Smoke detector

A smoke detector is a device that responds to particles produced by combustion and supplies a signal when its alarm criterion is reached. Common sensing principles use changes in ion current, scattered light or transmitted light. The response depends on the particles reaching the sensing region and interacting with the particular sensor.

Smoke varies with the fuel, burning mode and conditions during transport. Two smokes with the same measured optical density can produce different responses in a scattering or ionisation detector. A calibration threshold therefore needs to be interpreted with the smoke and test conditions used to establish it.

Detection time also includes transport and, for enclosed sensing chambers, smoke-entry delay. Air sampling and projected-beam arrangements introduce different measurement paths. Engineering assessment combines those physical processes with the detector's response criterion and any later alarm processing, while recognising the limits of available input data.

How smoke becomes a detection signal

Smoke contains particles whose properties depend on the fuel and the way it burns. Detectors can respond through changes in electrical current or changes in light reaching a receiver. The SFPE Handbook distinguishes the response of the sensor from the transport of smoke to it. Both processes affect the time at which an alarm signal becomes available.

An ionisation detector uses an ion current within its sensing chamber. Smoke particles intercept ions and reduce the current until the alarm criterion is reached, while a light-scattering detector instead measures light redirected by particles within the chamber. The two mechanisms respond differently to particle size, concentration and other smoke characteristics.

A Beam smoke detector measures attenuation of a light beam across a space. Attenuation is the reduction of transmitted light caused by the smoke along its path. Aspirating smoke detection uses a sampling arrangement that draws air towards a sensing chamber. These arrangements introduce different measurement geometries and, for sampling systems, additional transport through the sampling network.

A detector supplies an input to the Fire alarm system, which processes the signal and produces its configured outputs. Detection and occupant notification are separate functions. The presence of smoke at the sensing location does not establish that the complete warning sequence is already finished. Processing and the intended human response remain part of the system assessment.

Smoke-detection principles measure different particle responses. Based on Mulholland (2002), pp. 2-266–2-267 and Schifiliti, Meacham, and Custer (2002), pp. 4-21–4-24 and Iqbal and Salley (2004), p. F-52.

Particle size, fuel and burning mode

Smoke is not a uniform substance with one universal detector response. Schifiliti and colleagues identify particle size distribution, concentration, composition, colour and refractive index as relevant characteristics. Refractive index describes how the material interacts with light. Fuel composition, flaming or smouldering combustion and the supply of combustion air affect the particles produced.

Mulholland compares ionisation and scattering responses for particular detectors. In that comparison, ionisation response is stronger for particles smaller than about 0.3 µm. Light-scattering response is stronger for particles larger than about 0.3 µm. This is an approximate comparison from the reported detector response functions, not a sharp dividing line for every device.

The chapter associates relatively small particles with examples of flaming paper and wood fires. Larger droplets from smouldering combustion can favour the response of scattering detectors. These relationships explain why the same nominal optical obscuration does not imply the same electrical signal in every sensor. The burning mode and the detector principle must be considered together.

Particle properties also change during transport. Particles can settle or combine, changing the number and size distribution that reaches the detector. The SFPE detection chapter therefore distinguishes smoke characteristics at the fire from those at the sensing location. Using generation data alone can omit changes relevant to the eventual response.

Detector sensitivity varies with particle size. Based on Mulholland (2002), pp. 2-266–2-267.

Getting smoke into the sensing region

A rising plume can carry smoke towards the ceiling, where a Ceiling jet distributes it away from the fire. Ceiling height, geometry and intervening barriers influence this route. Natural and mechanical air currents also affect the movement of smoke, particularly during early, low-energy fires. The transport time cannot be assumed identical for different detector locations.

Smoke reaching the outside of a spot detector must still enter its sensing chamber. Screens, chamber geometry and the flow around the housing can resist entry. The SFPE Handbook calls this entry resistance. During an increasing external smoke concentration, the concentration within the chamber can lag behind the surrounding smoke.

This lag is conceptually separate from the detector's response threshold. A device may be sensitive to a particular smoke once it is inside the chamber but still receive that smoke slowly. The local gas velocity and the device's aerodynamic characteristics therefore matter. Treating external smoke concentration as instantaneously identical to chamber concentration can understate the response time.

