Fire Safety Academy Wiki - User contributions [en]

Wildland-urban interface

2026-11-02T17:15:16Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''wildland-urban interface''' ('''WUI''') is the term used for areas where buildings are exposed to fires in natural vegetation. In such areas a [[wildfire]] can spread from the vegetation to buildings and from one building to the next.<ref name="McG-ix">McGrattan et al. (2026), p. ix.</ref><ref name="IFEG-1.6-5">National Research Council of Canada et al. (2005), p. 1.6-5.</ref> Fire engineering guidance notes that flying brands and embers may need to be considered to limit fire spread to adjoining property, and that this can have design implications in areas prone to bushfires and wildfires.<ref name="IFEG-1.6-5" />

Fire can reach a building in three ways: through burning embers and other debris, through radiation and through direct flame contact.<ref name="IFEG-1.6-7">National Research Council of Canada et al. (2005), p. 1.6-7.</ref> Wind can carry embers, also called [[firebrands]], over great distances and start spot fires at other locations.<ref name="Qui-255">Quintiere (2002b), p. 2-255.</ref> The United States Forest Service supported the development of sub-models in the Fire Dynamics Simulator for fire spread in the wildland-urban interface.<ref name="McG-ix" /> Once buildings are alight, the methods for [[fire spread between buildings]] apply, and these rest on the same processes of radiant ignition and brand transport.<ref name="Kan-229">Kanury (2002a), p. 2-229.</ref><ref name="AR-396">Albini and Rand, as cited in Ramachandran (2002c), p. 3-396.</ref>

== Wildfire behaviour ==
Flame spread in a forest depends on the type and arrangement of the fuel, the terrain, the wind and the humidity. It is driven by radiant heat transfer and by convective heating caused by wind or by the slope of the terrain.<ref name="Qui-255" /> In most fires the porous brush on the forest floor burns. In severe fires the crowns of the trees also burn, or only the crowns.<ref name="Qui-255" /> Forest fires accelerate up slopes because the flames lean, an effect that was also seen in the escalator trench in the fire at King's Cross underground station.<ref name="CK-207">Cox and Kumar (2002), p. 3-207.</ref>

Thomas proposed an approximate relation for the rate of spread of wildland fires, wood cribs and urban conflagrations. In it, the spread rate times the bulk density of the fuel is proportional to one plus the wind speed, with a constant of 0.07 for wildland fires.<ref name="Tho-255">Thomas (1971), as cited in Quintiere (2002b), p. 2-255.</ref> The relation shows two trends: the spread rate rises with the wind speed, and it falls as the fuel is packed more densely. Quintiere also compares spread rates for cribs, gorse and heather and urban fires in Japan at wind speeds of up to 30 m/s.<ref name="Qui-255" /> He lists the spread rate of forest and urban fires as of the order of 1 to 100 cm/s.<ref name="Qui-247">Quintiere (2002b), p. 2-247.</ref> A glossary in NUREG-1805 gives 1 to 30 cm/s for wind-driven spread through forest debris.<ref name="Nol-F31">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-31.</ref>

The United States Department of Agriculture Forest Service has developed practical methods for estimating flame spread in forests under a range of field conditions. Rothermel reviewed the theory and models for fire danger rating and fire behaviour prediction, including the computer model BEHAVE.<ref name="Qui-255" /> Breaks in the fuel can stop the spread, but the width needed depends on the size of the fire and its heat transfer.<ref name="Qui-255" /> When a forest fire jumps across a firebreak, this is generally regarded as ignition by radiation.<ref name="Kan-229" />

The advancing front of a forest fire is an example of a line fire, like a row of burning townhouses, and like other natural fires it burns as a diffusion flame.<ref name="IS-9-2">Iqbal and Salley (2004), p. 9-2.</ref><ref name="IS-3-12">Iqbal and Salley (2004), p. 3-12.</ref> Behind the front, the passage of a forest fire often ignites the duff layer, the porous layer of decaying organic material on the forest floor, to smouldering. This smouldering is very irregular in extent and can reignite a flaming fire.<ref name="Ohl-206">Ohlemiller (2002), p. 2-206.</ref>

== How fire reaches buildings ==
The International Fire Engineering Guidelines name three routes by which fire spreads to an adjacent enclosure, building or property boundary: burning embers and other debris, radiation, and direct flame contact.<ref name="IFEG-1.6-7" /> Wind speed and direction influence how far flames project from the windows of a burning building, and openings on both the windward and the leeward side make the effect of wind more significant.<ref name="IFEG-1.6-5" /> A firebrand is a flaming or smouldering airborne object that emerges from a fire and can sometimes ignite remote combustibles. An ember is a particle of solid material that emits radiant energy because of its temperature or because of combustion at its surface.<ref name="IS-F-24">Iqbal and Salley (2004), p. F-24.</ref><ref name="IS-F-22">Iqbal and Salley (2004), p. F-22.</ref> Burning brands that are lifted by the convective flow of a fire and fall onto combustible materials are a common mechanism of fire spread, observed in wildland fires and in warehouse fires.<ref name="Ham-30">Hamins, Maranghides, McGrattan, Ohlemiller, and Anleitner (2005), p. 30.</ref> How far a burning brand rising in the plume may be carried by the wind is often difficult to establish with any certainty.<ref name="IFEG-2.6-12">National Research Council of Canada et al. (2005), p. 2.6-12.</ref>

[[File:fse-wildland-urban-interface-v1.svg|thumb|upright=1.5|alt=A vegetation fire front with wind blowing embers towards a house, heat rays reaching the facade and flames touching nearby vegetation and the wall.|The three routes by which fire spreads from burning vegetation to a building: embers, radiation and direct flame contact. Based on National Research Council of Canada et al. (2005), p. 1.6-7, and Quintiere (2002b), p. 2-255.]]

Radiation from the flames heats building surfaces until they ignite. For fire growth over separate items, such as buildings in a city, radiant heating to ignition is the main process.<ref name="Kan-229" /> Kanury distinguishes piloted ignition, caused by a small flame, a heated wire or a spark, from spontaneous ignition without such a pilot.<ref name="Kan-230">Kanury (2002a), p. 2-230.</ref> Wood heated for less than 30 min needs a minimum heat flux of about 12 kW/m² for piloted ignition and 20 kW/m² for autoignition. For exposures lasting hours or days, a value of 4.3 kW/m² should be used.<ref name="Bab-3.3-4">Babrauskas (2003), as cited in National Research Council of Canada et al. (2005), p. 3.3-4.</ref> Tewarson's critical heat flux for ignition of wood is 10 kW/m²; below this level, ignition and flame spread are not possible.<ref name="Tew-6-4">Tewarson (1995), as cited in Iqbal and Salley (2004), p. 6-4.</ref>

[[File:fse-wildland-urban-interface-v2.svg|thumb|upright=1.5|alt=Bars of heat flux in kW per square metre: wood ignition 4.3, 10, 12 and 20; New Zealand boundary limit 30.|Radiant heat fluxes for the ignition of wood and a regulatory limit for building fires. Based on Babrauskas (2003), as cited in National Research Council of Canada et al. (2005), p. 3.3-4, and other sources in the chart data.]]

The [[combustibility]] of the materials on and around a building therefore matters. Embers that land on combustible surfaces may ignite them, and a radiant flux of 10 to 20 kW/m² may cause objects to ignite.<ref name="Ham-30" /><ref name="Nol-F35">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-35.</ref> The heat flux that a surface receives from a fire depends on the [[flame radiation]] and on the geometry between flame and surface, which engineers express with a view factor.<ref name="IFEG-2.6-12" /><ref name="VR-5.4">Van Rossum (2027), §5.4.</ref>

== Building protection ==
Roofs are exposed to both embers and radiation. British Standard BS 476: Part 3 describes a test of the ignitability and fire spread of external roof surfaces under radiation from a fire in an adjacent building and piloted ignition from flying brands.<ref name="IFEG-2.6-12" /> In the United States, roof coverings are rated in the test of ASTM E108, UL 790 and NFPA 256, which measures the spread of flame, intermittent flame, burning brands and flying brands. Class A roof coverings must limit flame spread to less than 1.82 m, Class B to less than 2.44 m and Class C to less than 3.96 m.<ref name="IS-C-11">Iqbal and Salley (2004), p. C-11.</ref><ref name="IS-C-12">Iqbal and Salley (2004), p. C-12.</ref> For all classes, no flying brands or particles may continue to flame or glow after reaching the floor.<ref name="IS-C-11" />

Separation distance limits the radiation that reaches a building. The International Fire Engineering Guidelines prefer a first-principles approach for each case to methods derived from codes. They refer to an SFPE guide for the time to piloted ignition under flame radiation and for safe separation distances, which often require view factor calculations.<ref name="IFEG-2.6-11">National Research Council of Canada et al. (2005), p. 2.6-11.</ref><ref name="IFEG-2.6-12" />

These methods were developed for fires in buildings. In New Zealand, for example, the Building Code limits the radiation from a building fire to 30 kW/m² at the relevant boundary and 16 kW/m² at 1 m beyond it. External walls closer than 1 m to the boundary must be non-combustible, or must not ignite within 15 or 30 min under a radiant flux of 30 kW/m². The required time depends on the importance level of the building.<ref name="MBIE-43">Ministry of Business, Innovation and Employment (2025b), p. 43.</ref>

Firefighting foams can protect exposed buildings. Class A foams have been used extensively to suppress wildland fires. In tests, Class A foam lengthened the time to ignition of a combustible surface compared with the same rate and duration of water.<ref name="SH-144">Scheffey and Hickey (2002), p. 4-144.</ref> In portable equipment, foam is mainly used against wildland fires in Class A fuels, mixed with water at 0.1 to 1.0 %.<ref name="IS-B-60">Iqbal and Salley (2004), p. B-60.</ref>

== Spread between buildings and conflagration ==
Once buildings burn, the fire can spread through a built-up area as it does through a forest. Ramachandran describes models in which buildings are linked by ''bonds'' whose chance of transmitting fire depends on the building and its contents, the wind and the distance between buildings. He notes that the same problem arises when predicting fire spread in a forest or from building to building in an urban area.<ref name="Ram-395">Ramachandran (2002c), p. 3-395.</ref>

Data from fires in Tokyo were used to express the probability of fire spread between two buildings as an exponential function that decreases with the distance between them. The analysis showed that the construction of the buildings was the main factor in fire spread.<ref name="SJ-396">Sasaki and Jin, as cited in Ramachandran (2002c), p. 3-396.</ref> In a forest model with randomly placed combustible and non-combustible blocks, fire stopped spreading without wind when about 39 % of the blocks were non-combustible. With wind, up to 65 % non-combustible blocks were needed to stop it.<ref name="Nah-396">Nahmias et al., as cited in Ramachandran (2002c), p. 3-396.</ref>

Albini and Rand modelled urban fire spread as depending on whether a burning building is closer to its unburnt neighbours than a safe distance for radiation or brand transfer.<ref name="AR-396" /> Thomas's spread relation was also applied to the Great Fire of London, with a constant of 0.046 at a wind speed of 4.5 m/s, close to the value of 0.07 for wildland fires.<ref name="Tho-255" /> Wildland fires are also listed among the natural catastrophes, such as lightning, earthquakes and wind storms, that make a fire risk of zero unachievable.<ref name="WH-4">Watts and Hall (2002), p. 5-4.</ref>

== Modelling ==
The Fire Dynamics Simulator offers two ways to model fire spread through vegetation. In the particle model, vegetation such as grass, leaves or airborne burning brands is represented by Lagrangian particles that exchange heat and mass with the gas.<ref name="McG-96">McGrattan et al. (2026), p. 96.</ref><ref name="McG-135">McGrattan et al. (2026), p. 135.</ref> For radiation, vegetation is treated as a collection of large, opaque particles that do not scatter.<ref name="McG-76">McGrattan et al. (2026), p. 76.</ref>

The level set model replaces the details of heat transfer and decomposition by an empirical spread rate, which greatly reduces the complexity of the model.<ref name="McG-135" /> By default each point of the fire front spreads as an ellipse, whose length-to-breadth ratio depends on the combined effect of wind and slope.<ref name="McG-135" /><ref name="McG-136">McGrattan et al. (2026), p. 136.</ref> The spread rate of the head of the fire follows Rothermel's model: a rate without wind and slope, increased by factors for wind and slope. The model treats the spread rate as the ratio of the energy received by unburnt fuel to the energy needed for its ignition.<ref name="McG-137">McGrattan et al. (2026), p. 137.</ref> The wind is taken from a reference height of 6.1 m, equivalent to a standard 20 ft weather station.<ref name="McG-138">McGrattan et al. (2026), p. 138.</ref>

[[File:fse-wildland-urban-interface-v3.svg|thumb|upright=1.5|alt=Plan view of a fire front made of small ellipses stretched in the wind direction, with the front at three successive times.|Level set model of wildland fire spread: each point of the front spreads as an ellipse elongated by wind and slope. Based on McGrattan et al. (2026), pp. 135–137.]]

== Limitations ==
Estimates of flame spread require complex analysis and specific data on the fuel, its arrangement and the ambient conditions, and the available formulas are limited.<ref name="Qui-256">Quintiere (2002b), p. 2-256.</ref> The distance over which brands are carried is difficult to establish.<ref name="IFEG-2.6-12" /> The ignition fluxes and radiation limits above come from studies of materials and building fires, not from wildfires, and Thomas's relation is an approximation.<ref name="Bab-3.3-4" /><ref name="MBIE-43" /><ref name="Tho-255" /> A common misconception is that a wildfire only threatens buildings that the flame front touches. Embers carried by the wind can start fires at a distance, and smouldering duff can reignite the fire after the front has passed.<ref name="Qui-255" /><ref name="Ohl-206" />

== See also ==
* [[Facade fire]]
* [[Warehouse fire]]
* [[Car park fire]]

== Notes ==
<references />

== References ==
* Cox, G., & Kumar, S. (2002). Modeling enclosure fires using CFD. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-194–3-218). National Fire Protection Association; Society of Fire Protection Engineers.
* Hamins, A., Maranghides, A., McGrattan, K. B., Ohlemiller, T. J., & Anleitner, R. L. (2005). ''Experiments and modeling of multiple workstations burning in a compartment'' (NIST NCSTAR 1-5E). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Kanury, A. M. (2002a). Flaming ignition of solid fuels. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-229–2-245). National Fire Protection Association; Society of Fire Protection Engineers.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Ohlemiller, T. J. (2002). Smoldering combustion. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-200–2-210). National Fire Protection Association; Society of Fire Protection Engineers.
* Quintiere, J. G. (2002b). Surface flame spread. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-246–2-257). National Fire Protection Association; Society of Fire Protection Engineers.
* Ramachandran, G. (2002c). Stochastic models of fire growth. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-381–3-401). National Fire Protection Association; Society of Fire Protection Engineers.
* Scheffey, J. L., & Hickey, H. E. (2002). Foam system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-123–4-148). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Special Hazards & Building Types]]

Warehouse fire

2026-11-02T17:15:14Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''warehouse fire''' is a fire in a building used to store goods, often in large undivided spaces with goods stacked on pallets or in racks to heights of several metres. The combination of a large floor area, a high fire load and tall vertical fuel surfaces makes such fires grow quickly and burn intensely.<ref name="MBIE13-55">Ministry of Business, Innovation and Employment (2013), p. 55.</ref><ref name="MBIE13-73">Ministry of Business, Innovation and Employment (2013), p. 73.</ref> Warehouse fires are relatively rare, but in an analysis of fire investigations in Greater London from 1996 to 2000, warehouses had the largest mean fire damage area of all occupancies studied.<ref name="MBIE13-73" /><ref name="Hol-B2-7">Holborn et al. (2004), as cited in Australian Building Codes Board (2020a), Data Sheet B2, p. 7.</ref>

Design guidance for warehouses therefore deals with the growth of the fire in the stored goods, the heat it can release, and its control by sprinklers, smoke venting and the fire service. In New Zealand, Verification Method C/VM2 (second edition, 2025) uses an ultra-fast design fire with a peak of 50 MW for storage 3.0 to 5.0 m high. Higher racks get an even faster growth.<ref name="MBIE25-20">Ministry of Business, Innovation and Employment (2025b), p. 20.</ref> In the United States, sprinkler criteria for storage are based on full-scale tests and were merged into NFPA 13 from its 1999 edition.<ref name="Fle-73">Fleming (2002), p. 4-73.</ref>

== Characteristics of warehouse fires ==
A small room of about 200 m² behaves roughly as one well-mixed fire zone. In a warehouse of more than about 1,000 m², the fire is more complex and its burning rate is not uniform. It is expected to travel around the space and to burn with varying intensity. The intensity depends on the distribution of fuel, the geometry, the combustibility of the goods and the ventilation through roof vents, make-up air and, eventually, a failed roof.<ref name="MBIE13-73" />

In most warehouse designs the roof has no fire rating and is expected to fail early. Before it fails, heat and smoke are held in and the walls are exposed more intensely; after it fails, the hot gases escape and flame radiation dominates.<ref name="MBIE13-73" /> The fire behaviour of metal-faced [[sandwich panel]]s cannot be predicted from small-scale cone calorimeter tests, so the usual correlation with the full-scale room test does not apply to them.<ref name="MBIE13-141">Ministry of Business, Innovation and Employment (2013), p. 141.</ref>

Large-scale data are scarce. Fire experiments at warehouse scale are uncommon, and taking such a building to burnout in a test is almost unheard of.<ref name="MBIE13-73" /> The International Fire Engineering Guidelines note that the usual relations for compartment temperatures come from rooms of about 10 m², sometimes up to 50 m². If simultaneous burning is assumed in a large enclosure, the predicted temperatures are unrealistically high. As a first approximation, the floor can be divided into grids of 10 to 50 m², whose heat release rates are added with time delays for the spread from grid to grid.<ref name="IFEG-2.6-12">National Research Council of Canada et al. (2005), p. 2.6-12.</ref>

Stored goods present large vertical surfaces, which favour fire growth far faster than the ''fast'' design fire used for most buildings.<ref name="MBIE13-55" /> Standard t-squared growth curves assume horizontal spread in all directions and do not suit fires that also spread upwards, as in high racks.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref> The point of ignition matters too. In one comparison, igniting a rack at its front face delayed the fire events by about 470 s compared with ignition in the centre of the flue spaces, the gaps between the stored loads.<ref name="Bab-18">Babrauskas (2002b), p. 3-18.</ref><ref name="Bab-19">Babrauskas (2002b), p. 3-19.</ref> Wooden pallets become a fire hazard mainly when idle pallets are stacked many units high.<ref name="Bab-22">Babrauskas (2002b), p. 3-22.</ref>

In a high distribution centre, the hot smoke layer can ignite goods stored high in the racks. Most materials ignite at about 200 to 500 °C, so a smoke layer above 200 °C with goods in it gives a chance of such secondary fires.<ref name="VR-9.3">Van Rossum (2027), §9.3 (Results from zone models).</ref> In the London fire investigation data, a fire damage area of 100 m² was used as the criterion for flashover in large spaces. Of 20 warehouse fires, 20.8 % exceeded this area, and the mean damage area was 170 m².<ref name="Hol-B2-7" /><ref name="Hol-B2-8">Holborn et al. (2004), as cited in Australian Building Codes Board (2020a), Data Sheet B2, p. 8.</ref>

== Fire load ==
The [[fire load density]] of storage depends on the height of the goods. In the International Fire Engineering Guidelines, the storage values from Swiss surveys of 1967 to 1969 are given per metre of storage height and must be multiplied by that height.<ref name="Eng-3.4-2">England et al. (2000), as cited in National Research Council of Canada et al. (2005), p. 3.4-2.</ref> Examples are 600 MJ/m² per metre for storage at a box manufacturer, 1,700 at a carpet manufacturer and 4,200 at a cardboard manufacturer.<ref name="Eng-3.4-3">England et al. (2000), as cited in National Research Council of Canada et al. (2005), p. 3.4-3.</ref> For occupancies with very different goods, such as industrial buildings, the coefficient of variation is 50 to 80 % of the average, and isolated peaks can reach 2.5 times the average.<ref name="IFEG-3.4-2">National Research Council of Canada et al. (2005), p. 3.4-2.</ref>

New Zealand's C/VM2 (second edition, 2025) prescribes a design fire load of 1,200 MJ/m² for storage up to 3.0 m high, other than foamed plastics. For storage of combustible materials above 3.0 m, it prescribes 800 MJ/m² per metre of height, with a minimum of 2,400 MJ/m².<ref name="MBIE25-22">Ministry of Business, Innovation and Employment (2025b), p. 22.</ref> A store of 6 m high goods therefore has a design fire load of 4,800 MJ/m². In Australia, the 2020 data sheets for the Fire Safety Verification Method assign a ''very high'' fire load distribution to storage buildings, with a mean of 1,000 MJ/m², but explicitly exclude high-rack storage from it.<ref name="ABCB-B2-9">Australian Building Codes Board (2020a), Data Sheet B2, p. 9.</ref>

The total energy in a store varies greatly with what is stored. The C/VM2 commentary compares 1,000 m³ of different goods: low-density polyurethane foam contains about 1.3 TJ, MDF board 11.5 TJ and rolled polyethylene about 20 TJ.<ref name="MBIE13-25">Ministry of Business, Innovation and Employment (2013), p. 25.</ref><ref name="MBIE13-27">Ministry of Business, Innovation and Employment (2013), p. 27.</ref> The book ''Introduction to Fire Safety Engineering'' gives the example of a shed with 3,500 m³ of stored goods with a heat of combustion of 23 MJ/kg and a density of 135 kg/m³. This contains 10,867,500 MJ, equivalent to about 572,000 kg of wood.<ref name="VR-4.4">Van Rossum (2027), §4.4.</ref>

== Fire growth and heat release ==
Design fires for warehouses usually describe the growth of the fire with a t-squared curve, in which the heat release rate increases with the square of the time since ignition. The growth coefficient expresses how quickly the fire grows; for a ''fast'' fire, for example, it is 0.047 kW/s²:<ref name="ABCB-B2-3">Australian Building Codes Board (2020a), Data Sheet B2, p. 3.</ref>

<math display="block">\dot{Q} = \alpha \, t^2</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\dot{Q}</math> || heat release rate || kW
|-
| <math>\alpha</math> || fire growth coefficient || kW/s²
|-
| <math>t</math> || time from effective ignition || s
|}

For an ''ultra-fast'' fire, α is 0.188 kW/s², which means the fire reaches 1,055 kW after 75 s.<ref name="ABCB-B2-3" /> C/VM2 uses this value for buildings that can store goods to a height of 3.0 to 5.0 m, with a peak of 50 MW. The growth stops at that peak, at flashover, at the limit set by the available ventilation or, for some calculations, at sprinkler activation. For goods stored above 5.0 m, C/VM2 uses a cubic growth that also increases with the storage height, so the t-squared curve is not valid for high racks.<ref name="MBIE25-20" /><ref name="MBIE25-21">Ministry of Business, Innovation and Employment (2025b), p. 21.</ref> The commentary explains that storage up to 5.0 m is treated as ultra-fast following NFPA 72 and NFPA 204, and that growth increases further with taller vertical surfaces.<ref name="MBIE13-55" />

As a worked example, take a store with goods stacked 4 m high. With α = 0.188 kW/s², the heat release rate after 2 min is 0.188 × 120² = 2,707 kW, or about 2.7 MW. After 5 min it is 0.188 × 300² = 16,900 kW. The 50 MW peak is reached after √(50,000 / 0.188) = 516 s, or about 8.6 min. For comparison, the general design fire for most other buildings in C/VM2 grows at 0.0469 kW/s² to 20 MW, which takes 653 s.<ref name="MBIE25-20" /> The book uses slightly different reference values: it defines the growth time as the time to reach 1 MW, 75 s for ultra-fast growth.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref>

[[File:fse-warehouse-fire-v1.svg|thumb|upright=1.5|alt=Line chart: storage fire grows as 0.188 t squared to 50 MW at 516 s; general fire as 0.0469 t squared to 20 MW at 653 s.|Design fires of New Zealand's C/VM2 (2025): storage 3.0 to 5.0 m high (ultra-fast, peak 50 MW) and most other buildings (peak 20 MW). Based on Ministry of Business, Innovation and Employment (2025b), p. 20.]]

Measurements show how fast warehouse goods burn. According to NUREG-1805, most warehouse materials in a table taken from NFPA 72 fall in the fast or ultra-fast range.<ref name="IS-B-6">Iqbal and Salley (2004), p. B-6.</ref> Stacks of wood pallets 3 m high reach about 1,055 kW in 80 to 110 s, and vertical rolls of paper 6 m high in only 15 to 28 s.<ref name="NFPA72-B6">NFPA 72 (1999), as cited in Iqbal and Salley (2004), p. B-6.</ref><ref name="NFPA72-B7">NFPA 72 (1999), as cited in Iqbal and Salley (2004), p. B-7.</ref> Wood pallets stacked to 3 m release about 7 MW, and polystyrene jars in cartons stacked 4.9 m high about 30 to 40 MW.<ref name="KQ-3-2">Karlsson and Quintiere (1999), as cited in Iqbal and Salley (2004), p. 3-2.</ref>

Peak heat release rates per unit floor area of stored commodities, analysed by Heskestad from Factory Mutual tests, range widely. Cardboard cartons with a metal liner stacked 2.95 m high gave about 1,490 to 1,680 kW/m², while polyethylene bins in cardboard cartons stacked 4.51 m high gave 28,900 kW/m².<ref name="Hes-18">Heskestad (1997), as cited in Babrauskas (2002b), p. 3-18.</ref> These values cannot be scaled: they may be applied conservatively to lower stacks, but not extrapolated to higher ones.<ref name="Bab-17">Babrauskas (2002b), p. 3-17.</ref> Factory Mutual classifies stored goods for sprinkler protection by how much heat they release in the presence of water. Raising the share of polystyrene in cardboard boxes from about 15 % to 100 % increases the heat release rate exponentially and raises the water demand.<ref name="Tew-153">Tewarson (2002), p. 3-153.</ref><ref name="Tew-154">Tewarson (2002), p. 3-154.</ref>

[[File:fse-warehouse-fire-v3.svg|thumb|upright=1.5|alt=Bars on a logarithmic scale from 250 kW per square metre for an average building to 28,900 kW per square metre for plastic bins in cartons.|Peak heat release rate per unit floor area of stored commodities from Factory Mutual tests, compared with 250 kW/m² for a developed fire in an average building. Based on Heskestad (1997), as cited in Babrauskas (2002b), p. 3-18, and Van Rossum (2027), §4.5.]]

== Sprinkler protection ==
Classifying the occupancy and the stored commodity is the most critical step in sprinkler design. If the hazard is underestimated, the fire can overpower the sprinklers.<ref name="Fle-73" /> This step is described under [[sprinkler hazard classification]]. Sprinkler design for fire suppression is largely empirical. The criteria for rack storage and for large-drop and early suppression sprinklers are based on full-scale tests, while most other criteria rest on experience and judgement.<ref name="Fle-73" />

Ordinary sprinklers are designed to control a fire. An [[ESFR sprinkler]], short for early suppression fast response, combines a fast response with large water drops and a high velocity, so that it can extinguish a high-challenge fire. Its installation rules are strict, and in-rack sprinklers are a separate type placed within the racks.<ref name="IS-10-3">Iqbal and Salley (2004), p. 10-3.</ref> The variables of storage protection include the water application rate, the response time index, the temperature rating and in-rack protection, and the storage arrangement matters as well.<ref name="Sch-113">Scheffey (2002), p. 4-113.</ref> Water from sprinklers also wets the goods, which helps to stop the fire from jumping across the aisles of a warehouse.<ref name="Tew-151">Tewarson (2002), p. 3-151.</ref>

Sprinklers in storage buildings perform slightly below average.<ref name="ABCB-C2-76">Australian Building Codes Board (2020a), Data Sheet C2, p. 76.</ref> The Australian data sheets give a reliability of 84 %, an efficacy of 97 % and an effectiveness of 83 % for storage buildings.<ref name="ABCB-C2-71">Australian Building Codes Board (2020a), Data Sheet C2, p. 71.</ref> United States data for 2010 to 2014 show that wet-pipe sprinklers in warehouses, excluding cold storage, operated in 84 % of fires and were effective in 82 %.<ref name="Ahr-C2-78">Ahrens (2017), as cited in Australian Building Codes Board (2020a), Data Sheet C2, p. 78.</ref>

== Smoke and heat venting ==
[[Smoke and heat exhaust ventilation]] in a warehouse aims to support escape and firefighting, so that firefighters can enter and see the seat of the fire, and to reduce property damage. A typical arrangement is a large space with vents in the roof, often divided by draft curtains into separate smoke reservoirs.<ref name="Coo-219">Cooper (2002b), p. 3-219.</ref> Without venting, smoke fills such a space quickly. Cooper considers goods that reach about 1 MW in 100 s, in a warehouse 6 m high with a floor area of 1,500 m². He calculates that the smoke layer becomes untenable at 183 °C after about 350 s. At that moment its lower edge is 2.7 m above the floor, and after about 600 s the layer fills the whole warehouse at about 430 °C.<ref name="Coo-265">Cooper (2002a), p. 3-265.</ref><ref name="Coo-266">Cooper (2002a), p. 3-266.</ref>

[[File:fse-warehouse-fire-v2.svg|thumb|upright=1.5|alt=Cross-section of a warehouse with high racks, a fire in one rack, a smoke layer under the roof, roof vents, draft curtains above the aisles and sprinklers.|Cross-section of a racked warehouse with a fire in one rack: smoke layer, roof vents, draft curtains above the aisles, ceiling and in-rack sprinklers. Based on Cooper (2002b), pp. 3-219 and 3-237, and Iqbal and Salley (2004), p. 10-3.]]

