IR3 or point IR as if they were universal fire or gas sensors for hydrogen.
• Declaring that one open-path beam retires all point detectors in a congested module without coverage engineering.
• Treating fog or rain beam-block as a high-high gas trip.
• Ignoring line-of-sight alignment and voting when the stem is about spurious trips.