A weak fire beneath a high ceiling may also produce smoke that loses buoyancy before entering the assumed ceiling flow. The handbook discusses this possibility when examining low-energy fires and stratification. Stratification means that smoke can collect at a level determined by the surrounding temperature conditions. A detector near the ceiling may therefore receive a different exposure from the one assumed in a simple spreading-layer calculation.

Air sampling introduces a different transport arrangement. C/VM2 identifies airflow, sampling-tube length and alarm threshold as factors in the response of very high sensitivity sampling systems. The method treats these systems as requiring specialised design. A sensitive measuring chamber alone does not determine the time needed for a sample to arrive and initiate the intended output.

Calibration does not define every fire response

Optical density describes how smoke attenuates a light beam over a specified distance, and it is useful for measuring smoke and for modelling projected-beam detectors. A scattering detector, however, measures redirected light rather than the same beam attenuation quantity. Its alarm threshold expressed as optical density therefore depends on the calibration smoke and the detector characteristics.

The SFPE Handbook gives an example of a scattering detector responding at 0.029 m⁻¹ to smoke from a smouldering grey cotton lamp wick. The same example describes a response at 0.15 m⁻¹ for kerosene-fire smoke. At those response points, the receiver detects the same scattered-light signal. The darker kerosene smoke scatters light differently from the lighter wick smoke.

These values are examples for the particular comparison, not universal detector sensitivities, and they show why a calibration value cannot be transferred indiscriminately between smoke types. Wavelength and scattering angle also affect the response. Appropriate input data need to represent both the smoke and the sensing mechanism being modelled.

Optical-density response depends on smoke type. Based on Schifiliti, Meacham, and Custer (2002), p. 4-23.

The handbook also discusses limits imposed by manufacturer algorithms: a product may process sensor readings over time rather than operate at one simple instantaneous threshold. Those algorithms can affect nuisance rejection and genuine-fire response. Without suitable product information, a concentration calculation may describe the exposure but not the complete activation decision.

A beam calculation with a worked example

For uniform smoke along a beam, optical density per unit length is calculated from the fraction of light transmitted. The calculation uses the base-ten logarithm of the intensity ratio and divides by beam length. Greater attenuation gives a larger optical density. C/VM2 provides this relationship when describing projected-beam detector response.

$$D_u=-\frac{1}{L}{\log }_{10}\left(\frac{I}{I_0}\right),\frac{I}{I_0}=\frac{100-O}{100}$$

Symbol	Description	Unit
$D_u$	Optical density per unit beam length	m−1
$L$	Beam path length	m
$I$	Intensity at the receiver with smoke present	Same intensity unit as $I_0$
$I_0$	Intensity at the receiver without smoke	Same intensity unit as $I$
$O$	Total obscuration expressed as a percentage	%

The worked example in New Zealand's C/VM2 (2025) uses a 10 m beam and an alarm setting of 50% total obscuration. Half of the original light intensity remains at the receiver. Substituting the source values gives approximately 0.030 m⁻¹. This is the uniform optical density corresponding to the stated path and setting.

$$D_u=-\frac{1}{10}{\log }_{10}\left(\frac{100-50}{100}\right)=0.0301{\mathrm{m}}^{-1}\approx 0.030{\mathrm{m}}^{-1}$$

The calculation does not predict when the smoke will reach that value; a fire and smoke-transport analysis is still needed to estimate activation time. Nor is the result a universal point-detector threshold. It describes attenuation along the specified beam, using the detector's total-obscuration setting.

Smoke may occupy only part of the path, as when a plume crosses the beam. C/VM2 describes dividing a non-uniform path into segments and summing each segment's optical-density contribution. Schifiliti and colleagues describe the same distinction between uniform and segmented representations. Assuming a uniform layer when the actual exposure is local can misrepresent the beam measurement.

Prescribed inputs in the New Zealand method

C/VM2, second edition (2025), specifies detector inputs for its verification calculations, and its smoke-detector representation includes ionisation, photoelectric spot and air sampling arrangements. The specified optical density at alarm is 0.097 m⁻¹. The table also gives a radial distance of 7.0 m and a position at least 25 mm below the ceiling.

The document requires these analysis values regardless of the actual detector make or model, so they represent the method's prescribed simulation, rather than measured properties of every installed device. Projected-beam detectors are treated separately using beam length and total obscuration. The distinction prevents the spot-detector input from being applied as though it described every sensing arrangement.