Roof vents rely mainly on the buoyancy of the smoke. Strong wind can create suction over the roof and increase the extraction, but the vent may then draw in clear air from below the smoke layer, so that a critical vent area exists.<ref name="Kan-24">Kandola (2002), p. 1-24.</ref><ref name="Kan-25">Kandola (2002), p. 1-25.</ref> Cooper gives a worked example for a 5 m × 5 m fire in a warehouse with a ceiling at 9 m and draft curtains reaching down to 6 m. To keep the smoke above the bottom of the curtains, each curtained space needs a total vent area of 18.6 m².<ref name="Coo-235">Cooper (2002b), p. 3-235.</ref>

The combination of vents and sprinklers was long controversial, and no generally accepted design guidelines existed when Cooper's chapter was written.<ref name="Coo-236">Cooper (2002b), p. 3-236.</ref> A review of 34 position papers, as cited by Cooper, concluded that venting does not harm sprinkler performance. It has no effect on the activation of the first sprinklers or on the total number activated, although a vent directly above the fire may slightly delay the first sprinklers. The review also advised placing draft curtains in aisles rather than over storage.<ref name="BC-237">Beyler and Cooper (in press), as cited in Cooper (2002b), p. 3-237.</ref> The Australian data sheets warn, on the other hand, that a vent may remove hot gases before sprinklers activate, which could compromise early suppression systems.<ref name="ABCB-B1-21">Australian Building Codes Board (2020a), Data Sheet B1, p. 21.</ref>

Sprinkler spray can drag the smoke layer down and mix it with the air below, an effect called smoke logging. Once this happens, the benefits of venting are lost, even with very large vents.<ref name="Coo-237">Cooper (2002b), p. 3-237.</ref><ref name="Coo-238">Cooper (2002b), p. 3-238.</ref> In the Netherlands, the book describes the criteria of the standard NEN 6093 for fire service intervention in a space with smoke and heat exhaust. Visibility up to a height of 1.8 m must be at least 100 m, and radiation from the smoke layer at that height at most 3.5 kW/m². The layer temperature may be at most 200 °C if goods are stored in the smoke layer and 500 °C otherwise.<ref name="VR-9.3" />

== Detection and other measures ==
Detection in high warehouses is a design problem of its own. In a design example for a furniture warehouse, Schifiliti and co-workers show that heat detectors must be spaced more closely as the distance from the fire to the ceiling grows. For a floor area of 5,000 m², about 48 fixed-temperature detectors or about 8 rate-of-rise detectors were needed to detect a fire of about 2,000 kW.<ref name="Sch-17">Schifiliti et al. (2002), p. 4-17.</ref><ref name="Sch-18">Schifiliti et al. (2002), p. 4-18.</ref> For a storage room of 50 m × 75 m × 10 m, a projected beam smoke detector had to be set to respond at about 30 % total obscuration or less.<ref name="Sch-26">Schifiliti et al. (2002), p. 4-26.</ref> Such detectors typically respond at 20 to 70 % total obscuration.<ref name="MBIE25-28">Ministry of Business, Innovation and Employment (2025b), p. 28.</ref>

In New Zealand, C/VM2 (second edition, 2025) allows manual activation of the fire alarm in buildings such as warehouses, where the ceiling is high and the occupant load is low. The conditions are an average ceiling height of at least 5 m and fewer than 50 occupants who are awake and familiar with the building.<ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref><ref name="MBIE25-28" /> The rule rests on 34 simulations with the Fire Dynamics Simulator. In these, the 12 m high building had the fastest ceiling jet, because the growth of a rack fire increases with rack height.<ref name="MBIE13-109">Ministry of Business, Innovation and Employment (2013), p. 109.</ref>

Fire spread can also be limited by the way goods are stored. The International Fire Engineering Guidelines mention changing from rack storage to palletised storage and storing items horizontally rather than vertically.<ref name="IFEG-1.4-7">National Research Council of Canada et al. (2005), p. 1.4-7.</ref> Another approach is to lower the oxygen level, the principle behind an [[oxygen reduction system]]. Air contains 23.3 % oxygen by mass.<ref name="Klo-279">Klote (2002), p. 4-279.</ref> Tewarson shows that flames of PMMA, a plastic, go out at an oxygen mass fraction of about 0.18 under normal conditions. When the fuel receives an external heat flux of 65 kW/m², it keeps burning down to 0.128.<ref name="Tew-155">Tewarson (2002), p. 3-155.</ref>

== Regulation and limits ==
Building codes set limits on large stores. In New Zealand, C/VM2 (second edition, 2025) sets conditions for unsprinklered firecells within 15 m of a relevant boundary that have a fire load above 20 TJ or a floor area above 5,000 m². When firefighters first apply water, the radiation at 1.5 m above the floor must not exceed 4.5 kW/m², and the smoke layer must be at least 2 m above the floor.<ref name="MBIE25-53">Ministry of Business, Innovation and Employment (2025b), p. 53.</ref> The commentary explains that the 5,000 m² limit was based on 800 MJ/m² per metre of storage height and a storage height of 5.0 m.<ref name="MBIE13-23">Ministry of Business, Innovation and Employment (2013), p. 23.</ref> With these values, 5,000 m² corresponds to 20 TJ.

The commentary works this out for an unsprinklered warehouse of 1,200 m² and 8.0 m high, with goods stored to 6.0 m. With an ultra-fast fire to 50 MW and water applied after 1,200 s, the criteria were met with automatic make-up air and 22 roof vents of 2.0 m², 3.7 % of the floor area.<ref name="MBIE13-150">Ministry of Business, Innovation and Employment (2013), p. 150.</ref> The Building Code of Australia, as described in the commentary, requires sprinklers where more than 1,000 m³ of combustible goods are stored higher than 4.0 m.<ref name="BCA-25">Building Code of Australia, as cited in Ministry of Business, Innovation and Employment (2013), p. 25.</ref> In the Netherlands, buildings with large fire compartments outside the scope of NEN 6060 are a typical field for fire safety engineering.<ref name="VR-1.5">Van Rossum (2027), §1.5.</ref>

The data behind warehouse design are limited. The heat release rates per unit area apply only to the tested storage heights, the Australian growth rate statistics for storage rest on only six fires, and full-scale burnout tests are almost never done.<ref name="Bab-17" /><ref name="Hol-B2-4">Holborn et al. (2004), as cited in Australian Building Codes Board (2020a), Data Sheet B2, p. 4.</ref><ref name="MBIE13-73" /> A common misconception is that a store full of goods burns like an office with a higher fire load. In reality, tall vertical fuel surfaces make the growth much faster, and the fire spreads upwards through the racks rather than only across the floor.<ref name="MBIE13-55" /><ref name="VR-8.6" />

== See also ==
* [[Beam smoke detector]]
* [[Sprinkler-vent interaction]]
* [[Car park fire]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Babrauskas, V. (2002b). Heat release rates. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-1–3-37). National Fire Protection Association; Society of Fire Protection Engineers.
* Cooper, L. Y. (2002a). Compartment fire-generated environment and smoke filling. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-243–3-267). National Fire Protection Association; Society of Fire Protection Engineers.
* Cooper, L. Y. (2002b). Smoke and heat venting. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-219–3-242). National Fire Protection Association; Society of Fire Protection Engineers.
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Kandola, B. S. (2002). Introduction to mechanics of fluids. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-1–1-26). National Fire Protection Association; Society of Fire Protection Engineers.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Scheffey, J. L. (2002). Foam agents and AFFF system design considerations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-88–4-122). National Fire Protection Association; Society of Fire Protection Engineers.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Tewarson, A. (2002). Generation of heat and chemical compounds in fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-82–3-161). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Firebrands

2026-11-02T17:15:12Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Firebrands''' are flaming or smouldering airborne objects that emerge from a fire and can sometimes ignite combustible material at a distance from it.<ref name="IS-F24">Iqbal and Salley (2004), p. F-24.</ref> The terms flying brands, burning brands and embers are also used.<ref name="Qui-255">Quintiere (2002b), p. 2-255.</ref><ref name="IFEG-2.6-12">National Research Council of Canada et al. (2005), p. 2.6-12.</ref><ref name="HMMOA-30">Hamins, Maranghides, McGrattan, Ohlemiller, and Anleitner (2005), p. 30.</ref> An ember is a particle of solid material that emits radiant energy because of its temperature or because it is burning at its surface.<ref name="IS-F22">Iqbal and Salley (2004), p. F-22.</ref> Firebrands rise with the hot gases above a fire and can be carried by the wind before they fall onto combustible material.<ref name="HMMOA-30" /><ref name="IFEG-2.6-12" />

Burning embers and other debris are one of three ways in which fire spreads from one building or property to another, besides radiation and direct flame contact.<ref name="IFEG-1.6-7">National Research Council of Canada et al. (2005), p. 1.6-7.</ref> In forest fires, wind currents can carry large embers over great distances and start spot fires at other locations.<ref name="Qui-255" /> In areas prone to bushfires or wildfires, the spread of flying brands and embers can therefore have consequences for building design.<ref name="IFEG-1.6-5">National Research Council of Canada et al. (2005), p. 1.6-5.</ref> The distance a burning brand can travel is difficult to establish with any certainty, which makes firebrands hard to quantify in design.<ref name="IFEG-2.6-12" />

== Formation and transport ==
Transport starts in the fire plume. Above an open fire, about 70% of the heat leaves by convection and about 30% by radiation, so most of the energy goes into a rising column of hot gas.<ref name="VR-5.3">Van Rossum (2027), §5.3.</ref> Convective forces in this plume lift burning brands, which fall onto combustible materials elsewhere. This lofting has been observed in warehouse fires and in wildland fires, both of which involve cellulosic materials.<ref name="HMMOA-30" /> While it rises and falls, a brand may be carried by the wind. ''Introduction to Fire Safety Engineering'' notes that wind is difficult to predict, and the International Fire Engineering Guidelines state that the travel distance of a brand rising in the plume is difficult to establish.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref><ref name="IFEG-2.6-12" />

[[File:fse-firebrands-v1.svg|thumb|upright=1.5|alt=Side view: a burning building sends a plume upward; small burning brands rise in it, drift downwind and land on a roof and on vegetation.|Firebrands are lofted by the convective plume, carried by the wind and can ignite combustible material where they land. Based on Hamins, Maranghides, McGrattan, Ohlemiller, and Anleitner (2005), p. 30, and National Research Council of Canada et al. (2005), p. 2.6-12.]]

Computer models treat firebrands as particles moving through the gas. In the Fire Dynamics Simulator, a solid-phase reaction can produce material that leaves the burning surface as particles, which the documentation calls a firebrand.<ref name="FDS-222">McGrattan et al. (2026), p. 222.</ref> Lagrangian particles can represent airborne vegetation such as burning brands. Each particle feels gravity and a drag force from the surrounding gas, which depends on its size, shape and speed relative to the flow.<ref name="FDS-95">McGrattan et al. (2026), pp. 95–96.</ref> Drag coefficients are given for cylindrical and for thin disk-shaped particles.<ref name="FDS-95" /> The United States Forest Service supported the development of such sub-models to simulate fire spread at the [[wildland-urban interface]].<ref name="FDS-ix">McGrattan et al. (2026), p. ix.</ref>

== Ignition by firebrands ==
A firebrand starts a new fire when it lands on a receptive fuel and causes [[ignition]]. A glowing brand can act as a pilot source. Piloted ignition is ignition in the presence of a small flame, a heated wire, an electric spark or an incandescent particle, as opposed to spontaneous ignition without such a source.<ref name="Kan-230">Kanury (2002a), p. 2-230.</ref><ref name="Kan-232">Kanury (2002a), p. 2-232.</ref> The pilot source is usually so small that its contribution to heating the solid can be ignored. Its role is to start combustion in a flammable mixture of fuel gases and air near the surface.<ref name="Kan-232" />

Ignition can take three forms. Combustion is called smouldering when it takes place within the solid below the surface, glowing when it takes place at the surface, and flaming when it takes place in the gas phase.<ref name="Kan-230" /> Because a firebrand may be smouldering rather than flaming, [[smouldering]] combustion is part of the picture.<ref name="IS-F24" /> Smouldering can turn into flaming when the supply of oxygen is increased, for example by airflow over the fuel.<ref name="Ohl-208">Ohlemiller (2002), p. 2-208.</ref> In experiments by Palmer on layers of wood sawdust, the transition occurred at airflow velocities of about 0.9 to 1.7 m/s, but only when the air moved in the same direction as the smoulder front. Ohlemiller compares this with starting a camp fire from tinder and sparks.<ref name="Ohl-208" />

[[File:fse-firebrands-v3.svg|thumb|upright=1.5|alt=Cross-section of a fuel bed with a landed brand: smouldering zone inside, glowing surface, and a flame above once air flows over the bed.|Smouldering, glowing and flaming combustion after a firebrand lands on a fuel bed; airflow in the direction of the smoulder front can trigger the transition to flaming. Based on Kanury (2002a), p. 2-230, and Ohlemiller (2002), p. 2-208.]]

The time to flaming can be long. In a literature survey of smouldering in upholstered chairs and beds, about two-thirds of the items tested changed to flaming combustion, after between 22 and 306 minutes.<ref name="Ohl-208" /> Smouldering ignition can therefore take a long time to develop into a flaming fire. Glowing particles from a fire can also ignite a flammable mixture of pyrolysis gases inside a space and cause a smoke gas explosion.<ref name="VR-4.10">Van Rossum (2027), §4.10.</ref> For comparison, most materials ignite at temperatures between about 200 and 500 °C when they are heated by a hot smoke layer.<ref name="VR-9.3b">Van Rossum (2027), §9.3 (Results from zone models).</ref>

== Firebrands in wildfires and building fires ==
In a [[wildfire]] in a forest, flame spread depends on radiant heat transfer and on convective heating by wind or by the slope of the terrain. In severe fires the crowns of the trees also burn.<ref name="Qui-255" /> Wind currents can then carry large embers great distances from the fire, where they start spot fires. Breaks in the fuel can be used to control the spread, but the break needed to interrupt the fire depends on the size of the fire and its heat transfer.<ref name="Qui-255" />

In building fires, firebrands are one route of [[fire spread between buildings]]. The International Fire Engineering Guidelines list burning embers and other debris, radiation and direct flame contact as the ways in which fire can spread to an adjacent enclosure, building or property boundary.<ref name="IFEG-1.6-7" /> Depending on how a building and its contents burn, the distribution of flying brands and embers may need to be considered if fire spread to adjoining property is to be limited.<ref name="IFEG-1.6-5" /> Protection of adjacent buildings against flying brands is also listed among the objectives that fire service intervention and firefighting facilities support.<ref name="IFEG-2.9-3">National Research Council of Canada et al. (2005), p. 2.9-3.</ref>

[[File:fse-firebrands-v2.svg|thumb|upright=1.5|alt=Three panels with two buildings each: embers flying to the neighbour's roof, heat radiation crossing the gap, and flames touching the neighbouring wall.|The three routes of fire spread to an adjacent building: burning embers and debris, radiation and direct flame contact. Based on National Research Council of Canada et al. (2005), p. 1.6-7.]]

Firebrands can also be produced by a building collapse. When the North Tower of the World Trade Center collapsed in 2001, most burning material would have been extinguished. NIST considered it possible, however, that many potential ignition sources survived and were spread with the dust- and debris-laden air. These flows were many storeys tall and passed around and beyond World Trade Center 7.<ref name="McA-189">McAllister et al. (2008), p. 189.</ref> Not every fire spreads by brands: in NIST's experiments with burning office workstations, the video record showed no evidence that flying brands played a role in flame spread.<ref name="HMMOA-30" />

== Roofs and testing ==
Several test standards expose roof coverings to brands. British Standard BS 476 Part 3 describes a test of the ignitability and fire spread of external roof surfaces. It exposes the roof to radiation from a fire in an adjacent building and to piloted ignition from flying brands.<ref name="IFEG-2.6-12" /> In the United States, the roof covering tests ASTM E108 and UL 790 include a burning brand test and a flying brand test besides tests for flame spread and intermittent flame.<ref name="IS-C12">Iqbal and Salley (2004), p. C-12.</ref>

The American roof covering standard NFPA 256 classifies roofs by the flame spread in the test. Class A allows less than 1.82 m (6 ft), class B less than 2.44 m (8 ft) and class C less than 3.96 m (13 ft).<ref name="IS-C11">Iqbal and Salley (2004), p. C-11.</ref> For all classes there may be no significant lateral flame spread and no flaming on the underside of the deck. Flying brands or particles from the roof covering may not continue to flame or glow after reaching the floor.<ref name="IS-C11" />

[[File:fse-firebrands-v4.svg|thumb|upright=1.5|alt=Bar chart of maximum flame spread for roof covering classes: class A 1.82 m, class B 2.44 m, class C 3.96 m.|Maximum flame spread for roof covering classes A, B and C in NFPA 256. Based on Iqbal and Salley (2004), p. C-11.]]

== Limitations and open questions ==
The fire safety engineering documents used for this article treat firebrands only briefly. They define the phenomenon, describe how brands are lofted and carried, and name roof tests and design considerations, but give no measured data on brand sizes, numbers or travel distances.<ref name="IFEG-2.6-12" /><ref name="IS-F24" /> The International Fire Engineering Guidelines state plainly that the distance a burning brand may be carried cannot be established with any certainty.<ref name="IFEG-2.6-12" /> In these sources, firebrand exposure is addressed through roof covering tests and design considerations for bushfire-prone areas rather than through calculation methods.<ref name="IFEG-1.6-5" /><ref name="IS-C11" />

Numerical models can follow individual brands as particles, but their results depend on the assumed shape, size and drag of each particle and on the wind field.<ref name="FDS-95" /> Whether a smouldering brand leads to flaming also depends on the airflow over the fuel.<ref name="Ohl-208" /> Firebrands are not limited to forest fires. The same mechanism has been observed in warehouse fires, and it is one of the routes by which fire spreads between buildings.<ref name="HMMOA-30" /><ref name="IFEG-1.6-7" />

== See also ==
* [[Fire spread]]
* [[Facade fire]]
* [[Warehouse fire]]

== Notes ==
<references />

== References ==
* Hamins, A., Maranghides, A., McGrattan, K. B., Ohlemiller, T. J., & Anleitner, R. L. (2005). ''Experiments and modeling of multiple workstations burning in a compartment'' (NIST NCSTAR 1-5E). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Kanury, A. M. (2002a). Flaming ignition of solid fuels. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-229–2-245). National Fire Protection Association; Society of Fire Protection Engineers.
* McAllister, T. P., Gann, R. G., Averill, J. D., Gross, J. L., Grosshandler, W. L., Lawson, J. R., McGrattan, K. B., Pitts, W. M., Prasad, K. R., Sadek, F. H., & Nelson, H. E. (2008). ''Structural fire response and probable collapse sequence of World Trade Center Building 7'' (NIST NCSTAR 1-9, Vols. 1–2). National Institute of Standards and Technology.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Ohlemiller, T. J. (2002). Smoldering combustion. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-200–2-210). National Fire Protection Association; Society of Fire Protection Engineers.
* Quintiere, J. G. (2002b). Surface flame spread. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-246–2-257). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Tunnel fire

2026-11-02T17:15:10Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''tunnel fire''' is a fire in a road, rail or service tunnel, usually involving one or more vehicles or the installations in the tunnel. Because a tunnel is enclosed, the smoke of the fire stays around the people inside it, and their ability to move away from the fire is restricted.<ref name="Buk-229">Bukowski (2002), p. 5-229.</ref> Temperatures can become far higher than in an ordinary room fire: in the Summit rail tunnel fire on British Railways in 1984, the faces of brick-lined ventilation shafts fused.<ref name="DoT-13-7">Department of Transport (1984), as cited in Iqbal and Salley (2004), p. 13-7.</ref>

In the Netherlands, tunnels do not fit well into the building regulations, which are written mainly for buildings in which people stay. Fire safety engineering is therefore applied to tunnels on several aspects, and tunnels are usually fitted with a smoke management system.<ref name="VR-11.2">Van Rossum (2027), §11.2.</ref> [[Tunnel ventilation]] can keep part of a tunnel free of smoke by blowing air against the smoke, but the same airflow brings oxygen to the fire.<ref name="Klo-279">Klote (2002), p. 4-279.</ref> In fire tests, the peak heat release rate of vehicles ranged from about 1.5 MW for an older car to about 120 MW for a loaded truck.<ref name="Bab-27">Babrauskas (2002b), p. 3-27.</ref><ref name="Bab-28">Babrauskas (2002b), p. 3-28.</ref>

== Why tunnels differ from buildings ==
Bukowski describes the special hazards of rail tunnels in his chapter on transportation vehicles. In an accident in a tunnel, the fire effluent stays around the train and keeps exposing passengers and crew who have left the vehicles. Tunnels, bridges and elevated track also restrict how far people can move away from the train, which must be addressed for long tunnels.<ref name="Buk-229" /> Some fire scenarios are a threat to passengers only when they occur in a tunnel, because the tunnel increases the exposure to smoke.<ref name="Buk-231">Bukowski (2002), p. 5-231.</ref>

Most fire scenarios in trains lead to short exposures, mainly to heat. The exceptions are sleeping cars without fire detection and incidents in a long tunnel. For this reason, the operating procedure is that a train in which an emergency occurs keeps moving until it is out of the tunnel, if possible.<ref name="Buk-230">Bukowski (2002), p. 5-230.</ref> Trains are long and narrow spaces, so some assumptions of zone models may not be appropriate for them.<ref name="Buk-229" />

The temperatures in a tunnel fire can be extreme. A post-flashover compartment fire typically reaches about 1,100 °C, according to Drysdale as cited by Iqbal and Salley, but temperatures of 1,300 to 1,400 °C occasionally occur.<ref name="Dry-13-7">Drysdale (1998), as cited in Iqbal and Salley (2004), p. 13-7.</ref> Such temperatures are enough to fuse the surface of bricks, as happened in the ventilation shafts of the Summit tunnel.<ref name="DoT-13-7" /> For comparison, the cellulosic fire curves used for ordinary fire resistance tests reach a maximum of just over 900 °C.<ref name="IS-F-13">Iqbal and Salley (2004), p. F-13.</ref>

Regulation treats tunnels as special structures. The book ''Introduction to Fire Safety Engineering'' lists tunnels among the examples of buildings for which the Dutch building regulations are not effective. Other examples are public transport terminals and towers with a floor above 200 m. For such buildings, fire safety engineering is used.<ref name="VR-1.5">Van Rossum (2027), §1.5.</ref>

In New Zealand, Verification Method C/VM2 (second edition, 2025) names tunnels as an example of buildings outside its scope. Their compliance can be shown with an alternative solution.<ref name="MBIE-6">Ministry of Business, Innovation and Employment (2025b), p. 6.</ref> The International Fire Engineering Guidelines (2005) were written for buildings, but state that their concepts may also help in the design of ships and tunnels.<ref name="IFEG-0.1-3">National Research Council of Canada et al. (2005), Part 0 (Australia), p. 0.1-3.</ref> In the United States, road tunnels are covered by NFPA 502, listed in its 2001 edition in NUREG-1805.<ref name="IS-E-8">Iqbal and Salley (2004), p. E-8.</ref>

== Design fires ==
The design fire of a tunnel is usually a vehicle fire. The book gives a peak [[heat release rate]] of 8.5 MW for a car, and an energy content of 6,000 to 12,000 MJ per car and 19,000 MJ for an unloaded truck.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref><ref name="VR-4.4">Van Rossum (2027), §4.4.</ref> Measured peaks for single cars vary widely. Tests at the Fire Research Station in the United Kingdom gave 8.5 MW for an Austin Maestro and 4.5 MW for a Citroën BX. Finnish tests of cars from the late 1970s gave about 1.5 to 2.0 MW.<ref name="Bab-27" />

Larger vehicles release much more heat. In a European research programme, nearly two dozen vehicles were burnt, including cars, buses, trams, trucks and rail vehicles. Many of the peak heat release rates lay between about 10 and 35 MW. One test involved a truck loaded with 2 tonnes of modern upholstered furniture, which reached about 120 MW.<ref name="Bab-27" /><ref name="Bab-28" /> That is about 14 times the peak of a car fire of 8.5 MW. Truck tyres can ignite from an overheated axle and release a substantial amount of heat.<ref name="Bab-27" />

[[File:fse-tunnel-fire-v3.svg|thumb|upright=1.5|alt=Bars of peak heat release rates: cars 1.5 to 8.5 MW, electric car about 8 MW, large vehicles 10 to 35 MW, furniture truck about 120 MW.|Peak heat release rates of vehicle fires from tests and reviews. Based on Babrauskas (2002b), pp. 3-27–3-28, and Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 25.]]

In the Netherlands, the car fire curve in the standard NEN 6098 for mechanically ventilated car parks is based on a TNO design fire from 1999. In those tests, fire could spread to neighbouring cars 10 to 15 min after the first car, and the curve assumes three burning cars after 15 min.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref> Modern cars, with their high plastic content, have a larger fire load and a higher heat release rate than older steel models.<ref name="VdG-11">Van der Graaf et al. (2021/2023), p. 11.</ref> The book therefore advises checking whether such curves still fit current vehicles.<ref name="VR-8.6" />

An [[electric vehicle fire]] has a similar peak to a conventional car fire, about 6 MW according to a review by Sun and co-workers. Vehicles with batteries of more than 100 kWh reach about 8 MW.<ref name="Sun-25">Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 25.</ref> Without intervention, an electric car burns out in about 90 min, and after extinguishing the battery can keep reigniting for several hours.<ref name="NFPA-26">NFPA (2013) and Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 26.</ref> A battery pack of 100 kWh can also release 2 to 20 kg of hydrogen fluoride, a toxic gas.<ref name="Lar-26">Larsson et al. (2017), as cited in Van der Graaf et al. (2021/2023), p. 26.</ref>

== Smoke control and critical velocity ==
Smoke can be stopped from flowing through an opening, such as a doorway or a corridor, by an airflow against it.<ref name="Klo-279" /> One criterion for the design air velocity is that no smoke flows back during the evacuation.<ref name="Klo-286">Klote (2002), p. 4-286.</ref> If the airflow is too weak, smoke flows back against it, a phenomenon known as [[backlayering]].<ref name="Klo-278">Klote (2002), p. 4-278.</ref> The lowest air velocity that prevents this is the [[critical velocity]]. Thomas developed an empirical relation for this velocity in a corridor, which Klote gives in a simplified form for normal air properties:<ref name="Klo-279" />

[[File:fse-tunnel-fire-v1.svg|thumb|upright=1.5|alt=Two longitudinal tunnel sections with a burning vehicle: weak airflow lets smoke flow back upstream; stronger airflow keeps all smoke downstream.|Airflow below the critical velocity (top): smoke flows back upstream against the air. Airflow at or above the critical velocity (bottom): the upstream side stays free of smoke. Based on Klote (2002), pp. 4-278–4-279.]]

<math display="block">V_k = 0.292 \left( \frac{\dot{Q}}{W} \right)^{1/3}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>V_k</math> || critical air velocity to prevent smoke backflow || m/s
|-
| <math>\dot{Q}</math> || heat release rate of the fire || kW
|-
| <math>W</math> || width of the corridor || m
|}

The coefficient 0.292 follows from an air density of 1.3 kg/m³, a specific heat of 1.005 kJ/(kg·K), a downstream temperature of 27 °C and a constant of 1. The results are approximate, because the constant is known only approximately, but they show the size of the air velocity needed for different fires.<ref name="Klo-279" /> The relation was developed for corridors. Klote's design chart covers widths of 1 to 4 m and fires of 500 to 2,000 kW, far below the tens of megawatts of a truck fire.<ref name="Klo-280">Klote (2002), p. 4-280.</ref> It is not appropriate for sprinklered fires in which the smoke is barely warmer than the air.<ref name="Klo-279" /> For tunnels it therefore illustrates the principle rather than giving design values.

Klote's worked example is a wastebasket fire of 150 kW in a corridor 1.22 m wide. The relation gives 0.292 × (150 / 1.22)1/3 = 0.292 × 4.97 = 1.45 m/s.<ref name="Klo-280" /> Doubling the heat release rate raises the critical velocity by only 26 %, because of the cube root. Within the range of the chart, a fire of 2,000 kW in a corridor 4 m wide needs 2.32 m/s. The cube root explains why larger fires need only moderately higher air speeds.

[[File:fse-tunnel-fire-v2.svg|thumb|upright=1.5|alt=Line chart of critical velocity rising slowly with heat release rate from 150 to 2,000 kW for widths of 1, 2 and 4 m.|Critical velocity from Thomas's corridor relation for three widths, within the range of Klote's design chart. Based on Klote (2002), pp. 4-279–4-280.]]

The airflow has a side effect: it brings oxygen to the fire. One cubic metre of air per second at 21 °C contains enough oxygen for a fire of 3,660 kW. In the worked example, the airflow of 4.3 m³/s that stops the smoke could support a fire of 15,800 kW, more than 100 times the wastebasket fire, so Klote advises extreme caution.<ref name="Klo-279" /><ref name="Klo-280" /> Strong ventilation also affects suppression and detection, as an example from a cable tunnel in the next section shows.<ref name="MB-328">Mawhinney and Back (2002), p. 4-328.</ref>

The same principle is used in some Dutch car parks. Displacement ventilation creates a high air speed over the whole cross-section, for example 1.5 m/s according to NEN 6098, which can easily mean 40 air changes per hour. It gives firefighters a clear view of the fire, but it works only in tunnel-shaped car parks and usually starts only after people have escaped.<ref name="VdG-38">Van der Graaf et al. (2021/2023), p. 38.</ref>

== Fire exposure of the tunnel structure ==
A tunnel lining must resist fires that are more severe than the standard fire of a building test. The book names two less common temperature–time curves that are more severe than the standard fire curve. One is the Rijkswaterstaat curve for tunnels, known as the [[RWS fire curve]]; the other is the hydrocarbon curve, which simulates an oil or gas fire.<ref name="VR-8.6" /> Fire curves are agreed temperature histories used to compare tested constructions; they do not reproduce the heat development of a real fire.<ref name="VR-8.6" /> Most furnace tests simulate a cellulosic fire, which grows more slowly than a hydrocarbon fire. In the United States, exposure ratings for hydrocarbon fires are given as H ratings under UL 1709.<ref name="IS-C-8">Iqbal and Salley (2004), p. C-8.</ref><ref name="IS-C-9">Iqbal and Salley (2004), p. C-9.</ref>

Concrete linings can suffer from [[spalling]], in which layers or pieces of concrete break off the heated surface, sometimes violently. Spalling is caused by a build-up of pore pressure from water vapour during heating, although some researchers attribute it to restrained thermal expansion near the surface. Extensive spalling can lead to early loss of stability and integrity, because it exposes the reinforcement.<ref name="KH-163">Kodur and Harmathy (2002), p. 1-163.</ref>

[[File:fse-tunnel-fire-v4.svg|thumb|upright=1.5|alt=Two cross-sections of a heated concrete lining: left, pieces break off and expose the reinforcement; right, melted fibres leave channels for steam.|Spalling of a heated concrete lining (left) and the effect of polypropylene fibres that melt at about 170 °C and let steam escape (right). Based on Kodur and Harmathy (2002), pp. 1-163 and 1-171.]]

