Smoke detector

A Smoke detector is a device that responds to smoke particles or aerosol produced by combustion and thermal decomposition. It can provide an earlier indication than a heat detector because smoke can be produced before the fire's heat output becomes large. Its response depends on particle properties, transport, entry into the detector and the detection algorithm.

Smoke detectors use several principles, including ionisation, light scattering, optical obscuration, projected beams, air sampling and multi-sensor detection. No principle responds identically to every smoke type. Large smouldering droplets and small flaming-combustion particles can produce very different responses.

The device does not measure fire size or safety directly. It indicates that smoke with particular properties has reached and entered the detector in sufficient quantity or form to meet a response condition. That condition must be evaluated within the fire scenario and the detector's actual installation.

Detection principles

Ionisation detectors respond when smoke particles intercept gaseous ions and reduce the detector's ion current. They tend to be sensitive to high concentrations of small particles, such as those from flaming paper and wood fires. Their response is less sensitive to low concentrations of large smouldering droplets in the cited comparison.

Photoelectric or light-scattering detectors respond to light scattered by smoke. Their sensitivity depends on particle size distribution, refractive index, wavelength, scattering angle and scattering volume. They can be more sensitive to large smouldering droplets and less sensitive to low-smoking flaming fires in the cited comparison.

Projected-beam detectors measure attenuation along a light path. Total optical density is determined across that path, so beam length and smoke distribution matter. This approach can cover a large volume without requiring smoke to enter a small sensing chamber, although the beam and its wavelength must match the assessment.

Aspirating or air-sampling detectors draw air through a sampling system for analysis. This can support very high sensitivity and different placement strategies, but sampling time, filter condition and system design affect response. Their suitability should be established using product-specific testing rather than assuming faster activation in every scenario.

Multi-sensor detector designs use more than one sensing principle or input to improve fire-signature recognition. They can help distinguish fire from nuisance conditions, but their logic and thresholds are manufacturer-specific. Combining technologies does not eliminate unwanted alarms or the need for correct placement and testing.

Optical density and light response

Optical density expresses light attenuation per unit distance and is related to particle concentration and extinction properties. A projected-beam detector evaluates it along the beam, with the total response depending on path length. Smoke that is not uniformly distributed requires segment-by-segment evaluation rather than one room-average value.

Wavelength is essential when comparing optical measurements with a detector. Data measured at another wavelength can introduce errors because scattering and absorption depend on it. The light source used in the test or model must match the detector being evaluated.

A scattering-type detector can respond at different optical densities for different smoke types. Schifiliti and colleagues give an example of a detector responding at 0.029 per metre to grey cotton-wick smoke but at 0.15 per metre for kerosene smoke. At those thresholds, both smoke types scatter the same amount of light to the receiver but differ in colour and reflection.

UL acceptance testing described by Mulholland uses optical density per metre of 0.06 for grey cellulosic smoke and 0.14 for black kerosene smoke. These source values are test criteria, not universal alarm thresholds for every detector or fire. Detector-specific calibration and smoke signature remain necessary.

Particle size and fire mode

Particle size distribution influences detector response. Ionisation detectors in the cited comparison are more sensitive to particles smaller than about 0.3 micrometres, while light-scattering detectors are more sensitive to larger particles. These tendencies help explain why detector choice should reflect the expected fire mode.

Smouldering cellulosic insulation produced a mass mean particle size of 2 to 3 micrometres in Ohlemiller's study, much larger than cigarette smoke and the sooty particles from flaming combustion. This contributed to poor ionisation-detector sensitivity and to filtration of aerosol by the fuel bed. A detector suitable for one fire mode can therefore be slow or insensitive to another.

Flaming fires can produce small carbonaceous particles and strong light emission, while pyrolysis and smouldering often produce larger droplets and light-coloured smoke. Those physical differences affect both scattering and ionisation. A detector threshold cannot be inferred from fuel name alone without knowing the combustion mode.

Smoke properties also change with transport. Particles can coagulate, evaporate, deposit or form through condensation, altering concentration and size distribution. A detector response measured near the source cannot simply be applied to an older or remote smoke cloud.

Transport and ceiling jet

A Ceiling jet carries hot gases and smoke away from the plume and across the ceiling. Detector response depends on when that flow reaches the device and how much smoke it contains. Location, spacing, fire size and ceiling geometry therefore affect activation time.

Smoke entry resistance adds another delay. Bug screens, chamber design and aerodynamic characteristics can make optical density inside the chamber lower than outside for a period. Schifiliti and colleagues describe this with a detector time constant, analogous to a response lag.

Weak smouldering plumes can fail to reach a ceiling detector at all. Ohlemiller notes that very weak smouldering sources may not produce a plume reaching the ceiling, while the fuel bed can filter the aerosol. Ceiling placement alone therefore does not guarantee early detection of every smouldering fire.

Transport lag is not the only delay. Smoke must then enter and produce a sufficient signal, after which the alarm algorithm and notification sequence begin. These stages should be separated when evaluating the time to occupant response.

Response and activation time

Detector activation time includes smoke production, transport, entry and signal processing. It varies with fire growth, detector location, particle properties and the response algorithm. A single generic activation time cannot be assigned to a detector model across different fires.

Smoke-detector models are less certain than heat-detector models. Early smoke production and transport are highly variable, and manufacturers often do not publish their response algorithms. Estimation methods should therefore be used mainly to compare changes in spacing or location while other variables are held constant.

Product tests can provide useful response ranges. Standards may require activation at particular levels or times under specified fuels and buildup rates, but those conditions may not match the intended fire. Test limits should be used as evidence of that product's tested performance rather than as universal values.

C/VM2 provides different response modelling constants for smoke, projected-beam and air-sampling detectors in its design scenarios. Those parameters belong to that 2025 New Zealand method and should not be reused as universal detector constants. Actual system design still requires scenario and product evidence.