C/VM2 permits the activation exposure to be obtained using a model with a ceiling-jet correlation or computational fluid dynamics. The latter approach calculates the gas flow and smoke or soot concentration directly. In either case, the response criterion and its application remain part of the method. A more detailed flow calculation does not by itself establish a different permitted detector threshold.

Prescribed regulatory inputs and experimental detector data answer different questions. The former provide consistency within a particular assessment route, while the latter describe response under the tested smoke and exposure conditions. A calculation identifies which type of evidence it uses so that its result can be interpreted within the correct scope.

Signal processing and unwanted alarms

Unwanted fire alarms influence how a detector's signal is processed. The SFPE Handbook describes algorithms intended to distinguish fire signatures while reducing false and nuisance alarms; their details vary between products. These processing functions can change response relative to a calculation based solely on one smoke concentration.

A Multi-sensor detector combines sensing information, so the response of the combined device needs its own evidence. Knowing the calibration of one sensing element does not describe the complete processing decision, and the same principle applies when an alarm system requires confirmation from another detector. IFEG describes coincident systems that wait for two similar or dissimilar detectors before recognising an alarm.

A specialised nuclear-industry example illustrates the consequence of changing sensitivity. The 2010 supplement to NUREG/CR-6850 requires reduced sensitivity in very early warning systems to be assessed for loss of early detection benefit. It distinguishes early alerts from later fire alarms and controls their settings. Those provisions concern the defined United States nuclear application, not a universal classification of building alarms.

The nuclear report also limits the cabinet applications in which sampling can provide the credited warning. Smoke needs a path from the developing fault to the sampling location, but a tightly sealed cabinet can prevent the transport on which the arrangement depends. This application-specific limitation reinforces the distinction between a sensitive instrument and an effective detection path through the protected equipment.

The full response sequence includes the sensor, processing and the intended protective action. A lower number of nuisance signals alone does not demonstrate that the required fire is detected soon enough. Equally, a sensitive sensor can provide limited benefit if transport or subsequent action is delayed. The performance objective remains the basis for assessing the complete arrangement.

Evidence needed for a response prediction

Schifiliti and colleagues identify a gap between common smoke measurements and the information needed to predict individual detector response. Optical density is widely available, but it does not fully characterise ionisation or scattering behaviour. Manufacturer response algorithms and fuel-specific thresholds may also be unavailable. These limitations need to remain visible in the interpretation of a calculated activation time.

The handbook discusses experiments with detectors present as another source of response evidence. Product performance tests can provide limiting conditions even when detailed response traces are unpublished. The test fuel, smoke production, airflow and geometry still define the scope of that evidence. A result from one controlled exposure does not establish performance for every fire scenario.

Mulholland gives an illustrative calculation linking smoke production, detector response and deteriorating visibility. He explicitly notes that the example omits travel time to the detector and entry delay. He also identifies missing information on particle distributions and detector response to smoke agglomerates. The example therefore demonstrates a method and its data needs rather than establishing a complete real-fire warning time.

A sound comparison holds the fire and other relevant conditions consistent while examining changes in detector location or spacing. Schifiliti and colleagues caution against comparing uncertain smoke-response estimates directly with other predicted safety times without understanding their accuracy. The model, its inputs and its intended use must be compatible. Reported precision in seconds cannot remove uncertainty in the underlying smoke and sensor behaviour.

See also

• Smouldering
• Heat detector
• 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. Australian Building Codes Board.
• Iqbal, N., & Salley, M. H. (2004). Fire dynamics tools (FDTs): Quantitative fire hazard analysis methods for the U.S. Nuclear Regulatory Commission fire protection inspection program (NUREG-1805). U.S. Nuclear Regulatory Commission.
• Ministry of Business, Innovation and Employment. (2025). C protection from fire: Verification method C/VM2 (2nd ed.).
• Mulholland, G. W. (2002). Smoke production and properties. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 2-258–2-268). 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., pp. 4-1–4-43). National Fire Protection Association.
• U.S. Nuclear Regulatory Commission, & Electric Power Research Institute. (2010). Fire probabilistic risk assessment methods enhancements (NUREG/CR-6850, Supplement 1).