The risk grows with a high moisture content, rapid heating, slender members and high stresses during the fire. Most members of normal-weight concrete withstand severe fires without spalling, but high-strength concrete of 50 to 120 MPa is more susceptible.<ref name="FB-255">Fleischmann and Buchanan (2002), p. 4-255.</ref> Design methods for concrete assume that the concrete stays intact, which no longer holds if the cover spalls off.<ref name="FB-255" />

A common remedy is to add polypropylene fibres to the concrete. The fibres melt at about 170 °C and leave channels through which steam can escape. Kodur and Harmathy give about 0.1 to 0.25 % by volume as the amount needed.<ref name="KH-171">Kodur and Harmathy (2002), p. 1-171.</ref> The International Fire Engineering Guidelines, citing Kodur, give 0.15 to 0.3 % without stating whether this is by volume or by mass.<ref name="IFEG-2.6-9">Kodur (1997), as cited in National Research Council of Canada et al. (2005), p. 2.6-9.</ref>

== Suppression and detection ==
A [[water mist system]] uses fine sprays in which no drop is larger than 1.0 mm.<ref name="MB-311">Mawhinney and Back (2002), p. 4-311.</ref> Test programmes for water mist have been carried out for high-voltage cable tunnels and for railway tunnels, including full-scale suppression tests for Eurotunnel and an onboard system for its shuttle trains for heavy goods vehicles.<ref name="MB-325">Mawhinney and Back (2002), p. 4-325.</ref> In Germany, VdS has test protocols for local application systems in electric cable tunnels.<ref name="MB-328" />

A cable tunnel of 9 km in Auckland shows the practical problems. Opening all nozzles along the tunnel at once would need very large water mains and pumps, and costly pumping out afterwards. The chosen solution released water in zones upstream and downstream of the fire.<ref name="MB-328" /> The ventilation air of 5 m/s would carry the heat along the tunnel, away from the nozzles closest to the fire, so individually heat-activated nozzles would not work. Turbulence under normal ventilation also made smoke and heat detection difficult. The detection system and the hydraulic system therefore had to be tested together.<ref name="MB-328" />

Vehicles with lithium-ion batteries raise further questions. A sprinkler or water mist system cannot extinguish a fire in an electric car, because the water droplets cannot reach the cells inside the closed battery pack. It can, however, reduce the chance that the fire grows and spreads to other vehicles and the structure.<ref name="VdG-37">Van der Graaf et al. (2021/2023), p. 37.</ref> For firefighting with hoses, the NIPV report gives two rules of thumb. A high-pressure jet of 125 L/min is generally sufficient for fires of up to 2.5 MW, and a low-pressure jet of 450 L/min for fires of up to 10 MW.<ref name="VdG-45">Van der Graaf et al. (2021/2023), p. 45.</ref> A truck fire of 120 MW is far beyond these values.<ref name="Bab-28" />

== Modelling ==
[[Computational fluid dynamics]] can be used to study how effective the smoke management system of a tunnel is.<ref name="VR-11.2" /> The book warns that the fire and its heat release rate must be prescribed by the modeller and that CFD is mainly reliable for qualitative comparisons of situations.<ref name="VR-11.4">Van Rossum (2027), §11.4.</ref> Cox and Kumar describe the King's Cross underground station fire, where CFD predicted that the flames would lie down in the trench of the escalator instead of rising. Physical tests confirmed this, and the authors stress that simple calculations should always shadow such simulations.<ref name="CK-207">Cox and Kumar (2002), p. 3-207.</ref>

Tunnels are numerically demanding. The Fire Dynamics Simulator notes that its standard pressure solver is excessively slow for tunnels made of dozens of computational meshes placed end to end, and offers an optional method for such geometries.<ref name="McG-43">McGrattan et al. (2026), p. 43.</ref> Its documentation also refers to work that couples a three-dimensional model of the fire region with a one-dimensional network model of the ventilation of a tunnel.<ref name="McG-111">McGrattan et al. (2026), p. 111.</ref> An older network program, the Subway Environment Simulation from the early 1970s, is used to size and locate ventilation shafts, evaluate tunnel geometry and fan size, and model heat and smoke from fires.<ref name="IS-B-83">Iqbal and Salley (2004), p. B-83.</ref>

== Limitations ==
The critical velocity relation was developed for corridors and covers fires up to 2,000 kW. The vehicle data come from a limited number of tests, and the book names the RWS fire curve without giving its values.<ref name="Klo-280" /><ref name="Bab-27" /><ref name="VR-8.6" /> Fire curves describe a test exposure, not a real fire, and CFD results depend on the prescribed fire.<ref name="VR-8.6" /><ref name="VR-11.4" /> A common misconception is that a tunnel fire can be treated as a room fire. The long geometry, the limited escape routes and the effect of ventilation on both smoke and fire make it a different problem.<ref name="Buk-229" /><ref name="Klo-279" />

== See also ==
* [[Car park fire]]
* [[Thermal runaway]]
* [[Lithium-ion battery fire]]

== Notes ==
<references />

== References ==
* Babrauskas, V. (2002b). Heat release rates. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-1–3-37). National Fire Protection Association; Society of Fire Protection Engineers.
* Bukowski, R. W. (2002). Fire hazard assessment for transportation vehicles. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-227–5-233). National Fire Protection Association; Society of Fire Protection Engineers.
* Cox, G., & Kumar, S. (2002). Modeling enclosure fires using CFD. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-194–3-218). National Fire Protection Association; Society of Fire Protection Engineers.
* Fleischmann, C., & Buchanan, A. (2002). Analytical methods for determining fire resistance of concrete members. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-239–4-256). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Kodur, V. K. R., & Harmathy, T. Z. (2002). Properties of building materials. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-155–1-181). National Fire Protection Association; Society of Fire Protection Engineers.
* Mawhinney, J. R., & Back, G. G., III. (2002). Water mist fire suppression systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-311–4-337). National Fire Protection Association; Society of Fire Protection Engineers.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Facade fire

2026-11-02T17:15:07Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''facade fire''' is a fire that spreads over or within the external wall of a building, usually upwards from storey to storey. New Zealand's Verification Method C/VM2 describes the scenario as a fire source that exposes the external wall and leads to significant vertical fire spread.<ref name="MBIE-47">Ministry of Business, Innovation and Employment (2025b), p. 47.</ref> The fire source can be a fire inside the building whose flames project out of a window, or a fire outside, such as a burning rubbish skip against the wall.<ref name="MBIE-48">Ministry of Business, Innovation and Employment (2025b), p. 48.</ref>

Two routes of spread are distinguished. In the first, flames leaving a window heat the wall above and enter the building again through openings on a higher floor. In the second, the external wall itself takes part in the fire, when combustible cladding or cavities in the wall carry the flames upwards.<ref name="MBIE13-134">Ministry of Business, Innovation and Employment (2013), p. 134.</ref><ref name="ABCB-81">Australian Building Codes Board (2020b), p. 81.</ref> Facade fires threaten escape routes, firefighters and neighbouring property, and can involve falling debris.<ref name="MBIE-47" /><ref name="ABCB-81" /> Codes such as C/VM2 therefore limit the combustibility of external wall systems and the vertical distance a fire may travel, which C/VM2 sets at 3.5 m.<ref name="MBIE-47" /><ref name="MBIE-48" />

== How a facade fire starts ==
One source of a facade fire is a fully developed fire inside a room. After flashover the fire is usually ventilation-controlled: there is more pyrolysis gas than the incoming air can burn, so part of the gas burns outside the room in flames from the openings.<ref name="VR-4.11">Van Rossum (2027), §4.11.</ref> This [[external flaming]] is typical of post-flashover fires, and the flames rise along the outside of the building because of buoyancy.<ref name="Lat-292">Lattimer (2002), p. 2-292.</ref> Wind speed and direction can change the extent of the flames projecting from windows, especially when there are openings on both the windward and leeward sides.<ref name="IFEG-1.6-5">National Research Council of Canada et al. (2005), p. 1.6-5.</ref>

The heat from these flames can be severe. Experiments on full-scale rooms with a wall extending up to two storeys above the window measured heat fluxes of up to 200 kW/m² on the wall above.<ref name="Lat-292" /> In Oleszkiewicz's tests with a window 2.6 m wide and 1.37 m high, the heat flux 0.5 m above the window was roughly 210 kW/m² for an ideal fire size of 10.3 MW. For a fire of 5.5 MW it was only about 25 kW/m², as read from the published graph. The heat flux fell with height, to roughly 55 kW/m² at 3.5 m above the window for the largest fire.<ref name="Lat-293">Lattimer (2002), p. 2-293.</ref>

[[File:fse-facade-fire-v2.svg|thumb|upright=1.5|alt=Chart of heat flux against height above a window for four fire sizes; about 210 kW/m² at 0.5 m for 10.3 MW, falling with height.|Heat flux on the wall above a 2.6 m by 1.37 m window for four fire sizes in the room, from Oleszkiewicz's tests. Based on Lattimer (2002), p. 2-293.]]

The flame length above a window depends on the heat release rate of the part of the fire that burns outside and on the size of the window. A relation by Yokoi describes this and is treated under external flaming.<ref name="Lat-293" /> Fires can also start outside the building. C/VM2 names a burning rubbish container or skip in close contact with the facade as a design fire source that could ignite the wall and spread fire upwards.<ref name="MBIE-48" />

== Routes of vertical spread ==
The first route runs from window to window. Flames from an opening heat the wall and the openings above, and the fire can enter the next floor even if the wall itself does not burn. Platt's probabilistic fire spread model expresses this route simply. The probability that fire spreads up the facade via windows is taken as the probability that the external flame is at least as high as the spandrel.<ref name="Ram-392">Ramachandran (2002c), p. 3-392.</ref> In New Zealand, C/VM2 (second edition, 2025) aims to prevent projecting fire plumes in unsprinklered buildings from reaching unprotected areas on upper floors within 1.5 m vertically of the plume source.<ref name="MBIE-49">Ministry of Business, Innovation and Employment (2025b), p. 49.</ref>

The second route runs over or through the wall. Flame spread over a surface is much faster upwards than sideways or downwards. Typical orders of magnitude are about 0.1 cm/s for lateral or downward spread on thick solids and 1 to 100 cm/s for upward spread.<ref name="Qui-247">Quintiere (2002b), p. 2-247.</ref> In [[upward flame spread]], the flames and the buoyant flow heat the material ahead of the burning zone, and this wind-aided spread can accelerate.<ref name="Qui-251">Quintiere (2002b), p. 2-251.</ref> Low-density materials are a particular concern, because for a given flame heat flux the spread rate increases as the density of the fuel decreases.<ref name="Qui-247" /> A material with a low thermal inertia, such as lightweight insulation, heats up quickly at the surface; if it is combustible, this contributes to rapid ignition.<ref name="VR-5.2">Van Rossum (2027), §5.2.</ref>

Hidden spaces form a third path. Concealed spaces behind exterior cladding on the inside of facades are treated as enclosures when fire spread is analysed.<ref name="IFEG-1.6-6">National Research Council of Canada et al. (2005), p. 1.6-6.</ref> C/VM2 therefore requires [[Cavity barrier|cavity barriers]] at each floor level to stop fire spreading through cavities in an external wall.<ref name="MBIE-48" /> A cavity barrier is a construction that closes openings within a concealed space against the passage of fire or restricts fire spread within such spaces.<ref name="MBIE-62">Ministry of Business, Innovation and Employment (2025b), p. 62.</ref> Fire can also travel from a low building to a higher wall. C/VM2 covers the case of a fire breaking through a lower roof that is not fire-rated and exposing a higher external wall next to it.<ref name="MBIE-49" />

[[File:fse-facade-fire-v1.svg|thumb|upright=1.5|alt=Section through three storeys: flames from a window rise past a spandrel to the window above; fire also runs up cladding and its cavity to a barrier.|Routes of vertical fire spread on a facade: flames from an opening re-entering a higher floor (left) and fire spread over cladding and through the cavity behind it, interrupted by a cavity barrier (right). Based on Ministry of Business, Innovation and Employment (2025b), pp. 48–49, and Ramachandran (2002c), p. 3-392.]]

== Examples from high-rise fires ==
Two office fires in the United States show the window-to-window route on a large scale. On 4 May 1988, a fire in the 62-storey First Interstate Bank Building in Los Angeles started on the 12th floor and burned out four floors. The fire spread upwards mostly as flames extended about 9 m (30 ft) up the face of the building and through the gap between the floors and the glass curtain wall.<ref name="McA-334">McAllister et al. (2008), pp. 334–335.</ref> Some spread also occurred through floor penetrations and shafts. The fire moved up at about 45 minutes per floor and burned intensely for about 90 minutes on each level, so that two floors were heavily involved at any time.<ref name="McA-334" />

On 23 February 1991, a fire on the 22nd floor of One Meridian Plaza, a 38-storey office building in Philadelphia, eventually consumed the 21st to 29th floors. Vertical spread came from unprotected openings in floors and shafts, failing fire-rated floors and flames lapping through windows on the outside of the building.<ref name="McA-331">McAllister et al. (2008), pp. 331–333.</ref> The fire climbed at about 90 minutes per floor and was stopped only at the 30th floor, which had working sprinklers.<ref name="McA-331" /> NIST concluded that in both buildings vertical fire spread was by external flame exposure. The flames broke the windows from the outside and ignited the office furnishings on the floor above.<ref name="McA-341">McAllister et al. (2008), p. 341.</ref>

== Materials and wall systems ==
The fire behaviour of a facade depends on its materials and on how they are put together. C/VM2 (second edition, 2025) requires the entire external wall cladding system to consist of non-combustible or limited combustible materials, or to meet the acceptance criteria of a recognised test or classification method.<ref name="MBIE-48" /> In C/VM2, a combustible material is one that is neither non-combustible nor limited combustible.<ref name="MBIE-62" /> [[Combustibility]] can be assessed with tests such as AS 1530 Part 1 or ISO 1182, and brick, concrete and steel count as non-combustible.<ref name="MBIE13-133">Ministry of Business, Innovation and Employment (2013), p. 133.</ref>

For combustible claddings, the first edition of C/VM2 used small-scale tests in the cone calorimeter. Samples were exposed to 50 kW/m² for 15 minutes, and the cladding was accepted if its peak heat release rate stayed at or below 100 kW/m².<ref name="MBIE13-135">Ministry of Business, Innovation and Employment (2013), p. 135.</ref> This limit was chosen to give a low probability of accelerating flame spread over the surface. The commentary warns that small-scale tests are not suitable for all materials, particularly where the fire behaviour depends on connections or joints.<ref name="MBIE13-135" /> A table in the first edition set two performance levels. Level A allowed a peak heat release rate of 100 kW/m² and a total of 25 MJ/m², and level B 150 kW/m² and 50 MJ/m².<ref name="MBIE13-137A">Ministry of Business, Innovation and Employment (2013), p. 137A.</ref>

Composite products show why the complete construction matters. In corner fire tests on combustible linings, one of the materials was an 88 mm [[Sandwich panel|sandwich composite]] with a balsa wood core and glass-fibre facings. The heat fluxes measured in the corner were slightly higher than predicted by correlations derived with non-combustible walls.<ref name="Lat-291">Lattimer (2002), pp. 2-291–2-292.</ref> For products whose fire performance depends on connections or joints, the C/VM2 commentary advises larger-scale tests instead of small-scale tests.<ref name="MBIE13-135" />

== Testing facades ==
Because the behaviour depends on the complete system, facades are often tested at large or medium scale. The first edition of C/VM2 accepted a [[large-scale facade fire test]] such as NFPA 285, ISO 13785-1 or the vertical channel test. The test had to show that vertical flame spread stays within 3.5 m above the fire source.<ref name="MBIE13-136">Ministry of Business, Innovation and Employment (2013), p. 136.</ref> Current large-scale facade tests expose the wall to heat fluxes in the range of 50 to 90 kW/m².<ref name="MBIE13-135" /> The second edition no longer names test standards such as NFPA 285 and BS 8414; it refers instead to a separate document, the Building Product Specifications.<ref name="MBIE-3">Ministry of Business, Innovation and Employment (2025b), p. 3.</ref>

Test facilities reproduce the window fire described above. Oleszkiewicz's facility consisted of a burn room with a window and a target wall above it, next to a vertical channel test apparatus.<ref name="Lat-293" /> An intermediate-scale calorimeter developed in Canada for facade systems exposes specimens of 1.22 by 2.08 m to a radiant heat flux of up to 30 kW/m² and measures their heat release rate.<ref name="Jan-57">Janssens (2002), p. 3-57.</ref> Fire resistance tests of facade elements use a separate temperature curve. The external fire exposure curve represents a fire in the open air; it levels off at about 680 °C, whereas the standard curve keeps rising.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref>

== Design and regulation ==
In New Zealand, C/VM2 (second edition, 2025) contains design scenario VS, external vertical fire spread. Building Code clause C3.5 requires that fire does not spread more than 3.5 m vertically in buildings taller than 10 m, or in buildings with upper floors that contain sleeping uses or other property. This aims to keep escape routes tenable until occupants have left and to protect firefighters working in or around the building.<ref name="MBIE-47" /> The scenario applies to multi-level buildings of more than 10 m and to other multi-level buildings with sleeping uses, other property or external escape routes on upper floors.<ref name="MBIE-47" />

The design fire for the facade materials is a radiant heat flux of 50 kW/m² on the facade for 15 minutes for buildings of importance levels 2 and 3. For importance level 4 the flux is 90 kW/m² for 15 minutes.<ref name="MBIE-48" /> Some damage to the exposed area is accepted, but the cladding must not spread flames significantly beyond it.<ref name="MBIE-48" /> According to the commentary on the first edition, the 3.5 m limit accepts that fire may reach the floor immediately above but not two floors above. The value of 50 kW/m² represents the heat flux from a fire plume projecting from an opening, although higher fluxes are possible.<ref name="MBIE13-136" /><ref name="MBIE13-135" />

For spread via openings, C/VM2 offers three options in unsprinklered buildings: an automatic sprinkler system, fire-rated construction such as aprons and spandrels, or calculations of radiation from the projecting fire plumes.<ref name="MBIE-49" /> In a sprinklered building this route is addressed by reducing the risk of flashover, and external drenchers are not needed.<ref name="MBIE13-135" /> For a lower roof next to a higher wall, the options are sprinklers below the roof or fire-rating the underside of the roof within 5.0 m of the wall. A third option is fire-rating the wall up to 9.0 m above the unprotected roof.<ref name="MBIE-49" />

[[File:fse-facade-fire-v3.svg|thumb|upright=1.5|alt=Three columns: facade materials, openings and lower roof, each with the C/VM2 exposure and the accepted means of compliance.|The three parts of design scenario VS in C/VM2 (second edition, 2025) and their means of compliance. Based on Ministry of Business, Innovation and Employment (2025b), pp. 47–49.]]

In Australia, the Fire Safety Verification Method includes a similar scenario VS. It compares the risk to life from a fire affecting the external wall, including penetrations, cladding and attachments, with that of a reference building to the deemed-to-satisfy provisions.<ref name="ABCB-80">Australian Building Codes Board (2020b), p. 80.</ref><ref name="ABCB22-14">Australian Building Codes Board (2022), p. 14.</ref> The handbook refers to Verification Method CV3 and to AS 5113. It notes that an external wall system tested to AS 5113 may still contain combustible elements whose risk needs further mitigation.<ref name="ABCB-80" /> Typical measures listed include control of combustibility, distances between vertical openings, non-combustible and fire-rated spandrels, horizontal projections from the facade and enhanced sprinkler protection.<ref name="ABCB-81" />

== Consequences and limits of protection ==
A facade fire can reach several floors quickly, which matters for the evacuation strategy of a [[high-rise building]]. The Australian data sheets note that phased evacuation, in which floors leave one after another, may be sensitive to rapid fire spread to several floors via the facade. The evacuation management strategy should describe what to do in that case.<ref name="ABCB-250">Australian Building Codes Board (2020a), p. 250.</ref> The criteria of C/VM2 likewise refer to tenable escape routes and to the safety of firefighters working in or around the building.<ref name="MBIE-47" />

Facade fires also threaten the surroundings. The Australian scenario names falling debris and spread to adjacent buildings as consequences of fire spread over the external wall.<ref name="ABCB-81" /> Fire spread between separate buildings across a boundary is, however, a different design scenario. Both C/VM2 and the Australian method treat it separately from vertical spread.<ref name="MBIE-47" /><ref name="ABCB22-14" />

Fire resistance and facade fire spread are different properties. The fire resistance of building elements is tested with a standard temperature curve.<ref name="VR-8.6" /> The facade scenarios, in contrast, address the combustibility of the whole cladding system, including penetrations, attachments and cavities, and the flames from openings.<ref name="MBIE-48" /><ref name="ABCB-80" /> Sprinklers inside the building protect the facade only indirectly, by reducing the risk of flashover in the room of origin.<ref name="MBIE13-135" />

== See also ==
* [[Fire spread]]
* [[Fire spread between buildings]]
* [[Stay-put strategy]]
* [[Glass breakage in fire]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* Janssens, M. (2002). Calorimetry. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-38–3-62). National Fire Protection Association; Society of Fire Protection Engineers.
* Lattimer, B. Y. (2002). Heat fluxes from fires to surfaces. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-269–2-296). National Fire Protection Association; Society of Fire Protection Engineers.
* McAllister, T. P., Gann, R. G., Averill, J. D., Gross, J. L., Grosshandler, W. L., Lawson, J. R., McGrattan, K. B., Pitts, W. M., Prasad, K. R., Sadek, F. H., & Nelson, H. E. (2008). ''Structural fire response and probable collapse sequence of World Trade Center Building 7'' (NIST NCSTAR 1-9, Vols. 1–2). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Quintiere, J. G. (2002b). Surface flame spread. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-246–2-257). National Fire Protection Association; Society of Fire Protection Engineers.
* Ramachandran, G. (2002c). Stochastic models of fire growth. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-381–3-401). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Thermal runaway

2026-11-02T17:15:05Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Thermal runaway''' is the uncontrolled rise in temperature of a reacting material or device when the heat produced by its own chemical reactions exceeds the heat it can lose to its surroundings. The higher temperature speeds up the reactions, which produce still more heat, until the material ignites or its reactive content is used up.<ref name="Gray-212">Gray (2002), p. 2-212.</ref><ref name="Kan-196">Kanury (2002b), p. 2-196.</ref> In the SFPE Handbook the term is used as a synonym of ''runaway reaction'', originally for chemical reactors.<ref name="Gray-212" /> Today it is best known as the failure mechanism behind a [[lithium-ion battery fire]], in which a battery cell heats itself and can set fire to its neighbours.<ref name="Col-22">Colella et al. (2016), as cited in Van der Graaf et al. (2021/2023), p. 22.</ref><ref name="VdG-22">Van der Graaf et al. (2021/2023), p. 22.</ref>

The same physics explains the [[self-heating]] of haystacks, oily rags and coal piles, so the general theory of self-heating is the starting point for understanding batteries.<ref name="Gray-211">Gray (2002), p. 2-211.</ref><ref name="Gray-223">Gray (2002), p. 2-223.</ref> In an electric vehicle, thermal runaway can start at temperatures of about 70 to 250 °C, depending on the type of battery.<ref name="Sun-22">Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 22.</ref> It releases flammable and toxic gases and can make the fire reignite until the battery's energy is used up.<ref name="VdG-22" /> The peak heat release rate of an [[electric vehicle fire]] is similar to that of a petrol or diesel car. Its slow start, long duration and the difficulty of extinguishing it are what set it apart.<ref name="Sun-25">Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 25.</ref><ref name="VdG-26">Van der Graaf et al. (2021/2023), p. 26.</ref><ref name="VdG-30">Van der Graaf et al. (2021/2023), p. 30.</ref>

== Definition ==
Gray describes spontaneous combustion as the reaction of an unstable, usually oxidisable, material that releases heat. Much of this heat stays inside the material because it, or its container, conducts heat poorly. If the process ends in flaming, it is called spontaneous ignition.<ref name="Gray-211" />

Mild self-heating that settles at a slightly raised temperature is called ''subcritical''. Self-heating that ends in ignition is called ''supercritical'', and the sharp boundary between the two is the ''critical condition''.<ref name="Gray-211" /> Runaway reaction and thermal runaway are the names used for supercritical behaviour in a chemical reactor, which is almost always well stirred.<ref name="Gray-212" /> Other chapters of the SFPE Handbook use the term in the same sense. In the thermal theory of autoignition, ignition is taken as the instant at which the temperature in the reacting mixture rises rapidly. This event is called thermal runaway and also marks the start of a rapid consumption of the reactants.<ref name="Kan-196" />

In the theory of heat conduction, thermal runaway is the case in which the temperature of a heat-generating solid grows exponentially with time and no steady state exists.<ref name="RM-36">Rockett and Milke (2002), p. 1-36.</ref> For batteries, a report by the Dutch Institute for Public Safety (NIPV) on car parks quotes a definition by Colella and co-workers. In that definition, thermal runaway is a failure mechanism that leads to self-heating in a battery or battery cell and that can result in fire.<ref name="Col-22" /> The NIPV report notes that thermal runaway and the burning of the released gases are different processes, and that the link between them is not yet well understood.<ref name="VdG-55">Van der Graaf et al. (2021/2023), p. 55.</ref>

== Heat generation and heat loss ==
The mechanism is a competition between two heat flows. Most chemical reaction rates increase almost exponentially with temperature, while heat loss by conduction increases only linearly. Once the heat production grows faster than the loss, the temperature can only rise.<ref name="Gray-212" /> A larger body and a higher ambient temperature both reduce the relative heat loss and so favour ignition. Inside a self-heating body the temperature profile is roughly parabolic, so runaway starts at the centre, where the temperature is highest.<ref name="Gray-212" />

Semenov's theory, the simplest description, writes this competition as a heat balance for a body with a uniform temperature. The body stores heat according to its heat capacity, gains heat from the reaction in its whole volume and loses heat through its surface. The rate of change of the stored heat equals the heat generated minus the heat lost:<ref name="Gray-216">Gray (2002), p. 2-216.</ref>

<math display="block">C_v \rho V \frac{dT}{dt} = V Q f(c)\, e^{-E/RT} - S \chi \left( T - T_a \right)</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>C_v</math> || heat capacity at constant volume, per unit mass || J/(kg·K)
|-
| <math>\rho</math> || density of the material || kg/m³
|-
| <math>V</math> || volume of the body || m³
|-
| <math>T</math> || temperature of the reacting material || K
|-
| <math>t</math> || time || s
|-
| <math>Q</math> || heat of reaction per unit concentration of fuel || J/mol
|-
| <math>f(c)</math> || kinetic rate law, a function of the fuel concentration <math>c</math> || mol/(m³·s)
|-
| <math>E</math> || activation energy of the reaction || J/mol
|-
| <math>R</math> || universal gas constant, 8.314 || J/(mol·K)
|-
| <math>S</math> || surface area across which heat is lost || m²
|-
| <math>\chi</math> || heat transfer coefficient || W/(m²·K)
|-
| <math>T_a</math> || ambient temperature, constant in time || K
|}

The first term on the right is the heat generated by the self-heating reaction, and the second is the heat lost to the surroundings. Their difference drives the change in temperature.<ref name="Gray-216" /> The equation cannot be solved by classical methods, but the critical condition can be found from it exactly.<ref name="Gray-216" />

The relation assumes a uniform temperature, a single reaction of simple order, and a heat of reaction and activation energy large enough to give ignition behaviour. It suits well-stirred liquids and gases. It is a poor approximation for poorly conducting materials such as cellulose, with a conductivity of about 0.05 W/(m·K), although it still gives semi-quantitative results there.<ref name="Gray-215">Gray (2002), p. 2-215.</ref><ref name="Gray-216" /> It is a general model of self-heating and not a model of a battery.

The balance is easiest to understand as two curves of heat flow against temperature, the Semenov diagram. At a low ambient temperature the straight heat-loss line cuts the curved heat-generation curve twice. The lower crossing is a stable state, usually 5 to 20 °C above ambient, in which the material may degrade or discolour but does not ignite.<ref name="Gray-214">Gray (2002), p. 2-214.</ref>

[[File:fse-thermal-runaway-v1.svg|thumb|upright=1.5|alt=Semenov diagram: a steeply rising heat generation curve and three parallel straight heat loss lines for low, critical and high ambient temperature.|Semenov diagram: heat generated by the reaction (curve) and heat lost to the surroundings (straight lines) for three ambient temperatures. Based on Gray (2002), pp. 2-213–2-214.]]

At the critical ambient temperature the line just touches the curve. Above it there is no balance point at all, and the temperature rises until ignition, at a rate proportional to the imbalance between heat production and loss.<ref name="Gray-214" /> The slope of the loss line depends on the ratio of surface area to volume. For a given ambient temperature there is therefore also a critical size, above which a body will run away.<ref name="Gray-214" />

A worked example shows why the critical point lies close to ambient conditions. Many spontaneous combustion reactions, in particular those of cellulosic materials, have activation energies of around 100 kJ/mol.<ref name="Gray-221">Gray (2002), p. 2-221.</ref> Gray shows that criticality exists only if the activation energy is at least four times the product of the gas constant and the critical ambient temperature.<ref name="Gray-217">Gray (2002), p. 2-217.</ref> At an ambient temperature of 30 °C (303 K) this product is 8.314 × 303 = 2.52 kJ/mol, so four times it is 10.1 kJ/mol. The activation energy of 100 kJ/mol is about ten times larger, so the reaction accelerates strongly enough for a sharp critical condition.

The ratio of the gas constant times the squared temperature to the activation energy is 8.314 × 303² / 100,000 = 7.6 K. Kanury points out that this ratio is generally small, which is why the critical temperature of a reacting mass lies close to the temperature of its surroundings.<ref name="Kan-196" /> As a rule of thumb for materials with an unknown activation energy, self-heating of more than 30 °C above ambient, typically a body temperature above 60 to 70 °C, indicates imminent spontaneous ignition.<ref name="Gray-217" />

== Size, storage and time to ignition ==
Because heat loss depends on surface area and heat generation on volume, size matters as much as temperature. Bodies placed in contact act as one larger body. Two bales that touch have a considerably lower critical ambient temperature than a single bale, so a test on one item says little about a stack unless a theory is used to scale it up.<ref name="Gray-215" /> Bodies that are not in contact can still interact by heating the air around them. In tests with kegs of 14 kg in still air, the critical ambient temperature fell from 62.5 °C for a single keg to 54 °C for 18 kegs packed in a steel box.<ref name="Gray-223" />

The time to ignition is very hard to predict. It becomes infinite at the critical condition and is extremely sensitive to conditions close to it.<ref name="Gray-223" /> Larger bodies generally take longer: oily rags in a wastepaper basket may ignite within a few hours, haystacks after a few weeks and coal stockpiles after a few months. Bodies that are stacked while still hot ignite much sooner.<ref name="Gray-223" /><ref name="Gray-224">Gray (2002), p. 2-224.</ref> Deep-seated burning that starts by spontaneous ignition is difficult to extinguish completely and often reignites days after it seemed to be out.<ref name="Gray-212" />

The International Fire Engineering Guidelines note that self-heating depends not only on the chemical nature of a substance but also on the size and shape of the aggregation.<ref name="IFEG-3.3-2">National Research Council of Canada et al. (2005), p. 3.3-2.</ref> In continuously stirred chemical reactors, the scale-up from a prototype to a commercial size has led to the critical condition being exceeded.<ref name="Gray-219">Gray (2002), p. 2-219.</ref> The same lesson applies to stacks and stores: the critical condition belongs to the whole assembly, not to a single item.<ref name="Gray-215" />

== Thermal runaway in lithium-ion batteries ==
A battery pack in an electric vehicle is built up in steps. Cells deliver about 4 V, cells are combined into modules of up to 60 V, and modules are connected into packs of 300 to 1,000 V. Packs in passenger cars can store 100 kWh or more.<ref name="VdG-21">Van der Graaf et al. (2021/2023), p. 21.</ref> Packs can also be connected into larger battery systems of 10 to 400 kWh, for example for trucks and electric ships.<ref name="VdG-21" /> Stationary installations of this kind are described under [[battery energy storage system]].

Each battery pack has its own safety window. If it is exceeded, the pack can heat up further and thermal runaway can occur.<ref name="ARB-22">Air Resources Board (2015), as cited in Van der Graaf et al. (2021/2023), p. 22.</ref> The onset temperature depends on the type of battery and ranges from 70 to 250 °C.<ref name="Sun-22" /> Elsewhere the NIPV report gives a range of 50 to 270 °C, and it stresses that "lithium-ion" covers many battery types with different fire behaviour.<ref name="VdG-18">Van der Graaf et al. (2021/2023), p. 18.</ref> A battery management system monitors the cells. When a threshold is exceeded it can apply active cooling during charging, stop the charging or disconnect the whole pack, and a fault in the system itself also switches the pack off.<ref name="VdG-23">Van der Graaf et al. (2021/2023), p. 23.</ref>

The causes can be internal or external. Internal causes are rare design or manufacturing errors in the electrode, separator or electrolyte, which can lead to failures of the management system and to overcharging; such errors usually lead to recalls. External causes include outside heat, misuse and extreme conditions such as a collision that damages the battery.<ref name="VdG-23" /> In car parks, damaged charging cables or stations, vehicles that hit a charging station and vehicles with an unnoticed defective pack add to these causes.<ref name="Sun-29">Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 29.</ref>

Published failure probabilities differ widely, from one in a million or lower to about one in 100 million.<ref name="Lar-23">Larsson et al. (2017) and Bullis (2013), as cited in Van der Graaf et al. (2021/2023), p. 23.</ref> Overall the probability of battery failure seems relatively low. The NIPV report adds, however, that no lithium-ion battery is intrinsically safe. The design of a pack cannot completely prevent the release of enough energy for thermal runaway.<ref name="VdG-23" />

Once a cell runs away, the process can spread. Heat from one cell warms the adjacent cells, which then also go into thermal runaway, in a chain reaction through the pack.<ref name="VdG-22" /> As the battery heats up it releases flammable gases, which ignite if enough oxygen is present and the ignition temperature is reached; the hot cell housing may be enough. Gases trapped in a pouch or cartridge can escape when the enclosure or a pressure valve fails. In the right mixture with air they can cause a flare fire or a gas cloud [[explosion]], and mechanical damage can throw out burning fragments.<ref name="VdG-22" />

[[File:fse-thermal-runaway-v2.svg|thumb|upright=1.5|alt=Battery pack with modules of cells; one cell in thermal runaway heats its neighbours and releases gas through a vent.|Structure of an electric vehicle battery pack and spread of thermal runaway from cell to cell. Based on Van der Graaf et al. (2021/2023), pp. 21–22.]]