Unwanted fire alarms

Unwanted fire alarms can result from non-fire signatures, test activations, pranks or unsuitable placement. Complex algorithms are intended to reduce nuisance alarms and improve signature matching, but they are often proprietary. Exact prediction of false-alarm frequency is therefore difficult without product evidence.

Frequent false alarms can make occupants less responsive. Proulx describes reduced credibility of the alarm signal, with occupants delaying or failing to look for further information. Correct detector selection and management should address nuisance causes without making genuine detection less effective.

Smoke-detector testing and sensitivity settings require particular care. NRC guidance for incipient systems says reduced sensitivity should be evaluated to ensure that early detection is not compromised. Testing, commissioning and maintenance should preserve that function rather than treating fewer alarms as automatically better.

Unwanted activation is not always a device fault. Smoke-like conditions, contamination and inappropriate siting can create valid signals under the wrong interpretation. The assessment should distinguish those conditions from a genuine fire and from a detector malfunction.

Placement and system design

A Fire alarm system integrates detectors with control, alarm and notification. Detector choice should follow the required response time and the fire scenario, and the complete chain must be tested. An early detector signal alone does not guarantee an effective evacuation response.

Spot detectors require smoke to reach and enter the sensing chamber, while projected beams can observe a path through a larger volume. Aspirating systems can use high sensitivity where appropriate, but all types have transport and maintenance requirements. Selection should be based on evidence for the intended environment.

Ceiling construction and height affect spacing. Smooth ceilings favour distribution of heated air and smoke, while obstructions and high ceilings change transport and may require different placement or detector types. The fire's location relative to the detector also matters.

Multi-sensor detection and redundancy can improve resilience, but they can also share dependencies. Control, power and communication failures can affect several devices at once. Reliability analysis should include those common paths and the system's failure indication.

Aspirating smoke detection draws air through a sampling system for analysis and can support very high sensitivity. Sampling transport, filter condition and system design affect the response, so product-specific testing is required. Beam smoke detector arrangements instead measure attenuation across a light path, with total optical density depending on path length and smoke distribution. These are different measurement problems from spot-chamber detection.

For maintenance, the complete alarm chain should be checked. A sensing element can operate while a panel, sounder or communication path fails, and a blocked sampling or optical path can alter the response without an obvious electronic fault. Inspection should include those paths as well as the detector itself.

Documentation should identify the detector model, principle, threshold, location and test evidence. Response algorithms and thresholds should not be invented when manufacturers do not publish them. Where exact activation is critical, product-specific tests and scenario analysis are needed.

For final acceptance, the detector should be evaluated by the response it enables. Early indication is useful only if it produces intelligible notification and enough time for action. That makes the detector one part of the alarm and evacuation system rather than a guarantee of protection.

Testing should preserve sensitivity and correct operation. Product-specific acceptance tests, sensitivity checks and maintenance should follow the manufacturer and applicable standard. Changes to sensitivity should be evaluated for their effect on early detection rather than accepted solely as a way to reduce nuisance alarms.

Smoke-entry paths and sensing chambers require maintenance. Contamination, blocked screens and altered airflow can change the response, and an optical path can be obstructed without an obvious device fault. Inspection should include those physical paths as well as electronics.

Documentation should identify the detector model, principle, threshold, location and test evidence. Response algorithms and thresholds should not be invented when manufacturers do not publish them. Where exact activation is critical, product-specific tests and scenario analysis are needed.

Maintenance should preserve the complete alarm chain. A detector can be operational while the panel, sounder or communication path fails, so testing should not stop at the sensing element. The system must produce intelligible notification and an appropriate response.

Limits and common misunderstandings

A smoke detector does not detect every fire early. Weak smouldering sources may not deliver smoke to the detector, and some particles produce a weak response. Detector principle and placement must match the expected fire mode.

Higher sensitivity is not automatically better. It can increase unwanted alarms and undermine credibility, while reducing sensitivity can delay genuine detection. Settings should be justified using product tests and the required response time.

Optical density is not a universal alarm threshold. It varies with wavelength, particle properties and smoke colour, and detector response algorithms differ. Manufacturer-specific data and test conditions are required for meaningful use.

Finally, a detector signal is not an evacuation instruction. The alarm system must provide intelligible notification, and occupants must understand what to do. Detection is only the first stage of that response.

For final acceptance, detector choice should be tied to the expected fire mode and measurement problem. Spot chambers, beams and aspirating systems each have different transport and response characteristics, and each needs compatible evidence. Selecting by sensitivity alone would ignore those differences and could produce unreliable activation timing.

See also

• Aspirating smoke detection
• Smoke stratification
• Smouldering-to-flaming transition

References

• Australian Building Codes Board, National Research Council of Canada, International Code Council, & Department of Building and Housing New Zealand. (2005). International fire engineering guidelines (2005 ed.). Australian Building Codes Board.
• Ministry of Business, Innovation and Employment. (2025). C protection from fire: Verification method C/VM2 (2nd ed.). New Zealand Government.
• Mulholland, G. W. (2002). Smoke production and properties. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed.). National Fire Protection Association.
• Ohlemiller, T. J. (2002). Smoldering combustion. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed.). National Fire Protection Association.
• Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed.). National Fire Protection Association.
• Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed.). National Fire Protection Association.
• U.S. Nuclear Regulatory Commission. (2005). Fire dynamics tools: Quantitative fire hazard analysis methods for the U.S. nuclear regulatory commission fire protection inspection program (NUREG-1805). Author.
• U.S. Nuclear Regulatory Commission. (2010). Fire PRA methods enhancements (NUREG/CR-6850, Supplement 1). Author.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering. Fire Safety Academy.