Ageing changes the behaviour. As a battery ages, its maximum state of charge falls, and since the intensity of thermal runaway depends on the state of charge, the intensity decreases.<ref name="Essl-28">Essl et al. (2021), as cited in Van der Graaf et al. (2021/2023), p. 28.</ref> An older battery in thermal runaway also produces fewer flammable gases than a new one.<ref name="VdG-28">Van der Graaf et al. (2021/2023), p. 28.</ref>

== Fire behaviour of electric vehicles ==
A battery fire develops in two stages. While the cells warm up and start to release gas, the temperature rises slowly, at about 0.2 °C per minute. Once thermal runaway occurs, the rise becomes exponential, at more than 10 °C per minute.<ref name="VdG-26" /> A conventional car fire burns for just over half an hour and peaks at about 5 MW after 10 to 15 min.<ref name="VdG-26" /> Without intervention an electric vehicle burns out in about 90 min.<ref name="NFPA-26">NFPA (2013), as cited in Van der Graaf et al. (2021/2023), p. 26.</ref> After extinguishing, the energy remains in the battery and the fire can keep reigniting for several hours.<ref name="Sun-26">Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 26.</ref>

[[File:fse-thermal-runaway-v3.svg|thumb|upright=1.5|alt=Schematic temperature-time line: a long shallow rise of about 0.2 °C per minute, then a steep rise of more than 10 °C per minute after thermal runaway.|Schematic development of a battery fire: a slow rise while the cells warm up and release gas, then a steep rise once thermal runaway occurs. Based on Van der Graaf et al. (2021/2023), p. 26.]]

The peak heat release rate is not much different from that of a conventional car. A review by Sun and co-workers found a similar peak of about 6 MW for both. The time to the peak was 6 to 40 min for electric vehicles and 6 to 35 min for conventional cars. The total heat released was 4.7 to 8.5 GJ and 3.3 to 10 GJ respectively.<ref name="Sun-25" /> Vehicles with packs of 15 to 25 kWh give about 4 to 6 MW, and the growth of packs to more than 100 kWh has raised this to about 8 MW.<ref name="Sun-25" />

[[File:fse-thermal-runaway-v4.svg|thumb|upright=1.5|alt=Dot plot of peak heat release rates of car fire tests: electric 4.2 to 7 MW, hybrid 6 to 8 MW, conventional 2.1 to 10.8 MW.|Peak heat release rate in full-scale fire tests of battery electric, plug-in hybrid and conventional cars. Based on Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 24.]]

Measured flame temperatures do not differ significantly either. In tests by Lam and co-workers, electric and conventional vehicles both reached about 800 ± 100 °C over periods of 5 to 30 min.<ref name="Lam-23">Lam et al. (2016), as cited in Van der Graaf et al. (2021/2023), p. 23.</ref> In the Netherlands, the car fire curve in the standard NEN 6098 for car parks is based on a TNO design fire from 1999. In those tests, fire could spread to neighbouring cars 10 to 15 min after the first car. The book ''Introduction to Fire Safety Engineering'' notes that these curves may be less conservative for modern cars with more plastics.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref>

The gases differ more than the heat. Burning batteries release methane, carbon monoxide, carbon dioxide and hydrogen, as well as the toxic gases hydrogen fluoride and hydrogen chloride.<ref name="Sun-26" /> A burning pack releases about 20 to 200 mg of hydrogen fluoride per Wh of capacity. For a 100 kWh pack, that amounts to 2 to 20 kg, depending on the state of charge.<ref name="Lar-26">Larsson et al. (2017), as cited in Van der Graaf et al. (2021/2023), p. 26.</ref> Hydrogen fluoride is a strong irritant. Purser lists 200 ppm as the concentration predicted to impair escape in half of the population and 900 ppm as the concentration predicted to incapacitate half.<ref name="Pur-121">Purser (2002), p. 2-121.</ref>

For people escaping, smoke from any vehicle fire is already very toxic because of the burning plastics. On that basis the extra hydrogen fluoride from the battery is considered to have no effect on escape safety.<ref name="RISE-30">RISE (2020), as cited in Van der Graaf et al. (2021/2023), p. 30.</ref> Smoke with more hydrogen fluoride can still spread from the car park to escape routes and stairs of buildings above it.<ref name="VdG-30" />

The released gases also create an explosion hazard. If they do not ignite at once, an explosive mixture can build up in a confined or poorly ventilated space.<ref name="Bis-27">Bisschop et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 27.</ref> When it reaches the lower explosion limit and meets an ignition source, a fire gas explosion can follow.<ref name="VdG-27">Van der Graaf et al. (2021/2023), p. 27.</ref> Hydrogen, one of these gases, has a very wide flammable range: at room temperature a flame can propagate upwards through mixtures with 4.1 to 74 % hydrogen by volume in air.<ref name="IS-16-2">Iqbal and Salley (2004), p. 16-2.</ref> Zalosh describes explosion protection as breaking the chain of events: preventing a fuel-air mixture by ventilation or inerting, removing ignition sources, or limiting the pressure rise by venting or suppression.<ref name="Zal-402">Zalosh (2002), p. 3-402.</ref> Fires of electric vehicles can involve repeated small explosions, flying parts and flash fires.<ref name="VdG-29">Van der Graaf et al. (2021/2023), p. 29.</ref>

== Firefighting and design measures ==
A fire in which the battery pack is involved is very difficult or impossible to extinguish. The pack is closed, and the gases released by the self-propagating thermal runaway keep reigniting.<ref name="VdG-30" /> The coolant, often water, cannot reach the cells directly because the pack is sealed against moisture and dirt.<ref name="IFV-22">IFV (2021b), as cited in Van der Graaf et al. (2021/2023), p. 22.</ref> Cooling therefore needs a lot of water for a long time, and the situation is safe only when all electrical and chemical energy in the battery has been used up.<ref name="VdG-22" /> Putting out the flames needs no more capacity than for a conventional car, but long-term cooling to prevent reignition needs much more water.<ref name="NFPA-46">NFPA (2013), as cited in Van der Graaf et al. (2021/2023), p. 46.</ref>

At present, an immersion container is the best available method for the long-term cooling of a battery pack.<ref name="IFV-46">IFV (2021b), as cited in Van der Graaf et al. (2021/2023), p. 46.</ref> Blanketing the vehicle with foam is not yet a useful tactic, because laboratory studies show that packs can reignite during thermal runaway even with little oxygen.<ref name="VdG-47">Van der Graaf et al. (2021/2023), p. 47.</ref> Sprinklers and water mist cannot extinguish a burning electric vehicle, because the droplets cannot enter the pack. They do reduce the probability that the fire grows and spreads to nearby vehicles and to the structure.<ref name="VdG-37">Van der Graaf et al. (2021/2023), p. 37.</ref>

Ways to bring water into the pack are being studied. One is a sprinkler inside the pack, which could cool the cells in thermal runaway more effectively and possibly extinguish them.<ref name="RISE-56">RISE (2019a), as cited in Van der Graaf et al. (2021/2023), p. 56.</ref> Another is a hose connection in the casing of the pack, tested by a car manufacturer.<ref name="VdG-56">Van der Graaf et al. (2021/2023), p. 56.</ref> In a study of several detector types, carbon monoxide detectors gave the fastest alarm.<ref name="DNV-37">DNV-GL (2019), as cited in Van der Graaf et al. (2021/2023), p. 37.</ref>

A fire in the Singelgarage in Alkmaar on 1 July 2020 illustrates these problems. A police robot was used to tow an electric vehicle out of the car park because its battery pack kept reigniting, and the pack was then cooled for a long time in an immersion container.<ref name="VdG-12">Van der Graaf et al. (2021/2023), p. 12.</ref> For indoor car parks in the Netherlands, the NIPV report suggests, among other measures, not placing charging bays near pedestrian escape routes, because toxic and caustic products of thermal runaway can spread there. It also suggests a means of switching off all charging points at once.<ref name="VdG-35">Van der Graaf et al. (2021/2023), p. 35.</ref><ref name="VdG-36">Van der Graaf et al. (2021/2023), p. 36.</ref>

== Limits and misconceptions ==
The theory of thermal runaway is well developed for self-heating materials and chemical reactors, but it rests on simplifications. Semenov's model assumes a uniform temperature and a single simple reaction, and the classical corrections are often impossible to apply because data are lacking or uncertain.<ref name="Gray-212" /><ref name="Gray-216" /> No formula is available in these sources for the runaway of a battery cell or for its spread from cell to cell. For batteries, the NIPV report lists open questions: the relation between thermal runaway and fire development, the conditions for reignition, the effect and duration of external cooling, and the thermal load on car park structures.<ref name="VdG-55" />

A common misconception is that electric vehicles burn hotter or with a much larger peak heat release rate than conventional cars. The data summarised by NIPV show similar peaks and similar flame temperatures.<ref name="Sun-25" /><ref name="Lam-23" /> The differences lie in the slow start, the long duration, the risk of reignition and explosion, and the larger amount of hydrogen fluoride.<ref name="VdG-26" /><ref name="VdG-29" /> Another misconception is that a fire that seems to be out is safe. Both in self-heating stockpiles and in battery packs, deep-seated reactions can restart hours or days later.<ref name="Gray-212" /><ref name="Sun-26" />

== See also ==
* [[Car park fire]]
* [[Tunnel fire]]
* [[Warehouse fire]]

== Notes ==
<references />

== References ==
* Gray, B. (2002). Spontaneous combustion and self-heating. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-211–2-228). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Kanury, A. M. (2002b). Ignition of liquid fuels. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-188–2-199). National Fire Protection Association; Society of Fire Protection Engineers.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Purser, D. A. (2002). Toxicity assessment of combustion products. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-83–2-171). National Fire Protection Association; Society of Fire Protection Engineers.
* Rockett, J. A., & Milke, J. A. (2002). Conduction of heat in solids. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-27–1-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Zalosh, R. (2002). Explosion protection. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-402–3-421). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Special Hazards & Building Types]]

Electric vehicle fire

2026-11-02T17:15:03Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

An '''electric vehicle fire''' is a fire in a car or other vehicle that is driven wholly or partly by electricity stored in a battery pack, usually a lithium-ion battery. The term covers fully electric cars and plug-in hybrids, which combine a battery pack with a fuel tank.<ref name="NIPV-18">Van der Graaf et al. (2021/2023), p. 18.</ref><ref name="NIPV-24">Van der Graaf et al. (2021/2023), p. 24.</ref> What sets such a fire apart is the battery pack. Once it is involved, the fire can sustain itself through [[thermal runaway]] and reignite until the stored energy has been used up.<ref name="NIPV-22">Van der Graaf et al. (2021/2023), p. 22.</ref>

In many respects an electric car burns like a conventional one. Reviews summarised by the Netherlands Institute for Public Safety (NIPV) found similar peak heat release rates of about 6 MW and similar temperatures of about 800 °C.<ref name="NIPV-23">Van der Graaf et al. (2021/2023), p. 23.</ref><ref name="NIPV-25">Van der Graaf et al. (2021/2023), p. 25.</ref> The differences lie in the course of the fire. A battery fire develops more slowly and lasts longer, can produce small explosions and flare flames, and releases more hydrogen fluoride. It is also very hard to extinguish, because water cannot reach the cells inside the sealed pack.<ref name="NIPV-29">Van der Graaf et al. (2021/2023), p. 29.</ref><ref name="NIPV-30">Van der Graaf et al. (2021/2023), p. 30.</ref> These features matter most in an indoor [[car park fire]], where the possibilities for firefighting are already limited.<ref name="NIPV-11">Van der Graaf et al. (2021/2023), p. 11.</ref>

== Battery packs and how they fail ==
The battery pack of an electric car is built up in stages. Single battery cells deliver about 4 V; cells are connected into modules of up to 60 V, and the modules form a pack of 300 to 1,000 V. Battery packs in passenger cars can now store 100 kWh or more.<ref name="NIPV-21">Van der Graaf et al. (2021/2023), p. 21.</ref> The battery is monitored by a battery management system, which watches the individual cells. If threshold values are exceeded it can cool the pack, stop charging or disconnect the pack completely.<ref name="NIPV-23" />

The central failure mechanism is thermal runaway. NIPV defines it as a failure mechanism that leads to self-heating in a battery or battery cell and that can result in fire.<ref name="NIPV-22" /> Each battery pack has a safe operating window. When this is exceeded, thermal runaway can start at temperatures that depend on the battery type and range from about 70 to 250 °C.<ref name="NIPV-22" /> Lithium-ion is an umbrella term for several chemistries, and elsewhere in the same report the onset range is given as 50 to 270 °C.<ref name="NIPV-18" />

During thermal runaway the heated cell releases flammable gases, which ignite if there is enough oxygen and the ignition temperature is reached. The heat of one cell warms the neighbouring cells, so that the runaway spreads as a chain reaction through the pack.<ref name="NIPV-22" /> If the gases are trapped, they can escape suddenly when the casing or a pressure valve fails. In the right mixture they can cause a flare fire or a gas cloud explosion, and burning fragments can be ejected from a damaged pack.<ref name="NIPV-22" /> If the released gases do not ignite at once, an explosive mixture can build up in a confined or poorly ventilated space.<ref name="NIPV-27">Van der Graaf et al. (2021/2023), p. 27.</ref> The underlying chemistry is treated in the article on [[lithium-ion battery fire]]s.

[[File:fse-electric-vehicle-fire-v1.svg|thumb|upright=1.5|alt=Cells of about 4 V form modules of up to 60 V and a pack of 300 to 1,000 V; one overheating cell heats its neighbours and vents gas.|Structure of an electric vehicle battery pack and the chain reaction of thermal runaway between cells. Based on Van der Graaf et al. (2021/2023), pp. 21–22.]]

Battery failures can have internal and external causes. Internal causes include design or manufacturing errors in the electrodes, separator or electrolyte, and failures of the battery management system that lead to overcharging. External causes include heat, misuse and damage in a collision.<ref name="NIPV-23" /> Published failure probabilities differ widely, from one in a million or lower to about one in 100 million. NIPV concludes that battery failure seems relatively rare, but notes that no lithium-ion battery is intrinsically safe.<ref name="NIPV-23" />

== How an electric vehicle fire develops ==
Temperatures in electric and conventional car fires are similar. Tests on both types of vehicle measured about 800 ± 100 °C over periods of 5 to 30 minutes. Other tests measured flame temperatures of 798 °C, 807 °C and 863 °C for batteries at 50%, 75% and 100% state of charge.<ref name="NIPV-23" /> The heat radiation from an electric car with a fully charged battery is similar to that from a conventional car.<ref name="NIPV-24" />

The peak [[heat release rate]] is also similar. In the review by Sun and co-workers, both types reached about 6 MW. The time to the peak ranged from 6 to 40 minutes for electric cars and from 6 to 35 minutes for conventional cars.<ref name="NIPV-25" /> Electric cars with 15 to 25 kWh batteries reached about 4 to 6 MW. With batteries of more than 100 kWh the peak rose to about 8 MW, against an average of about 6.8 MW for a mid-range conventional car.<ref name="NIPV-25" /> These values are of the same order as the reference value of 8.5 MW for a burning car used in ''Introduction to Fire Safety Engineering''.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref>

[[File:fse-electric-vehicle-fire-v2.svg|thumb|upright=1.5|alt=Dot plot of peak heat release rates: electric cars 4.2 to 7 MW, plug-in hybrids 6 to 8 MW, conventional cars 2.1 to 10.8 MW.|Peak heat release rates in full-scale fire tests of electric, plug-in hybrid and conventional cars. Based on Sun et al. (2020), as cited in Van der Graaf et al. (2021/2023), p. 24.]]

The course of the fire differs. A petrol or diesel fire develops quickly, because a plastic fuel tank can fail and release its fuel all at once. Battery cells first have to warm up and release gas before they burn. In the first minutes the temperature therefore rises slowly, by about 0.2 °C/min, and only at thermal runaway does it rise by more than 10 °C/min.<ref name="NIPV-26">Van der Graaf et al. (2021/2023), p. 26.</ref> A normal car fire burns for just over half an hour and peaks at about 5 MW after 10 to 15 minutes. Without intervention, an electric car burns out in about 90 minutes, and after extinguishing it can keep reigniting for several hours.<ref name="NIPV-26" />

[[File:fse-electric-vehicle-fire-v4.svg|thumb|upright=1.5|alt=Two timelines: conventional car peaks at about 5 MW after 10 to 15 min, out after about 30 min; electric car burns about 90 min, may reignite.|Typical course of a conventional car fire and an electric car fire without intervention. Based on Van der Graaf et al. (2021/2023), p. 26.]]

Fires in electric cars can also include repeated small explosions, parts flying about and flash fires.<ref name="NIPV-29" /> Because the battery pack burns for a long time, nearby vehicles are exposed to heat radiation for longer. NIPV therefore expects electric cars to make "travelling car fires" more likely: fires that pass from vehicle to vehicle through a car park.<ref name="NIPV-29" /> The opposite path also exists. A conventional car burning next to an electric car can heat its battery pack and trigger thermal runaway there.<ref name="NIPV-32">Van der Graaf et al. (2021/2023), p. 32.</ref>

== Energy content of the battery ==
The battery adds energy that is released slowly and is hard to remove. NIPV, following Sun and co-workers, estimates the heat released by a burning battery pack from its storage capacity. The stored electrical energy in kWh is converted to GJ and multiplied by a factor of 7, which Sun and co-workers assume for the heat of flaming combustion. The factor accounts for the difference between the energy stored and the energy released when the pack materials burn:<ref name="NIPV-25" />

<math display="block">E = f \cdot 0.0036 \cdot C</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>E</math> || heat released by combustion of the battery pack || GJ
|-
| <math>f</math> || factor for the heat of flaming combustion, assumed as 7 || none
|-
| <math>0.0036</math> || conversion from kWh to GJ || GJ/kWh
|-
| <math>C</math> || energy storage capacity of the battery pack || kWh
|}

For a battery of 90 kWh, which gives a range of about 400 km, the estimate is 90 × 0.0036 × 7 ≈ 2.3 GJ. A conventional car needs a 30-litre petrol tank for the same distance, and at about 32 MJ per litre this holds about 1.0 GJ.<ref name="NIPV-25" /> On this basis NIPV concludes that an electric car fire requires a larger firefighting effort, while noting that many fuel tanks hold more than 30 litres.<ref name="NIPV-25" /> A 100 kWh pack gives about 2.5 GJ by the same estimate.

[[File:fse-electric-vehicle-fire-v3.svg|thumb|upright=1.5|alt=Bar chart: 90 kWh battery about 2.3 GJ, 30-litre petrol tank about 1.0 GJ; whole cars 4.7 to 8.5 GJ electric and 3.3 to 10 GJ conventional.|Estimated heat released by a 90 kWh battery pack and a 30-litre petrol tank for a range of about 400 km, with the total heat of whole cars. Based on Van der Graaf et al. (2021/2023), p. 25.]]

The factor of 7 is an assumption from a single review and applies to the battery pack only, not to the whole car. The total heat released by complete electric cars in fire tests ranged from 4.7 to 8.5 GJ, against 3.3 to 10 GJ for conventional cars.<ref name="NIPV-25" /> For comparison, ''Introduction to Fire Safety Engineering'' gives a fire load of 6,000 to 12,000 MJ for a car.<ref name="VR-4.4">Van Rossum (2027), §4.4.</ref> NIPV notes that the heat release rate rises with battery capacity, so its values have to be updated as battery packs change.<ref name="NIPV-54">Van der Graaf et al. (2021/2023), p. 54.</ref>

== Toxic and corrosive products ==
Burning battery packs release corrosive substances from the electrolyte and products from the burning plastics. Besides methane, carbon monoxide, carbon dioxide and hydrogen, these include toxic gases such as hydrogen fluoride (HF) and hydrogen chloride.<ref name="NIPV-26" /> Measurements found about 20 to 200 mg of HF per Wh of battery capacity, so a 100 kWh pack can emit 2 to 20 kg of HF, depending on its state of charge.<ref name="NIPV-26" /> Comparative tests found about 1.8 times more HF from an electric car than from a similar conventional car, and hardly any difference in the other toxic gases.<ref name="NIPV-26" />

The effect on escape is limited by the fact that smoke from any car fire is already severely toxic. NIPV, citing Swedish research, considers the effect of the extra HF on escape safety to be nil. Anyone who inhales smoke from a vehicle fire inhales a harmful mixture of toxic gases, however the vehicle is powered.<ref name="NIPV-30" /> HF is reactive and settles on floors, walls, ceilings and clothing. Its concentration in smoke falls by more than half within 10 to 20 minutes, while firefighting clothing becomes more contaminated.<ref name="NIPV-26" />

Water used on an electric car fire also needs attention. Extinguishing water has hardly more environmental impact than water used on conventional car fires and may be drained to the sewer.<ref name="NIPV-29" /> Water in an immersion container is different. Lithium oxide in the burnt pack reacts to lithium hydroxide, so the pH keeps rising, and this water may not be discharged directly to the sewer.<ref name="NIPV-29" /><ref name="NIPV-47">Van der Graaf et al. (2021/2023), p. 47.</ref>

== Firefighting ==
[[Firefighting]] tactics for electric vehicle fires distinguish between putting out the flames and stopping the battery. Putting out the flames needs no more water or crew than for a conventional car. Long-term cooling of the battery pack to prevent reignition needs much more water, depending on the pack and its state of charge.<ref name="NIPV-46">Van der Graaf et al. (2021/2023), p. 46.</ref> Fire services match the flow rate to the expected heat release rate.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref> As rules of thumb, a high-pressure jet of 125 L/min is generally sufficient for up to 2.5 MW, and a low-pressure jet of 450 L/min for up to 10 MW.<ref name="NIPV-45">Van der Graaf et al. (2021/2023), p. 45.</ref>

The battery pack itself is very difficult or impossible to extinguish. It is sealed against moisture and dirt, so water cannot be applied directly to the cells.<ref name="NIPV-22" /><ref name="NIPV-30" /> Once the flames are out, the pack can reignite within a short time and requires continuous cooling and temperature monitoring for a long period.<ref name="NIPV-46" /> At present the best method for long-term cooling is to immerse the vehicle or its battery in a water-filled container. Until the car can be removed, a mobile sprinkler is the best available way for the fire service to keep the pack cool.<ref name="NIPV-46" />

Several common methods do not work as expected. A blanket of foam is not useful, because the pack can reignite under the foam: laboratory studies show that thermal runaway can restart even with little oxygen.<ref name="NIPV-47" /> A sprinkler system cannot extinguish a burning battery pack, because the droplets cannot reach the seat of the fire. It can, however, reduce growth and spread to adjacent vehicles and structures.<ref name="NIPV-37">Van der Graaf et al. (2021/2023), p. 37.</ref>

Other hazards are specific to electric cars. The high-voltage system means a risk of electric shock, although this is negligible when firefighters extinguish from a distance.<ref name="NIPV-43">Van der Graaf et al. (2021/2023), p. 43.</ref> A burning electric car may start to move, and firefighting clothing needs decontamination because of the HF released.<ref name="NIPV-30" /> No additional personal protective equipment is needed compared with a conventional car fire.<ref name="NIPV-47" /> In the Singelgarage fire in Alkmaar in 2020, a police robot towed a burning electric car out of the car park because its battery kept reigniting. The car was then cooled for a long time in an immersion container.<ref name="NIPV-12">Van der Graaf et al. (2021/2023), p. 12.</ref>

== Electric vehicles in car parks and tunnels ==
Charging adds causes of fire that do not exist for conventional cars. These include damaged cables or charging stations, failures of the battery management system, collisions with charging stations and cars with an unnoticed defective battery entering the car park.<ref name="NIPV-29" /> The probability of fire in a parked electric car is unlikely to be higher than in a parked fossil-fuelled car. The charging process, however, is expected to raise the probability of fire, partly because the technology is still in a phase of teething problems.<ref name="NIPV-28">Van der Graaf et al. (2021/2023), p. 28.</ref>

For indoor car parks, NIPV proposes a range of measures rather than a fixed package. Examples are a single facility that cuts the power to all charging points, and charging bays close to the vehicle entrance and at street level. NIPV also advises keeping such bays away from pedestrian escape routes and ventilation inlets.<ref name="NIPV-38">Van der Graaf et al. (2021/2023), pp. 38–39.</ref><ref name="NIPV-41">Van der Graaf et al. (2021/2023), pp. 41–42.</ref> Sprinklers or water mist are recommended because they keep the fire small, which matters more when the fire is harder to fight.<ref name="NIPV-5">Van der Graaf et al. (2021/2023), p. 5.</ref> A plan for removing a burnt car from the car park is also advised, because it must be taken out to stop it reigniting inside.<ref name="NIPV-41" />

Electric vehicles also use road tunnels. The sources available for this wiki do not treat electric vehicle fires in tunnels specifically. Tunnels do not fit building regulations in several respects, so a fire safety engineering approach is applied, and they are usually provided with a smoke control system; this is discussed under [[tunnel fire]].<ref name="VR-11.2">Van Rossum (2027), §11.2.</ref>

== Knowledge gaps and common misconceptions ==
A common misconception is that electric cars are much more likely to catch fire than conventional cars. In 2021 there were no reliable datasets on the probability of electric vehicle fires, either when parked or when charging. The role of production defects and ageing is still unknown.<ref name="NIPV-11" /><ref name="NIPV-54" /> In 2019 about 218,000 electric cars were registered in the Netherlands, about 2.5% of all passenger cars, so experience is still limited.<ref name="NIPV-10">Van der Graaf et al. (2021/2023), p. 10.</ref>

A second misconception is that an electric car fire is far more intense. Peak heat release rates and temperatures are similar; the main differences are duration, reignition and the difficulty of cooling the battery.<ref name="NIPV-23" /><ref name="NIPV-26" /> Ageing may even reduce the intensity of thermal runaway, because an older battery reaches a lower maximum state of charge and produces fewer flammable gases.<ref name="NIPV-28" />

NIPV lists several open research questions. These include the exact process of thermal runaway, the thermal load of electric car fires on car park structures and travelling fires involving several electric cars. Further questions concern the best detection, the effect of different suppression systems, external cooling of the pack and salvage after a fire.<ref name="NIPV-55">Van der Graaf et al. (2021/2023), pp. 55–56.</ref> Manufacturers are testing ways to get water into the pack, such as a sprinkler inside the battery, a hose connection on the casing and a marked spot where the casing can be pierced.<ref name="NIPV-56">Van der Graaf et al. (2021/2023), p. 56.</ref>

== See also ==
* [[Battery energy storage system]]
* [[Photovoltaic system fire]]
* [[Warehouse fire]]

== Notes ==
<references />

== References ==
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Car park fire

2026-11-02T17:15:01Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''car park fire''' is a fire in a building or structure used for parking motor vehicles, usually starting in one car and able to spread to the cars parked next to it. In the Netherlands, the 2012 Building Decree classifies an indoor car park as an "other functional use for parking motor vehicles".<ref name="NIPV-14">Van der Graaf et al. (2021/2023), p. 14.</ref> A single burning car reaches a peak [[heat release rate]] of several megawatts; ''Introduction to Fire Safety Engineering'' uses 8.5 MW as a reference value.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref> Because an indoor car park is a large space of limited height, such a fire quickly produces very high temperatures and dense black smoke.<ref name="NIPV-11">Van der Graaf et al. (2021/2023), p. 11.</ref>

Car park fires are not frequent, but large ones can involve 10 to 50 or more vehicles and damage the building and the functions above it.<ref name="NIPV-11" /><ref name="NIPV-12">Van der Graaf et al. (2021/2023), p. 12.</ref> Major fires in Paris, Liverpool and Stavanger showed that standard design principles had not kept pace with the real behaviour of modern cars. These cars contain more plastic and therefore have a greater fire load and a higher heat release rate than older steel models.<ref name="NIPV-11" /> Design guidance responds with design fires for car parks, [[Sprinkler system|sprinklers]], ventilation and compartmentation, while fire services have to accept that an attack inside a large car park is often not safe.<ref name="NIPV-36">Van der Graaf et al. (2021/2023), p. 36.</ref><ref name="NIPV-45">Van der Graaf et al. (2021/2023), p. 45.</ref> Parking and charging of electric cars adds new questions, which are treated in the article on [[electric vehicle fire]]s.<ref name="NIPV-29">Van der Graaf et al. (2021/2023), p. 29.</ref>

== Why car park fires are a special case ==
Fire safety has always been a dilemma in indoor car parks, even before electric vehicles appeared. The design parameters of most existing car parks were based on older cars with little plastic, burning petrol or diesel.<ref name="NIPV-11" /> According to a 2015 study by DGMR, cited by the Netherlands Institute for Public Safety (NIPV), the assumptions behind the safety of existing car parks are now demonstrably outdated. A different fire development and higher heat release rates have reduced the safety of car parks and of the buildings above them.<ref name="NIPV-12" />

The geometry of a car park makes conditions worse for the fire service. The heat, the loss of visibility, the long hose distances and the difficulty of finding one's way all make reconnaissance and firefighting difficult and risky.<ref name="NIPV-11" /> In the fire in the Singelgarage in Alkmaar on 1 July 2020, poor visibility made it hard to find the burning cars. A firefighting robot was used, and a police robot towed out an electric car whose battery kept reigniting.<ref name="NIPV-12" />

A car park fire can also affect people far from the parked cars. On 26 February 1993 a bomb in a van on the B2 level of the parking garage under the World Trade Center caused an explosion and a subsequent fire. Smoke entered damaged lift shafts and reached the 44th floor of the North Tower about one minute after the explosion.<ref name="FS-57">Ferreira and Strege (2005), p. 57.</ref> Under the 1968 New York City Building Code, the parking garage was the only part of the two towers that was originally sprinklered.<ref name="Lew-162">Lew et al. (2005), p. 162.</ref>

The probability of a car park fire is still uncertain. There are no reliable datasets on the occurrence of fires in electric vehicles, and the role of production defects and ageing is unknown.<ref name="NIPV-11" /> NIPV therefore frames the main question as one of consequences. Even though fires in indoor car parks are not frequent, their effects must be socially acceptable and the measures must be tailored to them.<ref name="NIPV-12" />

== Fire load and heat release rate of cars ==
The fire load of one car is between 6,000 and 12,000 MJ, compared with about 1,900 MJ for a three-seat sofa.<ref name="VR-4.4">Van Rossum (2027), §4.4.</ref> Expressed per floor area, a Swiss survey of 1967 to 1969 found a fire load of 200 MJ/m² in parking buildings.<ref name="IFEG-3.4-6">National Research Council of Canada et al. (2005), p. 3.4-6.</ref> In New Zealand, C/VM2 (second edition, 2025) prescribes a design fire load energy density of 400 MJ/m² for car parks. For car stacking systems it prescribes 400 MJ/m² per tier of car storage.<ref name="MBIE-22">Ministry of Business, Innovation and Employment (2025b), p. 22.</ref>

Measured peak heat release rates of single cars vary widely. Tests at the Fire Research Station gave a peak of 8.5 MW for a 1982 Austin Maestro and 4.5 MW for a 1986 Citroën BX. Tests at VTT on cars from the late 1970s gave peaks of only about 1.5 to 2.0 MW. Babrauskas notes that the reason for the spread is not fully clear and that it is not due to the polymer content.<ref name="Bab-3-27">Babrauskas (2002b), p. 3-27.</ref> The highest of these values equals the reference peak of 8.5 MW for a car in ''Introduction to Fire Safety Engineering''.<ref name="VR-4.5" />

More recent reviews give similar orders of magnitude. A literature review by Sun and co-workers, summarised by NIPV, found a peak of about 6 MW for both electric and conventional cars. The total heat released by conventional cars was 3.3 to 10 GJ. The average for a mid-range conventional car was about 6.8 MW.<ref name="NIPV-25">Van der Graaf et al. (2021/2023), p. 25.</ref>

A typical passenger car fire burns for just over half an hour and reaches a maximum of about 5 MW after 10 to 15 minutes.<ref name="NIPV-26">Van der Graaf et al. (2021/2023), p. 26.</ref> For car parks, the fire scenario is usually described by a heat release rate curve rather than a temperature curve. The heat release rate determines how quickly air and objects heat up, how fast the fire grows and how much smoke is produced.<ref name="NIPV-24">Van der Graaf et al. (2021/2023), p. 24.</ref>

== Fire spread between vehicles ==
The main hazard in a car park is [[fire spread]] from car to car. In tests by TNO in the late 1990s, fire could spread to the adjacent parked cars 10 to 15 minutes after the first car ignited, depending among other things on the distance between them.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref> The Dutch standard NEN 6098 for smoke control in mechanically ventilated car parks contains a design curve based on these tests. In the curve, three cars are burning 15 minutes after ignition, which the book regards as a conservative scenario.<ref name="VR-8.6" />

The NEN 6098 curve combines the curves of three cars that ignite one after the other. Read from the graph in the book, the first car peaks at about 6 MW, and the total reaches about 9.4 MW some 22 minutes after ignition. The curve then falls to zero at about 32 minutes, because the intervention of the fire service is included.<ref name="VR-8.6" /> Because cars now contain more plastic, the book warns that such curves may be less conservative than intended and should be checked against current vehicle scenarios.<ref name="VR-8.6" />

[[File:fse-car-park-fire-v1.svg|thumb|upright=1.5|alt=Line chart of heat release rate against time for three cars and their total, peaking at about 9.4 MW after about 22 minutes.|Design curve for a car park fire in NEN 6098: three cars ignite one after another and the total peaks at about 9.4 MW. Based on Van Rossum (2027), §8.6.]]

NIPV calls fires that pass from vehicle to vehicle "travelling car fires", because they move through the car park, as it were.<ref name="NIPV-29" /> The greater speed of fire spread and the higher heat release rate of modern cars make a long, intense fire involving 10 to 50 or more vehicles more likely. This applies to closed and open car parks and even to outdoor car parks.<ref name="NIPV-11" /> In an open car park with largely open façades, heat and smoke can escape to the outside. Strong winds, however, can increase the probability of fire spreading from one car to the next.<ref name="NIPV-48">Van der Graaf et al. (2021/2023), p. 48.</ref>

Automatic and semi-automatic parking systems increase the risk of large fires involving several cars. The cars are much closer together, on top of, underneath and next to each other. The structure heats up more quickly, the hydraulic oil of the system adds fire load, and the cars are hidden from view.<ref name="NIPV-50">Van der Graaf et al. (2021/2023), pp. 50–51.</ref> Firefighters cannot enter such systems because of the risk of falling and being trapped.<ref name="NIPV-50" />

== Design fires for car parks ==
Fire engineers describe the growth of a car park fire with a design fire. In New Zealand, C/VM2 (second edition, 2025) uses a t-squared fire for car parks without stacking: the heat release rate grows with the square of the time since ignition. The growth stops at flashover, at the peak heat release rate, at the ventilation limit or, in a life safety analysis of a sprinklered building, when the sprinklers operate.<ref name="MBIE-20">Ministry of Business, Innovation and Employment (2025b), p. 20.</ref>

<math display="block">\dot{Q} = \alpha\, t^2</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\dot{Q}</math> || heat release rate of the design fire || kW
|-
| <math>\alpha</math> || fire growth coefficient; 0.0117 for car parks without stacking || kW/s²
|-
| <math>t</math> || time since the start of the growth phase || s
|}

For car parks without stacking, C/VM2 sets the growth coefficient at 0.0117 kW/s² and caps the fire at a peak of 20 MW. This coefficient is a quarter of the fast value of 0.0469 kW/s² used for most other buildings, so any given heat release rate is reached in twice the time.<ref name="MBIE-20" /> The 2013 commentary on the first edition of C/VM2 explains why. Car parks are one of the few uses in which the fuel can be bounded and its layout is controlled. A literature review showed that the early growth of car fires fits a medium t-squared rate.<ref name="MBIE13-55">Ministry of Business, Innovation and Employment (2013), p. 55.</ref> For car parks with stacking systems, the growth is much faster, with a t-cubed relation that depends on the storage height and a peak of 50 MW.<ref name="MBIE-20" />

As a worked example, take the time for the design fire to reach 6 MW, the peak of a typical single car. With <math>\alpha = 0.0117</math> kW/s², the time is the square root of 6,000 divided by 0.0117, which gives about 716 s or 12 minutes. The 20 MW cap is reached after about 1,307 s, or 22 minutes. After 10 minutes the design fire releases about 4.2 MW, which is close to the measured maximum of about 5 MW after 10 to 15 minutes for a typical car.<ref name="MBIE-20" /><ref name="NIPV-26" />

[[File:fse-car-park-fire-v2.svg|thumb|upright=1.5|alt=Two t-squared curves: car park growth reaches 6 MW after 12 minutes, the fast curve after 6 minutes; dashed lines mark car peaks.|C/VM2 design fire for car parks without stacking (0.0117 t²) compared with the fast fire (0.0469 t²), with measured peaks of single cars. Based on Ministry of Business, Innovation and Employment (2025b), p. 20, and Babrauskas (2002b), p. 3-27.]]

The relation describes only the growth phase and starts from an established fire. The commentary notes that tests behind such growth rates use ignition burners of 10 to 40 kW, which skip the slow incipient phase. The method may therefore slightly over-predict the early growth.<ref name="MBIE13-55" /> C/VM2 also sets rules for the ventilation available to a car park fire. Only permanent openings in the perimeter walls and access ramps count, and car parks with overlapping interconnected floors are assessed for the worst floor or the worst pair of adjacent floors.<ref name="MBIE-18">Ministry of Business, Innovation and Employment (2025b), p. 18.</ref>

== Protection measures ==
Automatic suppression is the measure most often recommended for closed car parks. A sprinkler or water mist system can contain a fire to a small number of cars for a long time and so prevent a travelling fire.<ref name="NIPV-36" /> The thermal load on the load-bearing structure then stays limited and short-lived. The risks of collapse, fire spread and smoke spread are reduced, escape and attack routes are more likely to remain intact, and the fire service is more likely to keep the situation manageable.<ref name="NIPV-36" /> In a closed car park, where heat is removed less quickly, NIPV regards a sprinkler system as the most obvious measure. In open car parks, weather makes sprinklers less suitable.<ref name="NIPV-49">Van der Graaf et al. (2021/2023), pp. 49–50.</ref>

Compartmentation is a second option. In the Netherlands, the 2012 Building Decree allows a fire compartment of up to 1,000 m² in a new car park and up to 3,000 m² in an existing one.<ref name="NIPV-14" /> Further division into compartments, or screens between vehicles, can confine a travelling fire. Smaller compartments, however, heat up faster, may reach flashover conditions and make the burning car harder to locate. NIPV therefore prefers an open, clearly arranged area with sprinklers.<ref name="NIPV-35">Van der Graaf et al. (2021/2023), p. 35.</ref>

Early detection gives more time for escape and response. The Dutch Building Decree requires an automatic fire alarm system with full detection in enclosed car parks larger than 1,000 m². In one comparative study, carbon monoxide detectors gave the fastest alert.<ref name="NIPV-37">Van der Graaf et al. (2021/2023), p. 37.</ref> Detection only helps if the alarm is followed up quickly, which is a particular concern in residential buildings.<ref name="NIPV-37" />

Ventilation is mainly a tool for the fire service. In tests of [[jet fan ventilation]] in closed car parks, TNO derived the design fire later adopted in NEN 6098.<ref name="VR-8.6" /> Displacement ventilation can create an air speed of about 1.5 m/s over the full cross-section, following NEN 6098, which can easily mean forty-fold ventilation. It gives firefighters a view of the fire but works only in tunnel-shaped car parks, and it usually starts after people have left.<ref name="NIPV-38">Van der Graaf et al. (2021/2023), p. 38.</ref> Smoke and heat extraction systems, mostly with zero- to ten-fold ventilation, are effective mainly for clearing smoke after the fire is out.<ref name="NIPV-38" /> For large underground car parks, Dutch guidelines sometimes require a computational fluid dynamics simulation, to show that the ventilation allows the fire service to work under acceptable conditions.<ref name="VR-11.2">Van Rossum (2027), §11.2.</ref>

== Firefighting in car parks ==
Fire services prefer to fight a fire from the outside. An interior attack is made only in a small structure, when an exterior attack fails and when it can be done safely.<ref name="NIPV-43">Van der Graaf et al. (2021/2023), p. 43.</ref> A car park is a large structure in which doors cannot be closed to limit the air supply, and the depth of attack is often considerable. NIPV therefore concludes that an offensive interior attack in a car park is generally not safely possible. The commanding officer on site decides whether to attempt one.<ref name="NIPV-45" /> If no offensive attack is possible, the fire service fights the fire defensively from outside, and the car park is considered beyond saving.<ref name="NIPV-44">Van der Graaf et al. (2021/2023), p. 44.</ref>

The type of car park changes the problem. In an underground car park, rising smoke hides the cars and the burning vehicle cannot be located from outside. A smoke-free stairwell with a [[Dry and wet risers|dry riser]] connection can serve as the attack route and shorten the attack depth.<ref name="NIPV-51">Van der Graaf et al. (2021/2023), p. 51.</ref> In a multi-storey car park, fire can spread to several storeys, and crews must find the stairwell, carry hoses and bridge the levels.<ref name="NIPV-52">Van der Graaf et al. (2021/2023), p. 52.</ref> Where a residential building stands above the car park, letting the fire burn out is hardly an option, so fire resistance needs extra attention.<ref name="NIPV-52" />

[[File:fse-car-park-fire-v3.svg|thumb|upright=1.5|alt=Two car park sections: open deck with smoke leaving sideways; smoke-filled underground level entered from a stairwell with a dry riser.|Open car park (left): heat and smoke leave through the open façades. Underground car park (right): smoke fills the level and the crew attacks from a smoke-free stairwell with a dry riser. Based on Van der Graaf et al. (2021/2023), pp. 48–51.]]

Water flow must match the size of the fire. As rules of thumb, a high-pressure jet of 125 L/min is generally sufficient for a fire of up to 2.5 MW. A low-pressure jet of 450 L/min is generally sufficient for up to 10 MW. International practice for a car close to the entrance is to use two low-pressure jets.<ref name="NIPV-45" /> The operating time of a sprinkler system, typically 30 to 120 minutes, limits the time available. If the fire service cannot enter within that time, several cars may catch fire and the car park may still burn out.<ref name="NIPV-47">Van der Graaf et al. (2021/2023), p. 47.</ref>

== Limitations and open questions ==
Design fires for car parks rest on tests of older cars. NIPV notes that Dutch regulations, standards and guidelines are still based on fires in conventionally fuelled vehicles, and it calls for a standard with heat release rates based on the latest information.<ref name="NIPV-11" /><ref name="NIPV-54">Van der Graaf et al. (2021/2023), p. 54.</ref> Measured car fires differ by a factor of four or more, so the choice of a single design value carries large uncertainty.<ref name="Bab-3-27" /> The 2015 DGMR study judged the probability of a large multi-vehicle fire to be higher than even recent statistics suggest, and too high to ignore in design regulations.<ref name="NIPV-11" />

A common misconception is that the fire service will always be able to put out a car park fire. Fire service deployment is a best-efforts obligation and not a substitute for prevention, and its outcome is by definition uncertain.<ref name="NIPV-41">Van der Graaf et al. (2021/2023), p. 41.</ref> NIPV stresses that developers, architects and owners should be told that firefighters will not automatically enter, and that a car park may burn out.<ref name="NIPV-47" /> The Dutch Building Decree aims at preventing casualties and spread to another plot; damage to the car park itself is left to agreements between owners and insurers.<ref name="NIPV-15">Van der Graaf et al. (2021/2023), pp. 15–16.</ref>

== See also ==
* [[Dry pipe sprinkler system]]
* [[Tunnel fire]]
* [[Lithium-ion battery fire]]

== Notes ==
<references />

== References ==
* Babrauskas, V. (2002b). Heat release rates. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-1–3-37). National Fire Protection Association; Society of Fire Protection Engineers.
* Ferreira, M. J., & Strege, S. M. (2005). ''Smoke management systems'' (NIST NCSTAR 1-4D). National Institute of Standards and Technology.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

High-rise building

2026-11-02T17:14:58Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''high-rise building''' is, in fire protection engineering, a building that is nominally taller than about 25 m and from which occupants cannot be rescued from the outside during a fire.<ref name="NIST08-xxxv">National Institute of Standards and Technology (2008), p. xxxv.</ref> Because rescue from outside is not possible, occupants must leave by the stairs, and firefighters often have to climb the same stairs to reach the fire.<ref name="NIST05-169">National Institute of Standards and Technology (2005), p. 169.</ref> Height changes several parts of the fire problem at once. The [[stack effect]] drives smoke up through stairs and lift shafts to floors far from the fire.<ref name="Klo-276">Klote (2002), p. 4-276.</ref>

A full evacuation can take a long time, so many high-rise buildings use a staged or [[phased evacuation]] in which the floors most at risk leave first.<ref name="IFEG-2.8-2">National Research Council of Canada et al. (2005), p. 2.8-2.</ref> Firefighters who climb with breathing apparatus need about 2 min per floor, which has led to calls for protected lifts for the fire service.<ref name="NIST05-170">National Institute of Standards and Technology (2005), p. 170.</ref><ref name="NIST05-218">National Institute of Standards and Technology (2005), p. 218.</ref> The 2001 collapse of World Trade Center Building 7, a 47-storey office building, was the first known total collapse of a tall building primarily due to fires.<ref name="NIST08-xxxv" />

== Definition and characteristics ==
The investigation of World Trade Center Building 7 explains two terms that are often mixed up. Architects and structural engineers speak of a ''tall building'' when it is taller than its surroundings, slender, or needs technologies such as wind bracing, and is nominally taller than 15 to 20 storeys. Fire protection engineers use ''high-rise building'' for buildings nominally taller than 25 m (75 ft) from which external rescue from fires is not possible.<ref name="NIST08-xxxv" /> The fire definition therefore depends on whether rescue from outside is possible, not on the shape of the building.

The main consequence of height is that the risk to an occupant grows with the floor on which that person is. The National Institute of Standards and Technology (NIST) notes that fire rating requirements in the United States were originally developed from experience with buildings of less than 20 storeys. For tall buildings, the consequences of a given threat are more severe for an occupant on an upper floor than on the first floor. With lifts out of service, both full evacuation from upper floors and fire service access to those floors take much longer.<ref name="NIST05-208">National Institute of Standards and Technology (2005), p. 208.</ref>

The fire load in a high-rise office floor is not unusual in itself. A table of fire load densities from Swiss studies of 1967 to 1969, reproduced in the International Fire Engineering Guidelines, gives an average of 800 MJ/m² for a high-rise office building.<ref name="IFEG-3.4-2">National Research Council of Canada et al. (2005), p. 3.4-2.</ref><ref name="IFEG-3.4-5">National Research Council of Canada et al. (2005), p. 3.4-5.</ref> Some performance-based methods do not cover such buildings. In New Zealand, Verification Method C/VM2 (second edition, 2025) lists tall buildings of more than 60 m or 20 storeys as outside its scope, together with buildings that need a managed evacuation.<ref name="MBIE-6">Ministry of Business, Innovation and Employment (2025b), p. 6.</ref>

== Smoke movement and the stack effect ==
Smoke spreads through a building under several driving forces: expansion of heated gases, the stack effect, wind, building services and pressure differences caused by temperature differences.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref> In tall buildings, smoke movement can be dominated by the stack effect.<ref name="Klo-276" /> When it is cold outside, the warmer and lighter air inside rises through shafts such as stairs, lift shafts and service ducts. This is called the normal stack effect. When the outside air is warmer than the inside air, the flow in the shafts reverses and runs downwards.<ref name="Klo-274">Klote (2002), p. 4-274.</ref>

The size of the pressure difference depends on the difference in density between the cold and warm air columns and on the height considered.<ref name="VR-3.2">Van Rossum (2027), §3.2.</ref> Klote gives a relation for the pressure difference between a shaft and the outside, at standard atmospheric pressure. It uses absolute temperatures and the height above the ''[[neutral plane]]'', the level where the pressure inside equals the pressure outside:<ref name="Klo-274" /><ref name="Klo-275">Klote (2002), p. 4-275.</ref>

<math display="block">\Delta P = K_s \left( \frac{1}{T_0} - \frac{1}{T_I} \right) h</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\Delta P</math> || pressure difference between the shaft and the outside || Pa
|-
| <math>K_s</math> || coefficient, 3460 in SI units || Pa·K/m
|-
| <math>T_0</math> || absolute temperature of the outside air || K
|-
| <math>T_I</math> || absolute temperature of the air inside the shaft || K
|-
| <math>h</math> || distance above the neutral plane || m
|}

Above the neutral plane the shaft pressure is higher than outside, and below it the shaft pressure is lower. If air leakage paths are fairly uniform over the height, the neutral plane lies near mid-height. Vented shafts and other uneven leakage can move it considerably. In unusually airtight buildings with exterior stairs, reverse stack effect has even been seen in cold weather.<ref name="Klo-275" /> The relation describes the undisturbed building; the shape and position of openings, flow resistance, wind and mechanical ventilation all change the real pressure distribution.<ref name="VR-3.2" />

[[File:fse-high-rise-building-v1.svg|thumb|upright=1.5|alt=Two building sections with a central shaft: in cold weather air rises in the shaft, in warm weather it falls; a dashed neutral plane at mid-height.|Air movement caused by the normal stack effect (left, cold outside) and the reverse stack effect (right, warm outside). Based on Klote (2002), pp. 4-274–4-275.]]

Klote's worked example is a building 60 m tall with the neutral plane at mid-height, an outside temperature of −18 °C and an inside temperature of 21 °C. At the top, the shaft is 55 Pa above the outside pressure, and at the bottom it is 55 Pa below it.<ref name="Klo-275" /> A recalculation in SI units gives 1.80 Pa per metre and 54 Pa at 30 m above the neutral plane; Klote's 55 Pa (0.22 in. of water) follows from the rounded imperial temperatures. On a mild winter day of 0 °C outside, the same building gives about 27 Pa. A summer day of 30 °C with 21 °C inside gives about −10 Pa, a reverse stack effect. These are of the same order as the 13 Pa that Klote calculates for the buoyancy of an 800 °C fire room, 1.52 m above its neutral plane.<ref name="Klo-276" />

[[File:fse-high-rise-building-v2.svg|thumb|upright=1.5|alt=Line chart of pressure difference against height in a 60 m building: straight lines through zero at 30 m for three outside temperatures.|Pressure difference between a shaft and the outside in a 60 m building with the neutral plane at mid-height and 21 °C inside, for three outside temperatures. Based on Klote (2002), pp. 4-274–4-275.]]

The stack effect can carry smoke far from the fire. With a fire below the neutral plane, smoke moves with the building air into the shafts and up them, and flows out onto upper floors above the neutral plane. If leakage between floors is small, the floors below the neutral plane apart from the fire floor stay relatively free of smoke, until the smoke production exceeds what the stack flows can carry. Smoke from a fire above the neutral plane is carried to the outside through openings in the facade.<ref name="Klo-276" /> Hot smoke can also rise through buoyancy against a reverse stack effect.<ref name="Klo-276" />

Wind adds to these pressures. If a window of the fire room breaks on the windward side, the wind forces smoke through the fire floor and even to other floors, endangering occupants and hampering firefighting. A broken window on the leeward side tends to vent the smoke instead.<ref name="Klo-277">Klote (2002), p. 4-277.</ref> Kandola points out that a pressurisation system in a high-rise building must account for wind-induced pressures, or it may create overpressure that prevents escape or underpressure that makes it ineffective.<ref name="Kan-24">Kandola (2002), p. 1-24.</ref> In cold weather, a broken window on a lower fire floor also increases the stack effect pressures.<ref name="Klo-286">Klote (2002), p. 4-286.</ref>

== Evacuation strategies ==
In multi-storey office buildings, Proulx distinguishes two types of evacuation. An ''uncontrolled total evacuation'' sends everybody to the stairs at the same time. A ''controlled selective evacuation'' clears floors in a planned order and needs a voice communication system to manage it.<ref name="Pro-355">Proulx (2002), p. 3-355.</ref> The International Fire Engineering Guidelines describe the same idea. In high-rise buildings with suitable warning and intercommunication systems, trained staff may manage the evacuation, with floors furthest from the fire first put on alert and evacuated progressively only if the fire keeps developing.<ref name="IFEG-2.8-2" />

Canadian drills show how the two approaches compare. Consider a 15-storey office building with 70 people per floor and two stairs 1,120 mm wide. A total evacuation then gives a flow of about one person per second per exit and a descent speed of 3.6 floors per minute. In both strategies, about 9 min were needed to move everybody below the presumed fire floor. The selective evacuation moved 490 people to safety in that time, while the total evacuation brought all 980 occupants to the ground.<ref name="Pro-355" />

[[File:fse-high-rise-building-v3.svg|thumb|upright=1.5|alt=Two time-floor diagrams of a 15-storey building: all floors leave at once (left); the fire floor and floors above leave in sequence (right).|Schematic movement of the last person leaving each floor in an uncontrolled total evacuation (left) and a controlled selective evacuation with a fire on the 10th floor (right). Based on Proulx (2002), p. 3-355.]]

In the total evacuation, however, people from the most threatened floors would queue, and stair doors opened on the fire floor would let smoke into the stairs above it.<ref name="Pro-356">Proulx (2002), p. 3-356.</ref> On the other hand, a total evacuation is simple and needs no costly systems or training, whereas a controlled evacuation may delay the moment at which occupants who should leave are told to do so.<ref name="Pro-356" /> In a real fire, delays are longer, smoke may block exits and occupants above the fire floor may be unable to leave, which can change the total evacuation time entirely.<ref name="Pro-358">Proulx (2002), p. 3-358.</ref>

A controlled evacuation depends on clear communication. In a 1971 drill in a 21-storey office building in Ottawa, the planned sequence cleared the fire floor first and then the floors above it from the top down. The first public address announcement came almost 3 min late and was ambiguous, and as many as 350 people left their floors before their turn.<ref name="Pro-358" /> The World Trade Center towers used prepared announcements from a Fire Command Desk in each lobby, asking occupants to wait until their floor number was called.<ref name="NIST05-60">National Institute of Standards and Technology (2005), p. 60.</ref> In Australia, the 2020 data sheets for the Fire Safety Verification Method note that alarm systems for phased evacuation commonly have an alert tone and an evacuation tone. Where alarms to some areas are delayed, the delay must be included in the analysis.<ref name="ABCB-243">Australian Building Codes Board (2020a), p. 243.</ref>

The scale of a full evacuation is shown by the NIST simulations of the World Trade Center towers. A phased evacuation of 600 occupants to three floors below the fire floor took 3.5 to 4 min without delay and 11 to 11.5 min with random delays of up to 10 min.<ref name="Ave-160">Averill et al. (2005), p. 160.</ref> A full evacuation of a fully occupied tower of 19,800 people took about 112 min in the buildingEXODUS model, with each extra thousand occupants adding 5.2 min. NIST estimated that such an evacuation on the day of the attacks would have taken over 3 hours.<ref name="Ave-162">Averill et al. (2005), p. 162.</ref> About 87 % of the estimated 17,400 occupants of the towers evacuated successfully, and 99 % of those below the impact floors.<ref name="NIST05-xxxviii">National Institute of Standards and Technology (2005), p. xxxviii.</ref><ref name="NIST05-xxxviii">National Institute of Standards and Technology (2005), p. xxxviii.</ref> NIST recommended that tall buildings be designed for a timely full evacuation, with stairs wide enough for firefighters moving up against the flow.<ref name="NIST05-215">National Institute of Standards and Technology (2005), p. 215.</ref>

[[File:fse-high-rise-building-v4.svg|thumb|upright=1.5|alt=Horizontal bars: phased evacuation of 600 people 11 min, full evacuations of 8,800, 19,800 and 25,500 people 55, 112 and 142 min.|Simulated evacuation times for a World Trade Center tower: phased evacuation of the floors at risk compared with full building evacuations (models with delays of up to 10 min). Based on Averill et al. (2005), pp. 160–162.]]

Each tower had three stairs, two 44 in. (1.12 m) wide and one 56 in. (1.42 m) wide, for up to 365 people per floor. Neither the 1968 New York City Building Code nor the more recent codes that NIST compared made the number and width of stairs depend on the number of storeys.<ref name="NIST05-58">National Institute of Standards and Technology (2005), p. 58.</ref> People with reduced mobility need other solutions, and the use of refuges and lifts for their evacuation is a growing area of research.<ref name="IFEG-2.8-2" /> Evacuation models increasingly allow lifts to be used for escape.<ref name="VR-8.3">Van Rossum (2027), §8.3.</ref>

== Smoke control in stairs and shafts ==
Movement through smoke is common in high-rise fires. In the 1993 bombing of the World Trade Center, 94 % of the occupants of Tower 1 moved through smoke. In two high-rise residential fires, about 96 % of the occupants above the fire floor did so. Proulx explains that repeated opening of doors on the fire floor and on upper floors lets smoke into the stairs, even when they are pressurised. In that sense, the stairs of a high-rise building may be the most dangerous place in a fire after the room of origin and the fire floor.<ref name="Pro-362">Proulx (2002), p. 3-362.</ref>

[[Stair pressurisation]] aims to keep a stair free of smoke by supplying air so that it is at a higher pressure than the fire floor. Its main goal is a smoke-free escape route, and a secondary goal is a smoke-free staging area for firefighters. Air may be injected at one point, usually the top, or at several points over the height. In tall stairs, a single injection point can fail when a few doors near it are open, because all the air escapes there.<ref name="Klo-286" /> Dividing the stair into compartments is an alternative, but it suits only buildings with a staged evacuation, since open doors between compartments cancel the effect.<ref name="Klo-287">Klote (2002), p. 4-287.</ref> The number of doors that will be open at once also depends on the strategy: in a densely occupied building with a total evacuation, all doors may be open together.<ref name="Klo-286" />

Klote's design example for a 20-storey stair shows the role of temperature. With −10 °C outside and 21 °C in the stair, the temperature factor is 1.39 Pa per metre. A pressure difference of 20 Pa at the bottom of the stair grows to 106 Pa at the top. About 3.9 m³/s of supply air is needed, and 6.5 m³/s if the leakage area around the doors is larger than assumed.<ref name="Klo-288">Klote (2002), p. 4-288.</ref> Because the leakage areas are difficult to estimate and control, Klote mentions fans with adjustable supply as one way to correct the airflow after installation.<ref name="Klo-288" />

Lift shafts are a second vertical route. Using lift shafts as smoke shafts prevents their use for evacuation and often spreads smoke to floors far from the fire.<ref name="Klo-278">Klote (2002), p. 4-278.</ref> The MGM Grand Hotel fire in Clark County, Nevada, in 1980 stayed on the casino level. Its heat and smoke, however, spread through seismic joints, lift shafts and stairs to the residential floors above.<ref name="Bry-326">Bryan (2002), p. 3-326.</ref> Of the 85 fatalities, most were found between the 20th and 25th floors, including 9 people in stairs and 5 in lifts.<ref name="Bry-326" /> The lifts had no special smoke protection and were one of the main paths of smoke to the upper floors.<ref name="Klo-288" />

== Firefighting and water supply ==
Fighting fires on the upper floors of tall buildings is not the same as in buildings less than about 30 m (100 ft) high.<ref name="NIST05-169" /> In the World Trade Center, responders without breathing apparatus climbed at about 1.4 min per floor, and firefighters with protective clothing, breathing apparatus and equipment at about 2 min per floor. Evacuees coming down, including people with disabilities, slowed them further in the 44 in. stairs.<ref name="NIST05-170" /> At 2 min per floor, reaching the 60th floor from the ground would take about two hours. NIST found that climbing more than 10 to 12 floors makes it difficult to start aggressive firefighting and rescue at once.<ref name="NIST05-219">National Institute of Standards and Technology (2005), p. 219.</ref>

For this reason NIST recommended fire-protected and structurally hardened lifts in tall buildings. These would give emergency responders timely access and allow the evacuation of occupants with reduced mobility, and they would be reserved for responders during emergencies.<ref name="NIST05-218" /> Lifts of this kind for fire service use are described under [[firefighting lift]]. Rescue by helicopter is not a general alternative. At the MGM Grand Hotel, about 300 guests were evacuated by helicopter and the fire department rescued about 900 others.<ref name="Bry-327">Bryan (2002), p. 3-327.</ref> In the World Trade Center, the New York police and fire departments did not consider roof rescue a viable strategy for a general evacuation.<ref name="NIST05-169" />

Firefighting water must be brought to the upper floors through vertical pipes, called standpipes in the World Trade Center reports; such systems are described under [[dry and wet risers]]. Each World Trade Center tower had an alarm system, automatic sprinklers, a standpipe system and a smoke venting system.<ref name="NIST05-61">National Institute of Standards and Technology (2005), p. 61.</ref> Each tower also had six storage tanks of 5,000 gal (about 19 m³), three on the 110th floor and one each on the 20th, 41st and 75th floors. Water could flow by gravity to the sprinklers, building staff could start pumps, and the fire department could add water through fire department connections.<ref name="NIST05-62">National Institute of Standards and Technology (2005), p. 62.</ref> Although there were redundant vertical supply pipes, each floor had only one connection to its sprinklers.<ref name="NIST05-63">National Institute of Standards and Technology (2005), p. 63.</ref> The standpipes stood in the stairs, so where stairs were close together the spacing between standpipes exceeded the 140 ft (43 m) maximum of the 1968 New York City code.<ref name="Lew-154">Lew et al. (2005), p. 154.</ref>

The reliability of [[Sprinkler system|automatic sprinklers]] has been studied for this building type. In Australia, the 2020 data sheets summarise a fault tree study of wet-pipe sprinkler systems in Australian high-rise office buildings. The general sprinkler reliability of 90 % used in the data sheets lies within the range calculated for these buildings.<ref name="ABCB-81">Australian Building Codes Board (2020a), p. 81.</ref>

== Structural fire resistance and collapse ==
High-rise buildings in the United States are designed to withstand a sizeable fire and remain standing. Their elements receive fire resistance ratings in hours from standard furnace tests. These ratings are relative: they do not mean that a structure will last that long in a real fire, which may be more or less severe than the test.<ref name="NIST05-11">National Institute of Standards and Technology (2005), p. 11.</ref>

World Trade Center Building 7 shows what can happen when this protection is overwhelmed. Debris from the collapse of the north tower ignited fires on at least 10 floors, and the fires on six floors grew and burned until the collapse almost seven hours later. Their growth and spread were consistent with ordinary office contents fires. Had the sprinkler system had a water supply and operated as designed, the fires would likely have been controlled and the collapse prevented.<ref name="NIST08-xxxv" /><ref name="NIST08-xxxvi">National Institute of Standards and Technology (2008), p. xxxvi.</ref>

Local fire damage, made worse by the thermal expansion of long-span floors, led to a cascade of local floor failures, which left a critical interior column unsupported over nine storeys. Its buckling started a [[progressive collapse]], in which local damage spreads from element to element until the whole structure or a large part of it falls.<ref name="NIST08-xxxvi" /> NIST concluded that structures must be designed for situations where sprinklers are absent, do not work or are overwhelmed.<ref name="NIST08-xxxvi" />

Fire can also spread vertically outside the building. The Australian data sheets warn that a phased evacuation strategy may be sensitive to rapid fire spread to several floors over the facade, a [[facade fire]], and that this scenario may need its own evaluation. The evacuation plan should then describe what happens when a fire spreads to several floors at once.<ref name="ABCB-250">Australian Building Codes Board (2020a), p. 250.</ref>

== Limitations ==
Much of the detailed knowledge about high-rise fires comes from a few cases. The World Trade Center reports describe one event with unusual causes, and the evacuation models NIST used had not been validated for buildings as tall as 110 storeys.<ref name="Ave-160" /> The stack effect relation assumes uniform leakage and a known neutral plane, and the real distribution of pressures depends on openings, wind and ventilation systems.<ref name="Klo-275" /><ref name="VR-3.2" /> Drill data, such as the Canadian studies, describe evacuations without smoke, which may differ greatly from a real fire.<ref name="Pro-358" /> A common misconception is that a total evacuation is always the safest choice. The drill data show that it can expose people on the most threatened floors to queues and smoke in the stairs, while a phased approach depends on reliable communication and trained staff.<ref name="Pro-356" /><ref name="Pro-358" />

== See also ==
* [[Fire service access]]
* [[Stay-put strategy]]
* [[Evacuation lift]]
* [[Mass timber]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Kandola, B. S. (2002). Introduction to mechanics of fluids. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-1–1-26). National Fire Protection Association; Society of Fire Protection Engineers.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Institute of Standards and Technology. (2008). ''Final report on the collapse of World Trade Center Building 7'' (NIST NCSTAR 1A).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Special Hazards & Building Types]]

Firefighting lift

2026-11-02T17:14:56Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''firefighting lift''' is a lift in a protected shaft that the fire service can control during a fire, to carry firefighters and equipment to the upper floors of a building. American reports speak of firefighter elevators or fire-protected elevators for emergency responders.<ref name="Lew-191">Lew et al. (2005), p. 191.</ref><ref name="Lew-187">Lew et al. (2005), p. 187.</ref><ref name="NIST-218">National Institute of Standards and Technology (2005), p. 218.</ref> Fire service personnel use a special key to operate such a lift for rescue and for moving personnel and equipment to fight the fire.<ref name="Klo-289">Klote (2002), p. 4-289.</ref>

The lift matters most in tall buildings. Without a working lift, firefighters with their gear climbed the stairs of the World Trade Center at about 1.4 to 2.0 minutes per floor. They had to rest after the first 10 to 12 floors.<ref name="LV-89">Lawson and Vettori (2005), p. 89.</ref><ref name="LV-91">Lawson and Vettori (2005), p. 91.</ref> At such rates a fire 60 floors up can take an hour or more to reach, while a fully furnished floor may burn out in about an hour.<ref name="LV-73">Lawson and Vettori (2005), p. 73.</ref> The United Kingdom and several countries that use British standards have required firefighter lifts in protected shafts for many years. After the World Trade Center disaster of 2001, the United States investigation recommended fire-protected and structurally hardened lifts for the exclusive use of emergency responders in tall buildings.<ref name="Lew-191" /><ref name="NIST-218" />

== Why tall buildings need a firefighting lift ==
Firefighting in tall buildings differs from firefighting in buildings below about 30 m (100 ft). A former New York fire chief defined a [[high-rise building]] as one that lacks viable exterior access to the upper floors and lies beyond the reach of the standard aerial ladder. Firefighters then have to attack the fire from inside and rely almost completely on the systems of the building.<ref name="LV-72">Lawson and Vettori (2005), pp. 72–73.</ref> Building height becomes an immediate problem when safe and working lifts are not available, because access for emergency responders becomes more limited as height increases.<ref name="LV-40">Lawson and Vettori (2005), p. 40.</ref>

The main problem is time and fatigue. Climbing stairs in protective clothing while carrying equipment has a significant physiological effect on firefighters. Fatigue sets in after about 12 floors and reduces what the crew can do once it reaches the fire floor.<ref name="LV-73" /> In the North Tower of the World Trade Center, firefighters climbed the first 10 to 12 storeys in 2 to 5 minutes and then had to stop. After a rest of several minutes they usually managed only 5 to 6 more floors, and some only 3 to 4, before the next rest.<ref name="LV-89" /> The NIST investigation concluded that climbing 20 or more storeys makes effective and timely firefighting and rescue difficult without working lifts.<ref name="NIST-219">National Institute of Standards and Technology (2005), p. 219.</ref>

A second problem is counterflow. When a building is being evacuated, firefighters climbing the stairs move against the stream of occupants coming down. The investigation found that the presence and movement of evacuees hampered the climbing rate of firefighters.<ref name="NIST-199">National Institute of Standards and Technology (2005), p. 199.</ref> Lifts used by first responders would reduce such counterflow problems in the stairs.<ref name="Lew-191" /> Even where protected lifts are provided, NIST recommended enough stair capacity for responders to enter while occupants leave.<ref name="NIST-216">National Institute of Standards and Technology (2005), p. 216.</ref>

The fire itself does not wait for the crews. A fuel load of about 48.8 kg/m² with normal finishes and furnishings takes about one hour to burn out, depending on ventilation. In that time the fire may also spread to other floors and other parts of the building.<ref name="LV-73" /> A firefighting lift therefore supports [[fire service access]] to the upper floors. NIST recommended such lifts precisely to give responders timely access in tall buildings.<ref name="NIST-218" />

== How a firefighting lift works ==
Ordinary passenger lifts are not meant for use in a fire. The American lift code ASME A17.1 requires signs at all passenger lifts stating that they may not be used during a fire.<ref name="Lew-161">Lew et al. (2005), p. 161.</ref> Fire detectors in every lift lobby and machine room initiate a recall. The cars return to the ground floor, where the doors open and the lift is taken out of service.<ref name="Lew-161" /> A more recent option is recall to an alternative floor when the fire is on the ground floor itself.<ref name="Klo-289" />

After recall, the fire service can take over. Fire service personnel use a special key to operate any individual car in a manual mode, as long as they judge it safe to do so.<ref name="Lew-161" /> In the American codes this manual use is called Phase II emergency operation. It covers both fire service access and the rescue of people with disabilities.<ref name="Lew-161" /> The fire service keys allow two tasks: operating lifts for rescue, and transporting personnel and equipment to fight the fire.<ref name="Klo-289" /> In the World Trade Center towers, all lifts had been upgraded for firefighter emergency operation under ASME A17.1 and a New York City law of 1973.<ref name="Lew-169">Lew et al. (2005), p. 169.</ref>

A firefighting lift adds protection to this basic control function, because it runs in a protected shaft.<ref name="Lew-191" /> The American codes require at least one lift serving every floor to be connected to emergency power.<ref name="Lew-161" /> Accessible lifts that the fire service may use to evacuate people with disabilities must also operate in a smoke-protected hoistway. They must be reachable from an area of refuge or a horizontal exit, unless the building is fully sprinklered.<ref name="Lew-161" /> In New Zealand, C/VM2 (second edition, 2025) defines a protected shaft as a space enclosed by fire separations or external walls that houses services, lifts or conveyors passing from one firecell to another.<ref name="MBIE-68">Ministry of Business, Innovation and Employment (2025b), p. 68.</ref>

[[File:fse-firefighting-lift-v1.svg|thumb|upright=1.5|alt=Section through a tall building: a lift in a protected shaft takes firefighters to two floors below the fire floor; they continue up the stair.|A firefighting lift runs in a protected shaft with emergency power; the crew leaves the car at least two floors below the fire floor and continues by stair. Based on Lew et al. (2005), pp. 161 and 191, and Lawson and Vettori (2005), p. 78.]]

Smoke protection of the shaft is a design task of its own. A lift shaft runs through every floor and can carry smoke far from the fire. In the MGM Grand Hotel fire the fire started on the ground floor, but smoke reached the upper floors, where most victims died. The lift shafts had no special smoke protection and were one of the main smoke paths.<ref name="Klo-288">Klote (2002), p. 4-288.</ref>

One method is lift shaft pressurisation: outdoor air is supplied to the shaft so that the pressure in it stays above that of the floors.<ref name="ABCB-127">Australian Building Codes Board (2020a), p. 127.</ref> The analysis is the same as for [[stair pressurisation]], with the stair replaced by the lift shaft.<ref name="Klo-289" /> A moving car pushes and pulls air, the so-called piston effect, which can draw smoke into a pressurised shaft or lobby. In a worked example the critical pressure difference for a car moving at 2.54 m/s was 6.9 Pa, so a shaft pressurised to 12.4 Pa kept the smoke out.<ref name="Klo-288" /><ref name="Klo-289" />

[[File:fse-firefighting-lift-v4.svg|thumb|upright=1.5|alt=Section of a pressurised lift shaft: fan blows outdoor air in, air leaks out at each floor, smoke stays on the fire floor; a moving car.|Lift shaft pressurisation: outdoor air keeps the shaft above the pressure of the floors, so smoke from the fire floor stays out. Based on Klote (2002), pp. 4-288–4-289, and Australian Building Codes Board (2020a), p. 127.]]

The machinery needs protection as well. A fire that affects the lift machinery can carry passengers or firefighters to the fire floor or trap them between floors. Smoke detectors in lift lobbies and machine rooms therefore recall the lifts to a safe level. Heat detectors can switch off the equipment before sprinkler water reaches energised controllers, motors and brakes.<ref name="Sch-4-21">Schifiliti et al. (2002), p. 4-21.</ref>

== Access time without a lift ==
The value of a firefighting lift is easiest to see by estimating the time needed to climb the stairs without one. If a crew climbs at a roughly constant rate, the time to reach a floor is the number of floors to be climbed multiplied by the time per floor. The NIST investigation of the World Trade Center gives the time per floor:<ref name="LV-91" />

<math display="block">t = N \cdot t_\mathrm{f}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>t</math> || time to climb to the target floor by stairs || min
|-
| <math>N</math> || number of floors to be climbed || none
|-
| <math>t_\mathrm{f}</math> || average climbing time per floor, including rests || min
|}

NIST estimated climbing times of about 1.4 minutes per floor for personnel without extra equipment. Firefighters in protective clothing who carried extra equipment needed about 2.0 minutes per floor. Both estimates have an error of about ±0.5 minutes per floor and were derived for climbs to floors in the 30s and 40s of the North Tower.<ref name="LV-91" /> The relation is therefore a rough estimate for fire floors about 30 or more storeys up. A rate of about one floor per minute is often quoted, but it is difficult to keep up beyond the first 20 storeys.<ref name="LV-73" /><ref name="Lew-191" />

As a worked example, consider a fire service crew in full gear that must reach a staging floor 60 floors above the entrance level. With <math>N = 60</math> and <math>t_\mathrm{f} = 2.0</math> min, the climb takes 60 × 2.0 = 120 minutes, or two hours. With the error of ±0.5 minutes per floor, the time lies between 90 and 150 minutes. A police officer without extra equipment, at 1.4 minutes per floor, would need about 84 minutes.<ref name="LV-91" />

[[File:fse-firefighting-lift-v2.svg|thumb|upright=1.5|alt=Line chart of stair climbing time against floors climbed, rising to 120 minutes at 60 floors for firefighters in full gear.|Time to climb the stairs without a lift, t = N × t_f, for 1.4 and 2.0 minutes per floor with the ±0.5 minute error band. Based on Lawson and Vettori (2005), p. 91.]]

The NIST final report gives between one and a half and two hours to reach the 60th floor, depending on the gear and equipment carried.<ref name="NIST-218" /> On 11 September 2001 the lowest impact floors were the 93rd in the North Tower and the 77th in the South Tower. NIST judged that firefighters would need two hours or more to reach them on foot.<ref name="LV-41">Lawson and Vettori (2005), p. 41.</ref>

The time from the equation is only the climb. Firefighters also needed about three to five minutes after arrival to reach the lobby command post and receive an assignment.<ref name="LV-91" /> After the climb, water supply has to be prepared and hoselines stretched before an attack can begin.<ref name="LV-73" /> In a tall building that water comes from standpipe systems: vertical pipe risers with fire hose connections on each floor, comparable to [[dry and wet risers]].<ref name="Eva-11">Evans et al. (2005), p. 11.</ref>

== Operating procedures ==
Many fire services do not fully trust lifts for critical tasks in tall buildings. Lift components are vulnerable to fire, smoke, water and earthquakes, and responders will not use systems with small safety margins when lives are at stake.<ref name="LV-77">Lawson and Vettori (2005), pp. 77–78.</ref> The Fire Department of New York (FDNY) therefore laid down rules for lift use in a training bulletin of 15 March 1997.<ref name="LV-78">Lawson and Vettori (2005), p. 78.</ref> The bulletin first requires crews to account for all lifts serving the fire floor and to check them for victims.<ref name="LV-78" />

The FDNY rules aim to keep the car well away from the fire. When the fire is confirmed on the 7th floor or below, crews avoid the lifts and use the stairs. Crews should not use a lift bank that serves the fire floor if a lower bank reaches within five floors of it. If a bank serving the fire floor must be used, a car with the fire service feature is preferred, because it gives the fire service exclusive control.<ref name="LV-78" /> The crew then selects a floor at least two floors below the fire floor and continues on foot.<ref name="LV-78" />

[[File:fse-firefighting-lift-v3.svg|thumb|upright=1.5|alt=Flowchart: fire on floor 7 or below means stairs; otherwise prefer a lower bank, a fire service car, and stop two floors below the fire.|Decision steps for lift use in the 1997 FDNY training bulletin. Based on Lawson and Vettori (2005), p. 78.]]

Further rules limit the consequences of a failure. Each car must carry a member with a portable radio and forcible entry tools, and no more than six members may ride in one car to prevent overloading. The car is stopped every five floors to confirm that it still responds to the selected floor.<ref name="LV-78" /> A service lift may only be used after the officer in command has declared it safe.<ref name="LV-78" /> NIST added that the use of lifts by emergency responders requires further operating procedures and protocols, including the release of lift door restrictors by response personnel.<ref name="NIST-219" />

== Lessons from the World Trade Center ==
The World Trade Center towers show what happens when lifts are not protected for fire service use. Each tower had 99 passenger lifts and 7 freight lifts, arranged in three zones with sky lobbies on the 44th and 78th floors.<ref name="Lew-169" /> The lifts had fire service recall, but they could not be used for fire service access because they were not fire-protected and did not run in protected shafts.<ref name="Lew-191" /> Most were damaged by the aircraft impacts.<ref name="Lew-191" />

After the impacts, building staff found no working lifts in either tower, and the fire department did not direct anyone to use lifts in the North Tower.<ref name="LV-79">Lawson and Vettori (2005), p. 79.</ref> FDNY personnel found one working lift in each tower. In the North Tower an elevator starter took a small group of firefighters to the 16th floor, but this lift was never reported to the command post.<ref name="LV-79" /> The large utility car that served all floors of the North Tower, normally the fire service's means of reaching the upper floors, fell down its shaft. Its loss seriously hindered FDNY operations.<ref name="LV-79" /><ref name="LV-80">Lawson and Vettori (2005), p. 80.</ref>

In the South Tower a firefighter found one lift running from the lobby to the 40th floor. It was used to shuttle personnel and equipment to the 40th floor and to bring injured occupants down. With its help some firefighters reached the 78th floor.<ref name="LV-31">Lawson and Vettori (2005), p. 31.</ref> The firefighter operating it became trapped in the shaft shortly before the tower collapsed.<ref name="LV-80" /> Before the second aircraft struck, lifts had also helped occupants of the South Tower to leave quickly. Together with early self-evacuation, this is estimated to have saved roughly 3,000 lives.<ref name="Ave-152">Averill et al. (2005), p. 152.</ref>

The 1993 bombing of the same towers had already shown the problem. The lifts were not working, and it took firefighters more than two hours to climb to the 100th floor. It took about five hours to locate and free a group of kindergarten children and adults trapped in a lift.<ref name="Ave-6">Averill et al. (2005), p. 6.</ref> Smoke from the basement also travelled upwards mainly through the lift shafts and stairs.<ref name="LV-42">Lawson and Vettori (2005), p. 42.</ref>

== Regulation and design practice ==
Requirements for firefighting lifts differ between countries, and the folder of sources covers them only partly. According to the NIST investigation of 2005, the United Kingdom and several countries that use British standards had required firefighter lifts in protected shafts for years. United States codes did not require fire-protected lifts for routine responder access, although research on specially protected lifts had been carried out in the late 1970s.<ref name="Lew-191" /> NFPA 101, the American Life Safety Code, permitted lifts with fire fighter service from areas of refuge in its 1991 edition. The 1997 edition permitted a fire fighter service lift with special features as a second means of egress from towers.<ref name="Bry-339">Bryan (2002), p. 3-339.</ref>

The NIST final report of 2005 made two related recommendations for the United States. Recommendation 20 asked for an evaluation of protected or hardened lifts as an evacuation technology. Recommendation 21 asked for fire-protected and structurally hardened lifts for emergency responders in tall buildings. These lifts would give responders timely access and allow mobility-impaired occupants to be evacuated.<ref name="NIST-218" /> Such lifts should be reserved for emergency responders during emergencies. Sufficient stair capacity should still be provided in case a protected lift becomes unusable.<ref name="NIST-218" /><ref name="NIST-216" />

In Australia, the ABCB Handbook for the Fire Safety Verification Method (2020, written for NCC 2019) states that there are no deemed-to-satisfy provisions for the use of lifts during a fire emergency. A performance requirement for lifts used in evacuation is included in the National Construction Code.<ref name="ABCB-58">Australian Building Codes Board (2020b), p. 58.</ref> For tall or innovative buildings, the handbook lists dedicated lifts for firefighters among the enhancements a design team may propose.<ref name="ABCB-49">Australian Building Codes Board (2020b), p. 49.</ref> The accompanying data sheets give a typical reliability of 70% for a lift shaft pressurisation system, with a range of 60% to 75%.<ref name="ABCB-134">Australian Building Codes Board (2020a), p. 134.</ref>

A firefighting lift is related to, but not the same as, an [[evacuation lift]]. A firefighting lift is operated by the fire service for access and rescue. An evacuation lift is intended to help occupants leave the building in addition to the required exits.<ref name="Lew-161" /><ref name="ABCB-106">Australian Building Codes Board (2020b), p. 106.</ref> Evacuation software increasingly offers lifts as an escape option.<ref name="VR-8.3">Van Rossum (2027), §8.3.</ref>

== Limitations and common misconceptions ==
A common misconception is that any lift with a fire service key switch is a firefighting lift. The World Trade Center lifts all had firefighter emergency operation, yet by code they could not be used for fire service access, because they were neither fire-protected nor in protected shafts.<ref name="Lew-169" /><ref name="Lew-191" /> The key switch only gives control. The protected shaft, the emergency power supply and the smoke protection are what set a firefighting lift apart.<ref name="Lew-161" /><ref name="Klo-289" />

A firefighting lift does not remove the need for stairs. NIST noted that protected lifts may become unusable through a malfunction or a threat larger than the one considered in design. Stair capacity should then still allow responders to enter while occupants leave.<ref name="NIST-216" /> The investigation also found that criteria for structurally hardened lifts had not yet been developed.<ref name="NIST-216" /> The climbing rates in the equation above come from a single event, with uncertain start and end times, and should be read as rough estimates.<ref name="LV-91" />

The protection of the shaft is also not absolute. In the Australian data sheets, the typical reliability of a lift shaft pressurisation system is 70%, so about three systems in ten are assumed not to perform as intended.<ref name="ABCB-134" /> The FDNY bulletin addressed the vulnerability of lift components with rules such as stopping at least two floors below the fire and checking the car every five floors.<ref name="LV-77" />

== See also ==
* [[Firefighting]]
* [[Fire hydrant]]
* [[Hose reel]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Lawson, J. R., & Vettori, R. L. (2005). ''The emergency response operations'' (NIST NCSTAR 1-8). National Institute of Standards and Technology.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Firefighting & Fire Service]]

Gaseous extinguishing system

2026-11-02T17:14:54Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''gaseous extinguishing system''' is a fixed [[fire suppression system]] that puts out a fire by discharging an [[extinguishing agent]] in gaseous form, usually into a closed room. A gaseous agent stays a gas at normal room temperature and pressure, spreads easily and distributes itself evenly through an enclosure.<ref name="IS-B41">Iqbal and Salley (2004), p. B-41.</ref> The agents fall into two classes: halocarbons, such as Halon 1301 and its replacements, and inert gases, such as nitrogen and mixtures containing argon. [[Carbon dioxide]] is a third, long-established agent.<ref name="IS-B41" /><ref name="DN-173">DiNenno (2002), p. 4-173.</ref>

Carbon dioxide systems are useful where an inert, electrically non-conductive, three-dimensional gas is needed and where clean-up must be minimal. Hazards protected include flammable liquids, electrical equipment such as transformers and switchgear, engines and computer rooms.<ref name="IS-B47">Iqbal and Salley (2004), p. B-47.</ref><ref name="HO-58">Hopkins et al. (2005), p. 58.</ref> An important advantage of gaseous agents is that no cleaning is needed after a release in the absence of a fire; a few minutes of venting is enough.<ref name="IS-B41" /> The protected room must be tight enough to hold the agent, and a halogenated system in most cases has only one chance to put the fire out. Several agents are also dangerous to people at the concentrations needed.<ref name="IS-B46">Iqbal and Salley (2004), p. B-46.</ref><ref name="IS-B49">Iqbal and Salley (2004), p. B-49.</ref>

Halon may contribute to the depletion of the Earth's ozone layer and is therefore considered harmful to the environment.<ref name="NO-F14">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-14.</ref> Production of halons ended in developed countries on 31 December 1993 under the Montreal Protocol. Since then, several clean halocarbon and inert gas alternatives to Halon 1301 have been commercialised.<ref name="DN-173" />

== Agents and how they work ==
Gaseous agents extinguish fires by different mechanisms. Inert gases act mainly by [[oxygen depletion]], while halocarbon agents act largely by absorbing heat, with some chemical effect on the flame reactions.<ref name="IS-B41" /> Inert gases lower the flame temperature below the level needed to sustain the combustion reactions. They do so by reducing the oxygen concentration and by raising the heat capacity of the atmosphere.<ref name="DN-174">DiNenno (2002), p. 4-174.</ref> Adding enough inert gas to bring the oxygen concentration in air below 12 % leads to [[flame extinction]] in flaming fires.<ref name="DN-174" />

Carbon dioxide displaces the normal atmosphere and reduces the oxygen content below the 15 % that a report of the US Nuclear Regulatory Commission (NRC) gives as necessary for a diffusion flame.<ref name="IS-B46" /> Its action is mainly thermophysical: the reacting gases are prevented from reaching a temperature high enough to keep up the free radicals that sustain the flame.<ref name="IS-B46" /> It is colourless, odourless, electrically non-conductive and stored as a liquid. After release it becomes a gas, with some small solid particles, about 50 % heavier than air.<ref name="IS-B46" />

Halon 1301 works chemically. It inhibits the chain reaction of combustion and therefore suppresses a fire very quickly, unlike agents that remove heat or oxygen.<ref name="IS-B42">Iqbal and Salley (2004), p. B-42.</ref> On a weight basis it is about 2.5 times as effective as carbon dioxide.<ref name="IS-B42" /> Its replacements, the halocarbon clean agents, act by a mix of chemical and physical mechanisms. Compounds containing bromine or iodine scavenge flame radicals like Halon 1301, while most others work mainly by extracting heat from the flame zone.<ref name="DN-174" /> Clean agents are defined as extinguishants that vaporise readily and leave no residue.<ref name="DN-173" />

Inert gas agents can also be blends. In World Trade Center Building 7 (WTC 7), a room containing a 6,000 gal fuel oil tank was protected by Inergen, a proprietary mixture of 52 % nitrogen, 40 % argon and 8 % carbon dioxide by volume.<ref name="MA-73">McAllister et al. (2008), p. 73.</ref><ref name="HO-71">Hopkins et al. (2005), p. 71.</ref> A small amount of carbon dioxide in an inert gas mixture appears to reduce the effects of low oxygen on people substantially.<ref name="DN-184">DiNenno (2002), p. 4-184.</ref>

== System types and components ==
A gaseous system can protect a room or an object. In total flooding, the whole enclosure is filled with the agent to a prescribed concentration, through nozzles placed to give a uniform concentration everywhere.<ref name="IS-B47" /> Local application directs the agent at a localised hazard without filling the enclosure. It suits surface fires in flammable liquids, gases and very shallow solids, for example spray booths, printing presses and rolling mills.<ref name="IS-B47" /> Total flooding is the most common method for halogenated agents.<ref name="IS-B43">Iqbal and Salley (2004), p. B-43.</ref>

[[File:fse-gaseous-extinguishing-system-v1.svg|thumb|upright=1.5|alt=Cross-section of a room with agent cylinders, pipework to ceiling nozzles, detectors, control panel, warning device and closed door.|Main parts of a total flooding gaseous extinguishing system. Based on Iqbal and Salley (2004), pp. B-47 to B-48, Hopkins et al. (2005), p. 58 and DiNenno (2002), p. 4-187.]]

The agent is stored in containers and released through pipes and nozzles. Carbon dioxide is held either in high-pressure steel cylinders at room temperature, at about 850 psi (5.9 MPa) at 21 °C, or in refrigerated low-pressure vessels at about −18 °C and 300 psi (2.1 MPa).<ref name="IS-B48">Iqbal and Salley (2004), p. B-48.</ref> Cylinders hold 50 to 100 lb (23 to 45 kg) of agent each, so most high-pressure systems use several cylinders manifolded together.<ref name="IS-B48" /> Halon 1301 is stored as a liquid under pressure and released as a vapour.<ref name="IS-B42" />

Detection starts the system. In the World Trade Center towers, smoke and heat detectors actuated the release of carbon dioxide into the protected computer rooms.<ref name="HO-58" /> The Inergen system in WTC 7 was to be actuated by two heat detectors at the ceiling above the fuel tank.<ref name="HO-71" /> In the United States, NFPA 12A (1997) of the National Fire Protection Association (NFPA) required automatic actuation of Halon 1301 total flooding systems for ordinary combustible materials, unless the authority having jurisdiction accepted manual actuation.<ref name="IS-B43" />

The discharge must be fast. A Halon 1301 discharge must be substantially complete in a nominal 10 seconds, to prevent the fire from becoming deep-seated, to limit decomposition products and to disperse the agent throughout the enclosure.<ref name="IS-B43" /> For halocarbon clean agents, the maximum discharge time is also 10 seconds. For inert gases, the requirement is currently 60 seconds, and longer times are typically used in Europe.<ref name="DN-189">DiNenno (2002), p. 4-189.</ref> For carbon dioxide on deep-seated fires, NFPA 12 requires the design concentration within 7 minutes and 30 % within 2 minutes. For surface fires, it requires the design concentration within 1 minute.<ref name="IS-B49" />

== Design concentration and agent quantity ==
The design starts from the concentration needed to extinguish the fuel. For halocarbon agents, the minimum extinguishing concentration is measured with a laboratory cup burner and then checked in full-scale approval tests.<ref name="DN-188">DiNenno (2002), p. 4-188.</ref> Halon 1301 design concentrations were historically set at the cup burner value plus a 20 % safety factor, with a minimum of 5 %.<ref name="DN-187">DiNenno (2002), p. 4-187.</ref> Full-scale tests later suggested that 20 % might be insufficient. The first edition of ISO 14520-1, a standard of the International Organization for Standardization (ISO), therefore required a safety factor of at least 30 %. NFPA 2001 requires 30 % for Class B hazards and for systems actuated only by hand.<ref name="DN-188" />

Carbon dioxide needs much higher concentrations. NFPA 12 sets a minimum of 34 % by volume for surface fires such as liquid fuels and 50 % for deep-seated hazards such as electrical equipment. Bulk paper requires 65 % and some special hazards 75 %.<ref name="IS-B47" /> Most surface burning stops at about 20 % carbon dioxide or less, so these values include a large margin of safety.<ref name="IS-B47" /> For inert gases, IG-541 is used at a minimum design concentration of 37.5 % where ordinary combustible materials are involved.<ref name="DN-189" />

[[File:fse-gaseous-extinguishing-system-v2.svg|thumb|upright=1.5|alt=Bar chart of design concentrations: Halon 1301 5 %, HFC-227ea 8.6 %, CO2 surface fires 34 %, IG-541 37.5 %, CO2 deep-seated 50 %; line at 9 %.|Minimum design concentrations of gaseous agents, with the carbon dioxide level above which most people lose consciousness. Based on DiNenno (2002), pp. 4-176 to 4-189 and Iqbal and Salley (2004), pp. B-47 and B-49.]]

Once the design concentration is known, the quantity of agent follows.<ref name="DN-189" /> The protected volume is the net volume of the space.<ref name="DN-189" /> For a halocarbon agent, the mass needed to reach the design concentration in a closed room depends on the room volume and the specific volume of the agent vapour:<ref name="DN-189" />

<math display="block">w = \frac{V}{S} \cdot \frac{C}{100 - C}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>w</math> || mass of agent required || kg
|-
| <math>V</math> || net volume of the protected space || m³
|-
| <math>S</math> || specific volume of the agent vapour at the minimum expected room temperature, <math>S = k_1 + k_2 T</math> || m³/kg
|-
| <math>C</math> || design concentration || % by volume
|}

The factor C/(100 − C) assumes that the agent and air mixture that escapes during discharge is well mixed at concentration C. The formula makes no allowance for leakage after discharge.<ref name="DN-189" /> For inert gases, a logarithmic version is used, which is identical to the expression for carbon dioxide and assumes that the displaced atmosphere is freely vented.<ref name="DN-189" /> The minimum room temperature is used because a colder room needs more agent.<ref name="DN-189" />

=== Worked example ===
HFC-227ea is one of the halocarbon agents acceptable for normally occupied areas.<ref name="DN-184" /> Its cup burner extinguishing concentration for n-heptane is 5.8 to 6.6 %.<ref name="DN-176">DiNenno (2002), p. 4-176.</ref> With the upper value and the 30 % safety factor for Class B hazards, the design concentration is 1.3 × 6.6 ≈ 8.6 %.<ref name="DN-188" /> This lies below the no observed adverse effect level (NOAEL) of HFC-227ea, which is 9.0 %.<ref name="DN-183">DiNenno (2002), p. 4-183.</ref>

For a minimum room temperature of 20 °C, the constants for HFC-227ea give S = 0.1269 + 0.0005 × 20 = 0.137 m³/kg.<ref name="DN-189" /> Assume a closed room with a net volume of 100 m³. The agent mass is then:

<math display="block">w = \frac{100}{0.137} \cdot \frac{8.6}{100 - 8.6} \approx 69\ \mathrm{kg}</math>

If the room can cool to 0 °C, S falls to 0.127 m³/kg and the required mass rises to about 75 kg. The calculation gives only the agent for the design concentration. Leakage, unclosable openings and the required hold time must be assessed separately.<ref name="DN-187" /><ref name="DN-189" />

== Enclosure integrity and hold time ==
The enclosure is a critical part of the system. It must be tight enough to prevent preferential loss of agent during discharge and excessive loss of the agent and air mixture afterwards, so that the concentration is held long enough.<ref name="DN-187" /> Doors, ventilation fans and other openings must be secured before discharge. Systems should not be installed in rooms with unclosable openings unless test data show that adequate concentrations can be reached.<ref name="DN-187" />

The density of the agent and air mixture is the driving force for leakage in still air.<ref name="DN-198">DiNenno (2002), p. 4-198.</ref> Halon 1301 is about five times as heavy as air, so the agent and air mixture flows out through low openings while fresh air enters at the top.<ref name="IS-B45">Iqbal and Salley (2004), p. B-45.</ref> An interface forms and descends; the space above it is unprotected, while the space below keeps the original concentration.<ref name="IS-B46" /> The concentration at the height of the highest combustible therefore matters most.<ref name="IS-B45" />

[[File:fse-gaseous-extinguishing-system-v3.svg|thumb|upright=1.5|alt=Two room sections: just after discharge the room is uniformly filled; later, agent leaks out at the bottom, air enters at the top and an interface descends.|Leakage of a heavy agent and air mixture from an enclosure: the protected zone shrinks as the interface descends. Based on Iqbal and Salley (2004), pp. B-45 to B-46 and DiNenno (2002), p. 4-198.]]

If the extinguishing atmosphere is lost too quickly, glowing embers may reignite when air returns. Openings must then be closed, or extra agent must be discharged.<ref name="IS-B48" /> For rotating electrical equipment such as generators, an extended discharge at a reduced rate maintains the concentration while leakage continues.<ref name="IS-B48" />

Deep-seated fires need long soaking times. If a 5 % concentration of Halon 1301 does not extinguish a fire within 10 minutes, NFPA classes it as deep-seated.<ref name="IS-B43" /> Visible flames on solid materials go out at once, but glowing combustion can continue inside the material.<ref name="IS-B43" /> In full-scale tests on cable trays, a 6 % Halon 1301 concentration held for 10 minutes extinguished fires in unqualified cables, while qualified cables needed 15 minutes.<ref name="IS-B44">Iqbal and Salley (2004), p. B-44.</ref> For carbon dioxide on deep-seated fires, the design concentration must be held for at least 20 minutes.<ref name="IS-B49" />

== Safety of people ==
Several gaseous agents are dangerous at the concentrations needed to put out a fire. Carbon dioxide replaces the oxygen in the air; harmful effects become noticeable at 6 to 7 %, and most people lose consciousness quickly above 9 %.<ref name="IS-B49" /> Since the minimum extinguishing concentrations exceed 9 %, every carbon dioxide system needs adequate safety precautions.<ref name="IS-B49" /> According to the NRC report, carbon dioxide should not be used in occupied areas unless the documented fire risk is greater than the risk to personnel and no viable alternative exists.<ref name="IS-B50">Iqbal and Salley (2004), p. B-50.</ref> In the United States, the Occupational Safety and Health Administration (OSHA) requires a predischarge alarm with enough time delay for people to leave when the design concentration exceeds 4 %.<ref name="IS-B50" />

Accidents show why these rules exist. A US Environmental Protection Agency review found 51 incidents from 1975 onwards, with 72 deaths and 145 injuries from discharges of carbon dioxide systems. All deaths attributed to carbon dioxide were caused by asphyxiation.<ref name="IS-B50" /> In 1998, a total flooding system discharged without warning during maintenance at a US national laboratory, killing one worker.<ref name="IS-B51">Iqbal and Salley (2004), p. B-51.</ref> In 1992, a test discharge at the Duane Arnold nuclear plant let carbon dioxide leak into the control room. The oxygen level fell to 17 % at chest height and 15 % at floor level, below the plant's criterion of 19.5 %.<ref name="IS-B52">Iqbal and Salley (2004), p. B-52.</ref>

Clean agents are limited by toxicity data. For halocarbons, the NOAEL is the highest concentration at which no adverse effect on the heart was observed in animal tests. The lowest observed adverse effect level (LOAEL) is the lowest concentration at which an effect was measured.<ref name="DN-183" /> Where no further modelling data exist, the design concentration in occupied areas must be below the NOAEL. On this basis, HFC-227ea, HFC-23 and FC-3-1-10 are acceptable for normally occupied areas.<ref name="DN-184" />

Inert gases are physiologically inert; their hazard is the low oxygen concentration that they create. Exposure is limited to 5 minutes up to 43 % agent (12 % residual oxygen), to 3 minutes between 43 and 52 %, and to 30 seconds above 52 %.<ref name="DN-184" /> Keeping a room permanently at a reduced oxygen level is a different approach, described under [[oxygen reduction system]].

[[File:fse-gaseous-extinguishing-system-v4.svg|thumb|upright=1.5|alt=Step chart of allowed exposure time against inert gas concentration: 5 min up to 43 %, 3 min from 43 to 52 %, 30 s above 52 %.|Exposure time limits for inert gas agents, with the corresponding residual oxygen concentration. Based on DiNenno (2002), p. 4-184.]]

== Reliability and performance ==
Gaseous systems are credited with a high, but not perfect, reliability. For fire risk studies of United States nuclear power plants, recommended unreliability values are 0.04 for carbon dioxide systems and 0.05 for halon systems.<ref name="PA-P6">Parkinson et al. (1994), as cited in Nowlen et al. (2005b), p. P-6.</ref> For a Halon 1301 system in a computer facility, Steciak and Zalosh estimated the probability that the system fails to protect against fire damage. They found 0.05 for an electrical cable fire, 0.13 for a waste paper fire and 0.08 for a fire outside the protected room.<ref name="SZ">Steciak and Zalosh (1992), as cited in National Research Council of Canada et al. (2005), p. 3.7-3.</ref>

A halogenated agent system in most cases has only a single chance to extinguish a fire. The NRC report therefore recommends testing systems and measuring concentrations at several heights, at least at the highest combustible.<ref name="IS-B46" /> Field experience includes both successes and failures. At Rancho Seco in 1984, a hydrogen fire at the main generator was extinguished by the fixed automatic carbon dioxide system within 14 minutes.<ref name="IS-H12">Iqbal and Salley (2004), p. H-12.</ref> At Millstone Unit 3 in 1999, an inadvertent carbon dioxide discharge in a cable spreading room migrated to other rooms. The control room reached more than 17,000 ppm, and operators wore breathing apparatus for about six hours.<ref name="IS-B51" />

Engineering analysis of these systems relies on standards rather than general methods. The International Fire Engineering Guidelines state that calculations for inert and chemical gaseous systems should be based on their design specifications and on standards such as AS 4214 and NFPA 2001. These calculations are not covered by the guidelines themselves.<ref name="IFEG-2.7-8">National Research Council of Canada et al. (2005), p. 2.7-8.</ref>

== See also ==
* [[Water mist system]]
* [[Extinguishing mechanisms]]
* [[Deluge system]]

== Notes ==
<references />

== References ==
* DiNenno, P. J. (2002). Halon replacement clean agent total flooding systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-173–4-200). National Fire Protection Association; Society of Fire Protection Engineers.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* McAllister, T. P., Gann, R. G., Averill, J. D., Gross, J. L., Grosshandler, W. L., Lawson, J. R., McGrattan, K. B., Pitts, W. M., Prasad, K. R., Sadek, F. H., & Nelson, H. E. (2008). ''Structural fire response and probable collapse sequence of World Trade Center Building 7'' (NIST NCSTAR 1-9, Vols. 1–2). National Institute of Standards and Technology.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nowlen, S. P., Najafi, B., Wyant, F. J., Forester, J., Kazarians, M., Kolaczkowski, A., Joglar, F., Funk, D., Hannaman, G., & Anoba, R. (2005b). ''EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Vol. 2. Detailed methodology'' (NUREG/CR-6850; EPRI 1011989). Electric Power Research Institute; U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.

[[Category:Fire Suppression]]

Required safe egress time

2026-11-02T17:14:51Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''required safe egress time''' ('''RSET''') is the time occupants need to leave a room or building in a fire. It is counted from ignition until all of them have left that space or reached a place of safety.<ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref><ref name="VR-7.1">Van Rossum (2027), §7.1.</ref> It includes not only the walking time but also everything that happens before people start to move, such as detection, the alarm and the time occupants take to respond.<ref name="VR-7.1" /> RSET is one half of the comparison at the centre of performance-based life safety design. The other half is the [[available safe egress time]] (ASET), the time until conditions in the space become untenable.<ref name="NM-367">Nelson and Mowrer (2002), p. 3-367.</ref>

A design is acceptable for life safety when ASET is longer than RSET, usually with a safety margin or safety factor.<ref name="VR-7.1" /> The RSET is calculated by adding the detection time, the notification time, the [[pre-movement time]] and the movement time.<ref name="MBIE25-26" /> The movement time in turn depends on the [[walking speed]] of the occupants and on the flow capacity of doors and stairs.<ref name="VR-7.4">Van Rossum (2027), §7.4.</ref> Because human behaviour before movement is hard to predict, and simple models assume ideal movement, a calculated RSET is best seen as a minimum time rather than a prediction of the real evacuation.<ref name="Proulx-364">Proulx (2002), p. 3-364.</ref>

== Definition and terminology ==
In New Zealand, C/VM2 (second edition, 2025) defines the RSET as the calculated time between ignition of the design fire and the moment when all occupants of a specified room or location have left it.<ref name="MBIE25-26" /> The commentary to the first edition adds that RSET is generally the time for all occupants to reach a place of safety. It can also be defined for one space, such as the room of fire origin.<ref name="MBIE13-198">Ministry of Business, Innovation and Employment (2013), p. 198.</ref> The ''Introduction to Fire Safety Engineering'' describes RSET as the time needed to leave the room or building, and stresses that its calculation rests on assumptions about how people will react.<ref name="VR-7.1" />

The International Fire Engineering Guidelines (2005) use the expansion "Required Safe Evacuation Time" for the same abbreviation.<ref name="IFEG-1.8-1">National Research Council of Canada et al. (2005), p. 1.8-1.</ref> They define it as the time from fire initiation needed for occupants to reach a place of safety, which is then compared with ASET.<ref name="IFEG-1.8-1" /><ref name="IFEG-1.8-3">National Research Council of Canada et al. (2005), p. 1.8-3.</ref> The guidelines note that the times may vary with the location of the occupants relative to the fire.<ref name="IFEG-1.8-1" />

== Components ==
Most sources split the RSET into four or five consecutive periods. C/VM2 uses a detection time, a notification time, a pre-travel activity time and a travel or flow time.<ref name="MBIE25-26" /> The Australian data sheets for the Fire Safety Verification Method use the same four parts.<ref name="ABCB-242">Australian Building Codes Board (2020a), Data Sheet D1, p. 242.</ref> Nelson and Mowrer split the time before movement into two parts: until occupants decide to act, and from that decision until evacuation starts.<ref name="NM-367" /> The International Fire Engineering Guidelines speak of a cue period, a response period, a delay period and a movement period.<ref name="IFEG-1.8-2">National Research Council of Canada et al. (2005), p. 1.8-2.</ref>

[[File:fse-required-safe-egress-time-v1.svg|thumb|upright=1.5|alt=Timeline from ignition: detection, notification, pre-movement and movement make up RSET; a longer ASET bar ends at untenable conditions.|The components of the required safe egress time and its comparison with the available safe egress time. Based on Ministry of Business, Innovation and Employment (2025b), p. 26, and Van Rossum (2027), §7.1.]]

The detection time runs from ignition until an automatic system or a person detects the fire. C/VM2 requires it to be found from deterministic modelling of the detectors or sprinklers, using fixed detector properties regardless of the products to be installed.<ref name="MBIE25-26" /><ref name="MBIE25-27">Ministry of Business, Innovation and Employment (2025b), p. 27.</ref> Methods to calculate it are described under [[detector activation time]]. The notification time covers the delay between detection and the alarm reaching the occupants. C/VM2 takes it as 30 s for standard evacuation strategies, and longer where the alarm is first investigated by staff.<ref name="MBIE25-29">Ministry of Business, Innovation and Employment (2025b), p. 29.</ref>

The pre-movement time follows: the time between the alarm or a cue and the moment occupants begin to move.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> In residential drills studied by Proulx, three-quarters of the evacuation time was spent before moving.<ref name="Proulx-348">Proulx (2002), p. 3-348.</ref> The last component is the movement time. Proulx distinguishes the time needed to go from a starting point to a destination such as a place of safety from the flow time. The flow time is the time a crowd needs to pass one element of the route, such as a door or stair.<ref name="Proulx-347">Proulx (2002), p. 3-347.</ref> The flow time follows from the number of people and the [[exit capacity]] of the usable width.<ref name="Proulx-347" />

== Calculation ==
C/VM2 (second edition, 2025) calls its standard method the simple hydraulic model for RSET.<ref name="MBIE25-26" /> The time before movement is the sum of the detection, notification and pre-travel times. The movement time is either the travel time to the exit or the flow time through the exit, whichever is greater.<ref name="MBIE25-26" />
<math display="block">\mathrm{RSET} = t_\mathrm{d} + t_\mathrm{n} + t_\mathrm{pre} + \max\left(t_\mathrm{trav},\, t_\mathrm{flow}\right)</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\mathrm{RSET}</math> || required safe egress time || s
|-
| <math>t_\mathrm{d}</math> || detection time, from ignition until detection || s
|-
| <math>t_\mathrm{n}</math> || notification time, from detection until the occupants are alerted || s
|-
| <math>t_\mathrm{pre}</math> || pre-travel activity time, from notification until evacuation begins || s
|-
| <math>t_\mathrm{trav}</math> || travel time to the exit || s
|-
| <math>t_\mathrm{flow}</math> || time spent in queues, controlled by the flow capacity of the exit || s
|}

The relation applies to the occupants of one room or location. For the room of fire origin, C/VM2 assumes that the occupants are spread evenly, so that the larger of the travel and flow times governs.<ref name="MBIE25-26" /> For routes through several spaces, the analysis must cover the whole route, so that a restriction closer to the final exit is not missed.<ref name="MBIE25-33">Ministry of Business, Innovation and Employment (2025b), p. 33.</ref> The travel speed falls with crowd density and is capped at 1.2 m/s on level routes.<ref name="MBIE25-32">Ministry of Business, Innovation and Employment (2025b), p. 32.</ref> The flow through a door is calculated from its effective width and a density of 1.9 persons/m², and is limited to 50 people per minute for each self-closing door leaf.<ref name="MBIE25-33" /><ref name="MBIE25-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref>

[[File:fse-required-safe-egress-time-v2.svg|thumb|upright=1.5|alt=Line chart of horizontal travel speed against crowd density: flat at 1.2 m/s up to 0.54 persons per m², then falling linearly to zero at about 3.8.|Design travel speed on level routes in C/VM2 (second edition, 2025) as a function of crowd density. Based on Ministry of Business, Innovation and Employment (2025b), pp. 32–33.]]

Like other hydraulic methods, this calculation assumes that all occupants start at the same moment and move without interruptions, and that most are free of disabilities that would slow the group.<ref name="NM-367" /> The ''Introduction to Fire Safety Engineering'' adds that a hand calculation requires a homogeneous group of people and a simple geometry.<ref name="VR-7.5">Van Rossum (2027), §7.5.</ref> Its eight-step method determines the people per space, the flow capacity of each exit, the distribution of people over the exits and the governing exit. It then adds the walking time of the first person, the flow time of the governing exit and the walking time of the last person to safety.<ref name="VR-7.5" />

== Worked example ==
The commentary to the first edition of C/VM2 calculates the RSET for a ballroom of 50 m by 20 m holding 1,000 people.<ref name="MBIE13-190">Ministry of Business, Innovation and Employment (2013), p. 190.</ref> In the sprinklered building, the design fire grows as a fast t-squared fire and is controlled by the sprinklers at a peak of 2,700 kW.<ref name="MBIE13-202">Ministry of Business, Innovation and Employment (2013), p. 202.</ref> Fire modelling shows that the smoke detectors respond after 66 s. The notification time is 30 s. Because the occupants are awake, unfamiliar with the building and alerted by a voice alarm, the pre-travel activity time is 30 s.<ref name="MBIE13-203">Ministry of Business, Innovation and Employment (2013), p. 203.</ref>

The travel time is checked first. At a density of 1 person/m², the travel speed is 1.03 m/s, and the longest route, taken as length plus width of the room, is 70 m. The travel time is therefore 68 s.<ref name="MBIE13-204">Ministry of Business, Innovation and Employment (2013), p. 204.</ref>

For the flow, each of four exit routes takes a quarter of the occupants, or 250 people.<ref name="MBIE13-204" /> A door leaf of 0.95 m has an effective width of 0.65 m. The flow equation gives about 51 people per minute, which is capped at 50, so a pair of doors passes 100 people per minute.<ref name="MBIE13-205">Ministry of Business, Innovation and Employment (2013), p. 205.</ref> The flow time of 150 s is longer than the travel time and governs:
<math display="block">\mathrm{RSET} = 66 + 30 + 30 + \max(68,\ 150) = 276\ \mathrm{s}</math>
For the people who leave through the foyer, the commentary adds another 150 s of flow time, giving 426 s.<ref name="MBIE13-206">Ministry of Business, Innovation and Employment (2013), p. 206.</ref> The modelled tenability limit for carbon monoxide was not reached within the 1,200 s simulated, so ASET exceeds RSET by more than 924 s in the ballroom.<ref name="MBIE13-206" /> The commentary notes that, strictly, the dose must be calculated at the locations of the moving occupants, which may mean using data from different rooms for different periods.<ref name="MBIE13-206" />

[[File:fse-required-safe-egress-time-v3.svg|thumb|upright=1.5|alt=Stacked bars: ballroom 66 + 30 + 30 + 150 = 276 s; via foyer 426 s; a vertical line marks ASET beyond 1,200 s.|RSET in the commentary example: detection, notification, pre-travel and flow time for the ballroom and for the route through the foyer, compared with the ASET. Based on Ministry of Business, Innovation and Employment (2013), pp. 203–206.]]

The ''Introduction to Fire Safety Engineering'' gives a simpler example of the movement part alone. Three hundred people are divided equally over four stairs, and the two narrowest stairs of 1.6 m govern. Each passes 1.2 persons per second, so 75 people need about 63 s.<ref name="VR-7.5" /> Adding 5 s for the first person to reach the stair and 10 s for the last person to reach safety gives a movement time of 78 s.<ref name="VR-7.5" />

== Methods ==
Escape safety can be assessed with methods of increasing complexity. The ''Introduction to Fire Safety Engineering'' lists the rules of the Dutch building regulations, a capacity model, hand calculations and evacuation simulations with programs such as Simulex, buildingEXODUS and Pathfinder.<ref name="VR-7.5" /> The more complex the building, the more detailed the model needs to be, and a simple method can check a complex one.<ref name="VR-7.5" />

[[Evacuation modelling]] with computer software rebuilds the geometry in three dimensions and fills it with people, often with several occupant profiles.<ref name="VR-10.1">Van Rossum (2027), §10.1.</ref> Its output includes the total evacuation time, queues, densities and bottlenecks, which can be used directly in an ASET/RSET comparison.<ref name="VR-10.1" /> Its main limitation is human behaviour. However convincing the output looks, it reflects the choices made by the modeller, such as which exit people use.<ref name="VR-10.2">Van Rossum (2027), §10.2.</ref> The International Fire Engineering Guidelines likewise warn that hydraulic and network models may give shorter times than real evacuations, because they leave out occupant behaviour.<ref name="IFEG-2.8-3">National Research Council of Canada et al. (2005), p. 2.8-3.</ref><ref name="IFEG-2.8-4">National Research Council of Canada et al. (2005), p. 2.8-4.</ref>

Nelson and Mowrer call the result of a hydraulic calculation the modelled evacuation time. The actual time will generally be longer.<ref name="NM-368">Nelson and Mowrer (2002), p. 3-368.</ref> They express the difference as an apparent evacuation efficiency. It covers delays caused by wardens, stops and restarts where flows merge, individual delays and unbalanced use of exits.<ref name="NM-368" /> Proulx therefore stresses that such calculations give only minimum evacuation times. Some consultants have wrongly presented them as maximum times.<ref name="Proulx-363">Proulx (2002), p. 3-363.</ref><ref name="Proulx-364" />

== RSET for different locations ==
An RSET is not a single number for a whole building. The International Fire Engineering Guidelines recommend carrying out the analysis for each enclosure, such as a room or a mall, or for groups of enclosures such as a floor or the whole building.<ref name="IFEG-1.8-7">National Research Council of Canada et al. (2005), p. 1.8-7.</ref> The analysis is repeated for each design occupant group defined in the fire engineering brief.<ref name="IFEG-1.8-4">National Research Council of Canada et al. (2005), p. 1.8-4.</ref> Occupants far from the fire usually receive fewer cues and therefore get longer pre-movement times than those in the room of fire origin.<ref name="MBIE25-31" />

In New Zealand, the challenging fire scenario of C/VM2 (second edition, 2025) places design fires in various locations and requires ASET to exceed RSET for each of them.<ref name="MBIE25-56">Ministry of Business, Innovation and Employment (2025b), p. 56.</ref> For some smaller rooms, tenability need not be shown for the occupants of the room of fire origin. This applies, for example, to rooms under 500 m² with fewer than 150 occupants and more than one direction of travel.<ref name="MBIE25-56" /> It must still be shown that a fire in such a room does not threaten the occupants in the rest of the building.<ref name="MBIE25-56" /> Tenability is then assessed in the escape routes over the time occupants need to escape, and at the locations of the occupants throughout the evacuation.<ref name="MBIE25-57">Ministry of Business, Innovation and Employment (2025b), p. 57.</ref>

Evacuation strategies also change the RSET from place to place. Where a building uses a phased evacuation, the alarm may sound later in some areas. The Australian data sheets require such delays to be included in the notification time for those areas.<ref name="ABCB-243">Australian Building Codes Board (2020a), Data Sheet D1, p. 243.</ref> In the commentary example above, the occupants who leave the ballroom through the foyer have a longer RSET than those who leave directly outside, because they pass two sets of doors.<ref name="MBIE13-205" /><ref name="MBIE13-206" />

== Comparison with ASET and safety margins ==
Because many inputs for both ASET and RSET are uncertain, a safety margin is important.<ref name="VR-7.1" /> The ''Introduction to Fire Safety Engineering'' gives two common forms: ASET at least equal to a safety factor times RSET, or ASET minus RSET at least equal to a margin.<ref name="VR-7.1" /> The more reliable the inputs, the smaller the margin needed, and a sensitivity analysis can show whether the margin is sufficient.<ref name="VR-7.1" />

In the Netherlands, the standards for large fire compartments, NEN 6060 and NEN 6079, include an ASET/RSET comparison with safety factors.<ref name="VR-7.1" /> For new buildings assessed with the rule-of-thumb method and conservative inputs, the factor is 1.5 when RSET is up to 3 minutes and 2.0 when it is longer.<ref name="VR-7.1" /> With nominal inputs, the factors are 2.0 and 2.5. Existing buildings have lower factors, between 1.0 and 1.5.<ref name="VR-7.1" /> Proulx suggests that, in a conservative approach, the time available should be at least twice the time required.<ref name="Proulx-363" />

[[File:fse-required-safe-egress-time-v4.svg|thumb|upright=1.5|alt=Grouped bars of required ASET/RSET factors for new and existing buildings, conservative and nominal inputs, RSET up to and over 3 minutes; values 1.0 to 2.5.|Safety factors on RSET in NEN 6060 and NEN 6079 for the rule-of-thumb method, as reproduced in the Introduction to Fire Safety Engineering. Based on Van Rossum (2027), §7.1.]]

Other frameworks compare designs rather than set a fixed factor. In Australia, the Fire Safety Verification Method expects the margin of safety, ASET minus RSET, of a proposed performance solution to be at least that of a reference building.<ref name="ABCB-241">Australian Building Codes Board (2020a), Data Sheet D1, p. 241.</ref> In New Zealand, C/VM2 (second edition, 2025) notes that its design fires leave out the incipient phase of a fire, which provides an inherent safety factor for its pre-travel activity times.<ref name="MBIE25-31">Ministry of Business, Innovation and Employment (2025b), p. 31.</ref>

== Limits and uncertainty ==
The largest uncertainty usually lies in the time before movement. Human behaviour is one of the least predictable aspects of fire safety, and the data available have not led to generally valid models for pre-evacuation times.<ref name="VR-7.3" /> The movement part can also be slower than design values suggest. In the first World Trade Center tower in 2001, the median occupant moved down the stairs at about 0.2 m/s, including all interruptions. This is at the slow end of the published values.<ref name="Averill-129">Averill et al. (2005), p. 129.</ref> The flow from the stairs was about 0.5 people per second per door, roughly half the common rule of thumb.<ref name="Averill-130">Averill et al. (2005), p. 130.</ref>

A single RSET also hides the spread between occupants. The Australian data sheets recommend evaluating several scenarios to reflect the different responses of occupant groups and individuals.<ref name="ABCB-241" /> The International Fire Engineering Guidelines list ways to shorten RSET. These include additional and less ambiguous cues, trained personnel, better signage and improved design of escape routes.<ref name="IFEG-1.8-13">National Research Council of Canada et al. (2005), p. 1.8-13.</ref>

== See also ==
* [[Evacuation]]
* [[Occupant load]]
* [[Hydraulic evacuation model]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nelson, H. E., & Mowrer, F. W. (2002). Emergency movement. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-367–3-380). National Fire Protection Association; Society of Fire Protection Engineers.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Dry and wet risers

2026-11-02T17:14:49Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Dry and wet risers''' are fixed vertical water pipes in a building with a hose outlet on each floor. They allow firefighters to connect their hoses close to a fire instead of laying hose from the street.<ref name="H-19">Hopkins et al. (2005), p. 19.</ref> A wet riser is filled with water at all times and connected to an automatic water supply. A dry riser contains air, and in its simplest form has no water supply of its own: the fire service pumps water in through an inlet at street level.<ref name="H-21">Hopkins et al. (2005), p. 21.</ref> In the United States the same installations are called standpipe systems, and the vertical pipes in them are called standpipe risers.<ref name="H-19" />

Risers exist because laying hose from an appliance in the street to an upper or underground floor takes time and effort.<ref name="VdG-66" /> Removing the need for long hose lays from the appliance to an upper floor is generally accepted to improve the initial attack, even though no performance statistics are kept for these systems.<ref name="H-21" /> In a [[high-rise building]], the fire service depends on fire pumps and standpipe systems that can supply enough water for fires beyond the reach of appliances on the ground.<ref name="LV-76">Lawson and Vettori (2005), p. 76.</ref> The pressure needed rises by about 9.8 kPa for each metre of height, which is why tall buildings divide their risers into vertical zones with tanks and pumps.<ref name="T-4-45">Titus (2002), p. 4-45.</ref><ref name="H-38">Hopkins et al. (2005), p. 38.</ref>

== Terminology and components ==
A standpipe system is a fixed piping system that delivers water to designated areas of a building to support manual [[firefighting]].<ref name="H-19" /> In a high-rise building it consists of risers with hose connections at each floor, supplied by pumps and by at least one fire department connection at street level.<ref name="H-19" /> The connections may or may not have hoses permanently attached, depending on the type of system and the local code.<ref name="H-19" /> The number of risers depends on the size and layout of the building, and the systems are usually pressure-monitored to show that they are ready for use.<ref name="H-19" />

The main parts are the riser itself, the hose valve or outlet on each floor, and the inlet through which the fire service pumps water into the system. In American usage this inlet is called a fire department connection or "siamese" connection.<ref name="H-19" /><ref name="H-35">Hopkins et al. (2005), p. 35.</ref> In the World Trade Center towers, each hose station had a valve, a 125 ft (38 m) hose and a nozzle, and the racks were mounted about 4.5 ft (1.4 m) above the stair landing.<ref name="H-10">Hopkins et al. (2005), p. 10.</ref><ref name="H-40">Hopkins et al. (2005), p. 40.</ref> This height allowed firefighters to reach the valve without walking up half a flight of stairs.<ref name="H-40" />

The basic concept has changed little in about 90 years. NFPA 14, the American standard for the installation of standpipe and hose systems, was first adopted in 1915.<ref name="H-19" /> New Zealand documents speak of "fire hydrant systems for buildings" and of an internal [[fire hydrant]] designed to NZS 4510.<ref name="MBIE-53">Ministry of Business, Innovation and Employment (2025b), p. 53.</ref><ref name="MBIE-54">Ministry of Business, Innovation and Employment (2025b), p. 54.</ref> A Dutch guideline for car parks, translated into English, uses the term "dry riser".<ref name="VdG-51">Van der Graaf et al. (2021/2023), p. 51.</ref>

== Wet, dry and other types ==
NFPA 14, as summarised in the NIST report on the World Trade Center suppression systems, describes standpipe systems by their water supply as wet, damp or dry.<ref name="H-21" /> Wet systems are the preferred arrangement. They are filled with water at all times and connected directly to an automatic water supply, such as the public water main, a tank with a pump, an elevated tank or a pressure tank.<ref name="H-21" /> Because the pipes are full, every part of the building that contains them must be heated to prevent freezing.<ref name="H-21" />

[[File:fse-dry-and-wet-risers-v1.svg|thumb|upright=1.5|alt=Two building sections: left, a water-filled riser fed by a tank and pump; right, an empty riser fed by a fire appliance at a street-level inlet.|A wet riser (left) is always full and connected to an automatic water supply; a manual dry riser (right) contains air until the fire service pumps water in through the inlet at street level. Based on Hopkins et al. (2005), pp. 19 and 21.]]

Dry systems contain air at atmospheric pressure or under pressure, and NFPA 14 divides them into manual, automatic and semi-automatic types.<ref name="H-21" /> A manual dry standpipe has only fire department connections and no connection to an automatic water supply: water reaches the outlets only when a fire service pump feeds the inlet.<ref name="H-21" /> An automatic dry standpipe holds pressurised air or nitrogen behind a dry-pipe valve; when a hose valve is opened, the air escapes and the water supply opens the valve and fills the pipe.<ref name="H-21" /> A semi-automatic dry standpipe has a connection to an automatic water supply that is kept closed until it is needed.<ref name="H-21" />

The choice between wet and dry involves a trade-off. A wet riser gives water at once, but needs a reliable automatic supply and protection against frost.<ref name="H-21" /> A manual dry riser needs no permanent supply, but delivers nothing until the fire service has connected a pumping appliance to its inlet.<ref name="H-21" /> Its use therefore depends on [[fire service access]] to the inlet. The New York City building code, in both editions compared by NIST, provides for wet standpipes.<ref name="RG-46">Razza and Grill (2005c), p. 46.</ref>

Standpipe systems are also classified by user.<ref name="H-19" /> Class I systems have 2½ in. (63.5 mm) hose valves for heavy streams used by the fire service in advanced stages of a fire.<ref name="H-20">Hopkins et al. (2005), p. 20.</ref> Class II systems have 1½ in. (38 mm) hose stations with hoses and nozzles for trained occupants or fire brigades to fight small fires.<ref name="H-20" /> Class III systems combine both and serve the fire service, trained occupants and fire brigades.<ref name="H-20" />

Whether occupants should rely on pre-connected hoses has long been debated.<ref name="H-21" /> Some jurisdictions permit only Class I systems, for use by the fire service alone.<ref name="H-21" /> The pre-connected hoses of the World Trade Center towers were intended for the trained fire brigade or the fire department.<ref name="H-10" />

== Pressure and building height ==
Water in a vertical pipe presses down with its own weight, so the pressure at the bottom of a riser is higher than at the top.<ref name="T-4-45" /> The pressure caused by a column of water alone is called static pressure, and the height of the column is called the head.<ref name="T-4-45" /> The SFPE handbook gives this relation from fluid statics: the pressure difference between two points equals the density of the water, times the acceleration of gravity, times their difference in height.<ref name="T-4-45" />

<math display="block">\Delta p = \rho\, g\, \Delta h</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\Delta p</math> || static pressure difference between two levels || Pa
|-
| <math>\rho</math> || density of water (about 1,000 kg/m³) || kg/m³
|-
| <math>g</math> || acceleration of gravity (9.81 m/s²) || m/s²
|-
| <math>\Delta h</math> || difference in height between the two levels || m
|}

The relation applies to water at rest or to the elevation part of the pressure in a flowing system; friction losses in pipes and fittings come on top of it.<ref name="T-4-45" /><ref name="H-20" /> Titus gives the same rule in American units: a column of water of ''h'' feet corresponds to 0.433''h'' psi, and 1 psi corresponds to 2.3 ft (0.7 m) of water.<ref name="T-4-45" /> In SI units this is about 9.8 kPa per metre of height. The pump ratings of the World Trade Center standpipe system follow this rule: a pump rated at 360 psi was also listed as 831 ft of head, and 0.433 × 831 ≈ 360.<ref name="H-34">Hopkins et al. (2005), p. 34.</ref>

'''Worked example.''' A standpipe must deliver its design flow at 100 psi (689 kPa) at the most remote outlet.<ref name="H-20" /> If that outlet is 300 ft (91.4 m) above the pump, the height alone costs:

<math display="block">\Delta p = 1000 \times 9.81 \times 91.4 \approx 897\ \mathrm{kPa}\ (130\ \mathrm{psi})</math>

The pump must therefore supply at least 689 + 897 ≈ 1,590 kPa, about 230 psi, before any friction loss is added (rounded up). The height of 300 ft is the limit used in the New York City building code, which allows an automatic fire pump to serve only the lowest 300 ft of a taller building.<ref name="RG-48">Razza and Grill (2005c), p. 48.</ref> Zones above that height must be supplied by gravity or pressure tanks.<ref name="RG-48" />

[[File:fse-dry-and-wet-risers-v2.svg|thumb|upright=1.5|alt=Straight line of static pressure against height, rising about 9.8 kPa per metre, with a marker at 91.4 m and 897 kPa.|Static pressure of a water column against height, about 9.8 kPa per metre; the marker shows the worked example at 300 ft (91.4 m). Friction losses and the outlet pressure come on top. Based on Titus (2002), p. 4-45.]]

The same physics explains the vertical zoning of tall buildings. The standpipes in each World Trade Center tower were divided into four vertical water supply zones: floors 1 to 8, 9 to 41, 42 to 76 and 77 to 110.<ref name="H-38" /> Check valves between the zones stopped water from flowing down from one zone into the next.<ref name="H-38" /> Because the working pressure varied with elevation, the riser pipes were heavier at the bottom. The lowest zone had extra-strong steel pipe with fittings rated at 1,000 psi, the top zone standard pipe with 350 psi fittings.<ref name="H-37">Hopkins et al. (2005), p. 37.</ref> Adjustable pressure-restricting devices were fitted to the hose racks on floors 77 to 94.<ref name="H-40" />

[[File:fse-dry-and-wet-risers-v3.svg|thumb|upright=1.5|alt=Tall tower schematic split into four standpipe zones, with water tanks near the top of zones, pumps on mechanical floors and inlets at street level.|Vertical zoning of the standpipe system in a World Trade Center tower: four supply zones, gravity tanks for the first minutes, manual fire pumps on mechanical floors and fire department connections at street level. Based on Hopkins et al. (2005), pp. 10, 30 and 38.]]

High pressures also limit what the fire service can supply. The 1997 high-rise firefighting procedures of the Fire Department of the City of New York require building fire pumps to deliver their rated capacity at 50 psi (345 kPa) at the highest outlet.<ref name="LV-I">Lawson and Vettori (2005), Appendix I.</ref> The same procedures warn that these pumps, and gravity tanks, may not give adequate nozzle pressure on the upper floors, so the fire service's own pumpers are the primary supply in most operations.<ref name="LV-I" /> As a rule of thumb, they recommend one pumper supplying the standpipe for each hand line in use.<ref name="LV-I" />

== Design requirements ==
In the United States, NFPA 14 sets the design flow for Class I and Class III systems at 500 US gal/min (about 1,900 L/min) at 100 psi at the most hydraulically remote standpipe.<ref name="H-20" /> Each additional standpipe adds 250 US gal/min (about 950 L/min), up to a total of 1,250 US gal/min (about 4,700 L/min).<ref name="H-20" /> This assumes two 2½ in. hose lines on the most remote standpipe and one on each of the others.<ref name="H-20" /> "Remote" means hydraulically remote: the outlet needing the highest pressure at the source, which may be the top floor or the far end of a building, so the design is checked with a hydraulic analysis.<ref name="H-20" /> Class II systems are designed for 100 US gal/min (about 380 L/min) at 65 psi (448 kPa).<ref name="H-20" />

The spacing of the outlets is expressed in lengths of hose. When the World Trade Center was built, NFPA 14 required Class I hose valves to be placed so that every part of a building lay within 30 ft (9.1 m) of a nozzle. That nozzle was assumed to be on a 100 ft (30.5 m) hose.<ref name="H-20" /> The New York City building code, in its 1968 and 2003 editions, requires that every point on every floor can be reached by a 20 ft (6.1 m) stream. The stream comes from a nozzle on not more than 125 ft (38 m) of hose.<ref name="RG-46" /> The risers and their 2½ in. hose valves must be inside the stair enclosures, with an outlet on each floor 5 to 6 ft above the landing.<ref name="RG-46" /><ref name="RG-47">Razza and Grill (2005c), p. 47.</ref>

The same code sets the riser size by height: at least 4 in. (about 100 mm) in diameter up to 150 ft (46 m) from the entrance floor to the highest outlet, and 6 in. (about 150 mm) above that.<ref name="RG-48" /> Where a system has several risers, they must be cross-connected at or below street level.<ref name="RG-47" /> Buildings over 300 ft (91 m) high need an additional water supply for their standpipes, supplemented by manually operated fire pumps.<ref name="RG-49">Razza and Grill (2005c), p. 49.</ref>

In New Zealand, the second edition of C/VM2 (2025) requires vehicle access to a hard-standing from which an unobstructed path leads to within 20 m of the inlets to sprinkler or fire hydrant systems.<ref name="MBIE-53" /> It also requires the means to deliver water for firefighting to all parts of the building.<ref name="MBIE-53" /> That water comes either from a pumping appliance parked so that every point in the building lies within 75 m, about three hose lengths, or from an internal hydrant installed to NZS 4510.<ref name="MBIE-54" /> For intermediate floors, a commentary in the same document accepts a hydrant in a safe path within 40 m, about two hose lengths, of the most remote point.<ref name="MBIE-55">Ministry of Business, Innovation and Employment (2025b), p. 55.</ref>

In the Netherlands, guidance on indoor car parks advises placing a dry riser where the fire service enters, because it shortens the hose to be laid out and so the attack time.<ref name="VdG-66">Van der Graaf et al. (2021/2023), p. 66.</ref> It recommends a smoke-free stairwell as an attack route to underground storeys, with a connection point for a dry riser, and notes that most car parks have such a stairwell.<ref name="VdG-51" /> The International Fire Engineering Guidelines (2005) list the location of hydrants and other fire service facilities, and the amount of firefighting water available, among the factors that govern the efficiency of fire service intervention.<ref name="IFEG-1.9-5">National Research Council of Canada et al. (2005), p. 1.9-5.</ref>

== The World Trade Center, 2001 ==
The towers of the World Trade Center show how a large riser system works and how it can fail. Each tower had three standpipe risers per zone, one in each exit stair, with Class III hose stations on every floor.<ref name="H-10" /><ref name="H-37" /> In the initial operating mode, water came by gravity from 5,000 US gal (about 19,000 L) tanks placed above each zone, enough for 500 US gal/min for about 10 minutes.<ref name="H-30">Hopkins et al. (2005), p. 30.</ref><ref name="H-39">Hopkins et al. (2005), p. 39.</ref> During that time, staff were expected to start manual 750 US gal/min fire pumps on the 7th, 41st and 75th floors, which drew on the city [[fire water supply]].<ref name="H-10" /><ref name="H-39" /> Fourteen fire department connection stations at street level allowed the fire service to add water and pressure.<ref name="H-10" />

On 11 September 2001 the aircraft impacts damaged the stairwells around the risers. In WTC 1, NIST concluded that two of the three standpipe risers were probably lost and the third possibly, so the standpipe water supply would have been lost after a limited time.<ref name="E-37">Evans et al. (2005), p. 37.</ref><ref name="E-38">Evans et al. (2005), p. 38.</ref> Isolation valves on the 88th and 99th floors could in principle have allowed a partial restoration, but the status of the systems was unknown.<ref name="E-38" /> The fire service's first size-up found the standpipe systems inoperable at the impact zones.<ref name="LV-77">Lawson and Vettori (2005), p. 77.</ref> Restoring them would have required building staff to close an isolation valve between the towers and to go up with firefighters to seal the broken pipes and run the pumps.<ref name="LV-77" />

The scale of the task shows why risers matter. NIST estimated that about 4,700 L/min would be needed to extinguish half a floor of a tower, using about five 63.5 mm hose lines, and about 9,400 L/min and ten lines for a whole floor.<ref name="LV-77" /> Such an operation would involve at least 15 to 20 engine companies, and if the building's pumps were not working, special high-rise pumpers would have to feed the standpipes.<ref name="LV-77" /> Fire departments also rely on lifts, such as a [[firefighting lift]], to carry personnel and equipment up tall buildings, and lifts are often an integral part of high-rise firefighting.<ref name="LV-77" /> In 7 World Trade Center later that day, a chief officer opened a standpipe on the 4th floor and found no water.<ref name="LV-110">Lawson and Vettori (2005), p. 110.</ref>

== Limitations ==
No performance records are kept for standpipe and hose systems, so their reliability cannot be quantified from the sources.<ref name="H-21" /> They are not an alternative to automatic sprinklers.<ref name="H-21" /> A riser only supplies water if its source works. In the World Trade Center towers the extra water depended on manually started pumps, whereas NFPA 14 requires an automatic supplementary supply as current best practice.<ref name="H-10" /><ref name="E-xxxi">Evans et al. (2005), p. xxxi.</ref> A dry riser depends entirely on the fire service reaching and charging its inlet.<ref name="H-21" />

Because risers run through the stairs, damage that affects the stair cores can disable several risers at once, as it did in 2001.<ref name="E-37" /> The NIST emergency response report lists the distance between stairwells containing standpipes, and redundant standpipe and water supply systems, among the difficulties seen in the responses of 1993 and 2001.<ref name="LV-l">Lawson and Vettori (2005), p. l.</ref> Sectional isolation valves, such as those on floors 88 and 99 of the World Trade Center risers, offer a way to shut off a damaged section and restore part of a system.<ref name="H-39" /><ref name="E-38" />

== See also ==
* [[Hose reel]]
* [[Fire extinguisher]]
* [[Positive pressure ventilation]]

== Notes ==
<references />

== References ==
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* Lawson, J. R., & Vettori, R. L. (2005). ''The emergency response operations'' (NIST NCSTAR 1-8). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Razza, J. C., & Grill, R. A. (2005c). ''Comparison of the 1968 and current (2003) New York City Building Code provisions'' (NIST NCSTAR 1-1F). National Institute of Standards and Technology.
* Titus, J. J. (2002). Hydraulics. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-44–4-71). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)

[[Category:Firefighting & Fire Service]]

Ethics in fire safety engineering

2026-11-02T17:14:47Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Ethics in fire safety engineering''' concerns the right conduct of engineers who design and assess fire safety: acting with integrity, working within their competence and being open about assumptions and uncertainty. Ethics is the branch of philosophy that reflects critically on right action. It seeks criteria for judging whether an action is good or wrong and for weighing its motives and consequences.<ref name="VR2">Van Rossum (2027), §2.</ref> In fire safety, integrity matters because a wrong decision can cost lives, while clients and employers may create incentives that tempt an engineer to act otherwise.<ref name="VR2" />

The issue is sharper in [[fire safety engineering]] than in rule-based design. A performance-based justification can be so complex that the authority reviewing it cannot check it, and must then rely on the integrity of the engineer.<ref name="VR2.1">Van Rossum (2027), §2.1.</ref> Guidance in several countries responds by setting requirements for competence and by providing for independent review. The Australian Fire Safety Verification Method, for example, may only be used by fire safety engineers with demonstrated competency. The need for a peer review must be considered at the design brief stage of each project.<ref name="FSVM5">Australian Building Codes Board (2022), p. 5.</ref><ref name="FSVM6">Australian Building Codes Board (2022), p. 6.</ref>

== Why ethics matters in fire safety engineering ==
In a prescriptive approach, the reviewing authority checks a design against fixed rules, which is relatively easy. In a scientific approach such as fire safety engineering, the justification provided by the applicant may be so complex that the authority cannot verify it, or can do so only with difficulty.<ref name="VR2.1" /> ''Introduction to Fire Safety Engineering'' gives computational fluid dynamics (CFD) as an example. Understanding how CFD works requires long scientific training and frequent use, and a reviewing authority seldom employs a CFD specialist.<ref name="VR2.1" /> The boundary conditions and assumptions of a simulation are then often unknown to the reviewer, although checking exactly these is important.<ref name="VR2.1" />

Where a design cannot be checked, the authority has to rely on the integrity of the fire safety engineer. A counter-expertise by a second engineer is possible, but it is not economically feasible for every project.<ref name="VR2.1" /> The book recommends one where there are good reasons, for example when results are unexpected or when the engineer is new and has no reputation yet.<ref name="VR2.1" />

The same point is made by the developers of fire models. The technical guides of CFAST and the Fire Dynamics Simulator (FDS) state that users bear sole responsibility for judging whether the model suits an application and for the conclusions they draw. The models are intended only for people competent in fluid dynamics, thermodynamics, heat transfer, combustion and fire science.<ref name="CFASTix">Peacock et al. (2026), p. ix.</ref><ref name="FDSvii">McGrattan et al. (2026), p. vii.</ref> A model "may or may not have predictive capability" in a particular case, and all results should be evaluated by an informed user.<ref name="CFASTix" />

== Interests and incentives ==
Almost every party in a building project has an economic interest in the outcome of a fire safety analysis. ''Introduction to Fire Safety Engineering'' lists what a client may gain: a permit, and savings on construction through cheaper building methods or materials, a lower level of fire protection installations or dropping costly measures.<ref name="VR2.2">Van Rossum (2027), §2.2.</ref> In a tender, the contractor bidding for the work also has an interest. An engineer employed by a firm may find that the employer favours certain outcomes for the same reasons, and in a government body the pressure may be political.<ref name="VR2.2" />

[[File:fse-ethics-in-fire-safety-engineering-v1.svg|thumb|upright=1.5|alt=An engineer pushed by client, contractor, employer and politics, and supported by integrity, competence and peer review.|Interests that may push a fire safety engineer towards a desired outcome (left) and the safeguards that counter them (right). Based on Van Rossum (2027), §2.2 and §2.5.]]

Giving in to these interests can bring short-term advantages, such as a follow-up commission, a bribe or faster promotion. Resisting them can mean that a career stalls or, in the worst case, that the engineer is pushed out or dismissed.<ref name="VR2.2" /> The book therefore treats integrity as a responsibility that remains even when nobody else can see that advice has been bent.<ref name="VR2.5">Van Rossum (2027), §2.5.</ref>

Watts and Hall describe the same tension in terms of values. Accepting a level of risk requires a value judgement, and technical experts, ordinary citizens, enforcing authorities and managers judge differently.<ref name="WH5-1">Watts and Hall (2002), pp. 5-1–5-2.</ref> An engineer performing a fire risk analysis usually works for a client, but cannot base the analysis on the values of that client alone. The analysis must address both the client's values and the values of society.<ref name="WH5-2">Watts and Hall (2002), p. 5-2.</ref>

[[File:fse-ethics-in-fire-safety-engineering-v2.svg|thumb|upright=1.5|alt=Four quadrants labelled technical, societal, enforcer and managerial value judgements, each with a short description, around a sprinkler symbol.|Four perspectives on the value of a safety measure such as residential sprinklers. An engineer cannot base an analysis on the client's values alone. Based on Watts and Hall (2002), pp. 5-1–5-2.]]

Enforcing authorities are often seen as the interpreters and guardians of those societal values.<ref name="WH5-2" /> Because they bear neither the costs nor the benefits of their decisions, they must also ask when it is acceptable to dictate safety choices to people. Examples are whether it is fair to require sprinklers in all homes, how costs should be shared and who is responsible for the reliability of the systems.<ref name="WH5-2" />

== Competence ==
Competence is the second pillar of professional ethics. ''Introduction to Fire Safety Engineering'' notes that in the Netherlands there are few mechanisms to assure the knowledge and skill of practitioners. There is no degree programme in fire safety in higher education and no established set of competences for the profession, so anyone can call themselves a fire safety adviser.<ref name="VR2.3">Van Rossum (2027), §2.3.</ref> It is therefore the engineer's own responsibility to be competent for the work.

Clients can help by asking for a CV with reference projects, which is usual on large, complex projects.<ref name="VR2.3" /> According to the book, an active role in fire safety engineering needs a higher technical education. People with secondary vocational training can still build the skills needed to check such analyses.<ref name="VR2.3" />

The book warns against overconfidence and names the Dunning–Kruger effect. People who are incompetent at a task tend to overestimate their performance, because their incompetence keeps them from seeing their own errors.<ref name="VR2.4">Van Rossum (2027), §2.4.</ref> The author, in the field since 2001, writes that further study has mainly shown him how much he does not yet know. He also reports meeting practitioners who gave themselves specialist titles after a course of three to five days.<ref name="VR2.4" /> For CFD in particular, the book advises a university background in physics or a related field and much practical experience. It refers back to its chapter on ethics for this point.<ref name="VR11.4">Van Rossum (2027), §11.4.</ref>

Other countries set formal requirements. In Australia, the Fire Safety Verification Method (FSVM) Standard 2022 states that the method may only be used by fire safety engineers who are suitably qualified and experienced. They must have demonstrated competency in fire safety engineering, be proficient in fire modelling methods and be familiar with fire testing and the validation of computational data.<ref name="FSVM5" /> In New Zealand, C/VM2 (second edition, 2025) is intended for professional fire engineers proficient in modelling methods, who are expected to apply appropriate judgement to model inputs and results.<ref name="VM2-7">Ministry of Business, Innovation and Employment (2025b), p. 7.</ref>

The FSVM handbook adds that some jurisdictions register fire safety engineers. National registers recognised in some jurisdictions serve as evidence of suitable qualification and experience.<ref name="HB6">Australian Building Codes Board (2020b), pp. 6–7.</ref> The Australian part of the ''International Fire Engineering Guidelines'' presupposes a level of competence and experience that would enable accreditation by an appropriate body.<ref name="IFEG013">National Research Council of Canada et al. (2005), p. 0.1-3 (Part 0, Australia).</ref>

== Honesty about assumptions and uncertainty ==
A fire safety engineering analysis rests on many assumptions, and ethical practice means making them visible. Notarianni writes that it is important to identify where value judgements enter a performance-based calculation. Assumptions about values, and their effect on the final design, should be made explicit.<ref name="No5-46">Notarianni (2002), p. 5-46.</ref> Decisions that would change if a value, attitude or risk perception changed must be explicit in the design, and agreement on them by all stakeholders is critical.<ref name="No5-46" /> This is a core requirement of [[performance-based design]].

Equivalence is a typical example. One person may consider non-combustible construction equivalent to a sprinkler system because both give enough time to leave the building. Another may disagree because the sprinkler system is less reliable.<ref name="No5-46" /> Designs may be equivalent for life safety but not for property protection or business interruption, so the assumptions behind an equivalence must be stated.<ref name="No5-46" />

In the Netherlands, the regulations do not say how safe is safe enough. Their limits apply to measures, such as travel distances, rather than to goals. Per hour of stay, for example, the risk of dying in a fire is much higher in a residential building than in an assembly building, yet detection requirements are stricter for assembly buildings. The acceptable residual risk must therefore be agreed in advance, for example by comparison with a reference design that meets the rules.<ref name="VR1.2">Van Rossum (2027), §1.2.</ref>

Uncertainty raises similar questions. Notarianni notes that [[uncertainty analysis]] is an uncomfortable topic for many stakeholders. One fear is that it would show that predictions of heat and toxic products are not accurate enough to judge a design with high confidence.<ref name="No5-44">Notarianni (2002), p. 5-44.</ref> She also warns that assumptions meant to be conservative may not be, for example when the reaction time of a resident is set equal to a travel time.<ref name="No5-44" /> ''Introduction to Fire Safety Engineering'' warns against false precision: an available escape time given as 187.4 s suggests an accuracy that the calculation cannot deliver. Two or at most three significant figures are realistic, and results compared with limits should be rounded to the safe side.<ref name="VR1.7">Van Rossum (2027), §1.7.</ref>

Models must also be fit for their use. Cox and Kumar note that the test cases supplied with commercial CFD software validate the technology in general terms only. The responsibility for ensuring validity for a fire application rests with the fire community.<ref name="CK3-203">Cox and Kumar (2002), p. 3-203.</ref> [[Verification and validation]] of models is one way to meet this responsibility. In the United States, the standard NFPA 805 for nuclear power plants requires that fire models are verified and validated and acceptable to the authority having jurisdiction.<ref name="NUREGxv">Najafi et al. (2007), p. xv.</ref>

== Review and independence ==
Independent review is the main safeguard against both error and undue influence. ''Introduction to Fire Safety Engineering'' advises engineers to have their work checked by colleagues.<ref name="VR2.5" /> In Australia, the FSVM Standard 2022 requires the design team to consider at the brief stage whether an independent fire safety engineer should review some or all of the proposed solutions. Where full agreement cannot be reached, dissenting views must be recorded and carried through the approval process.<ref name="FSVM6" />

Independence must be protected. The FSVM handbook states that the approving authority should not take part in design decisions for matters under its jurisdiction. Doing so creates a conflict, or a perceived conflict, of interest, because the authority must be independent and act in the public interest.<ref name="HB38">Australian Building Codes Board (2020b), p. 38.</ref> The same applies to a peer reviewer, who should not be involved in design decisions.<ref name="HB40">Australian Building Codes Board (2020b), p. 40.</ref>

[[File:fse-ethics-in-fire-safety-engineering-v3.svg|thumb|upright=1.5|alt=Flowchart from design team and fire safety engineer through the design brief and analysis to an independent peer reviewer and the approving authority.|Roles in an Australian FSVM project: the peer reviewer and the approving authority comment on the brief but stay out of design decisions. Based on Australian Building Codes Board (2022), p. 6, and Australian Building Codes Board (2020b), pp. 38–40.]]

The Australian part of the IFEG describes the attitude expected of a third party reviewer. The review should be constructive and help the authority approve a design, and it may help the engineer to justify expert judgement properly.<ref name="IFEG037">National Research Council of Canada et al. (2005), pp. 0.3-7–0.3-8 (Part 0, Australia).</ref> Professional detachment, flexibility and an open mind are essential. The reviewer must be independent of the project, keep the documents confidential and carry out check calculations where appropriate.<ref name="IFEG037" /> The conclusions of the review should be documented. Assertions and assumptions must be substantiated and referenced, and the remedial actions needed must be clearly identified.<ref name="IFEG037" />

The investigation of the World Trade Center disaster shows why a documented basis matters. NIST found no technical basis for the choice of the sprayed fire-resistive material used on the long-span floor trusses of the towers, or for its thickness.<ref name="NIST197">National Institute of Standards and Technology (2005), p. 197.</ref> The same report notes that, from 1993, new or modified fire safety systems in the towers required third-party review by a licensed architect or engineer.<ref name="NIST197" />

== Responsibilities of the engineer ==
''Introduction to Fire Safety Engineering'' sums up the responsibility of the fire safety engineer as giving advice that is both competent and honest. This applies despite the incentives and despite the fact that others cannot see whether advice has been compromised.<ref name="VR2.5" /> Dishonest advice may bring short-term gains but disastrous consequences if it leads to casualties. In the Netherlands, an adviser can be prosecuted if an investigation after a fire shows that the advice was demonstrably wrong by the standards of knowledge that could be expected at the time.<ref name="VR2.5" /> There is also the risk of damage to professional reputation.<ref name="VR2.5" />

The point is not limited to fire safety. Writing on structural engineering in the SFPE Handbook, Fitzgerald notes that safety is a major part of the professional responsibility.<ref name="Fi1-132">Fitzgerald (2002), p. 1-132.</ref> Because professionals do not know everything and mistakes remain human, the book recommends five habits. Engineers should dare to question their own knowledge and skill, and have their work checked by peers. They should also join professional associations to exchange knowledge, follow regular training and keep up with publications.<ref name="VR2.5" />

Professional societies such as the Society of Fire Protection Engineers ([[SFPE]]) publish engineering guidance. The Australian part of the IFEG, for example, lists the SFPE engineering guide to performance-based fire protection among its reference documents.<ref name="IFEG013" /> The SFPE Handbook is also one of the main sources of the calculation methods in the Fire Dynamics Tools of the US Nuclear Regulatory Commission.<ref name="IS-iii">Iqbal and Salley (2004), p. iii.</ref>

== See also ==
* [[Prescriptive design]]
* [[Building regulations]]
* [[Acceptance criteria]]
* [[Equivalency]]
* [[Qualitative design review]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* Cox, G., & Kumar, S. (2002). Modeling enclosure fires using CFD. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-194–3-218). National Fire Protection Association; Society of Fire Protection Engineers.
* Fitzgerald, R. W. (2002). Structural mechanics. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-132–1-143). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* Najafi, B., Joglar, F., & Dreisbach, J. (2007). ''Verification and validation of selected fire models for nuclear power plant applications: Vol. 1. Main report'' (NUREG-1824; EPRI 1011999). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Notarianni, K. A. (2002). Uncertainty. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-40–5-64). National Fire Protection Association; Society of Fire Protection Engineers.
* Peacock, R. D., McGrattan, K. B., Forney, G. P., & Reneke, P. A. (2026). ''CFAST – Consolidated Fire and Smoke Transport: Vol. 1. Technical reference guide'' (Version 7, NIST Technical Note 1889v1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.TN.1889v1
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Safety Design & Regulation]]

Aqueous film-forming foam

2026-11-02T17:14:45Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Aqueous film-forming foam''' ('''AFFF''') is a synthetic [[firefighting foam]] whose solution forms a thin watery film that spreads across the surface of a burning hydrocarbon fuel.<ref name="Sch-4-89">Scheffey (2002), p. 4-89.</ref> The concentrate combines fluorine-free hydrocarbon foaming compounds with highly fluorinated surfactants.<ref name="Sch-4-89" /> Mixed with water and aerated, it produces a foam that extinguishes a fire by covering the fuel and stopping it from vaporising, like other foams.<ref name="Sch-4-88">Scheffey (2002), p. 4-88.</ref><ref name="Sch-4-89" /> In addition, the liquid that drains from the bubbles forms a stable film on the fuel, and this film formation is the feature that sets AFFF apart.<ref name="Sch-4-89" />

AFFF is used mainly on fires of flammable liquids, such as fuel spills, and its typical application is the two-dimensional [[pool fire]].<ref name="Sch-4-88" /> In large aviation fire tests it controlled fires at lower application rates than protein foam, and US airports were allowed a 30% reduction in water when AFFF replaced protein foam.<ref name="Sch-4-100">Scheffey (2002), p. 4-100.</ref><ref name="Sch-4-101">Scheffey (2002), p. 4-101.</ref> It can also be discharged through ordinary sprinklers and spray systems, without special foam-making devices.<ref name="IS-B-64">Iqbal and Salley (2004), p. B-64.</ref> Whether a solution can form a film on a given fuel is described by its spreading coefficient, a simple balance of surface tensions.<ref name="Sch-4-93">Scheffey (2002), p. 4-93.</ref> Its limits are those of foam in general. It is less effective on three-dimensional fires of running or sprayed fuel, needs special formulations for alcohols, and must not be used on liquids hotter than 100 °C.<ref name="Sch-4-88" /><ref name="IS-B-60">Iqbal and Salley (2004), p. B-60.</ref>

== Composition and types of foam ==
Fire-fighting foam consists of air-filled bubbles formed from an aqueous solution.<ref name="Sch-4-88" /> The solution is made by mixing a foam concentrate with water, typically at 1, 3 or 6% concentrate.<ref name="Sch-4-88" /> The solution is then aerated to form the bubbles.<ref name="Sch-4-88" /> Foam is lighter than the solution that drains from it and lighter than flammable liquids, so it floats on the fuel.<ref name="Sch-4-88" /> Low-expansion foams, with a foam volume of up to about ten times the solution volume, are the type used on pool fires.<ref name="Sch-4-88" />

There are no universally agreed definitions of foam agents, and definitions differ even between NFPA standards.<ref name="Sch-4-88" /> Scheffey, in the SFPE Handbook, describes the main families as follows. Protein foam is made from hydrolysed proteins, such as hoof and horn meal, with iron salts, and forms no film on the fuel.<ref name="Sch-4-88" /> Fluoroprotein foam adds fluorinated surfactants to a protein base; this lowers the surface tension, but the drained solution still forms no film on hydrocarbons.<ref name="Sch-4-89" /> Aqueous film-forming foam is fully synthetic and is formulated so that its solution has the surface and interfacial tensions needed to spread as a film.<ref name="Sch-4-89" /> Film-forming fluoroprotein foam (FFFP) is a protein foam with enough surfactant added that its drained solution can also spread across a hydrocarbon fuel.<ref name="Sch-4-89" />

The key ingredient of AFFF is a fluorinated surfactant. In a typical molecule, a perfluorooctyl group forms one end, which repels both water and oil, while the other end is attracted to water.<ref name="Sch-4-89" /> In solution these molecules gather at the surface with their water-repelling ends towards the air.<ref name="Sch-4-89" /> Hydrocarbon surfactants in the concentrate make the expanded foam more stable.<ref name="Sch-4-89" /> The combination gives a solution with a surface tension well below that of water and a foam that resists breakdown by heat, fuels and dry chemical agents.<ref name="Sch-4-89" />

The effect on surface tension is large. A 6% AFFF solution has a surface tension of about 15 to 16 dynes/cm at 15 to 32 °C, against 73 dynes/cm for water.<ref name="GW-2-298">Gottuk and White (2002), p. 2-298.</ref> One dyne per centimetre equals one millinewton per metre.<ref name="DiN-A-10">DiNenno et al. (2002), p. A-10.</ref> In unconfined spill tests, water and 6% AFFF solution gave spill depths that bound those of most fuels.<ref name="GW-2-298" />

Several variants exist for special conditions. Regular foams do not perform well on polar solvents such as alcohols, and alcohol-resistant AFFF has been developed for them.<ref name="IS-B-64" /> Such concentrates form a floating polymeric skin that protects the foam from breakdown by solvents such as acetone, methanol and ethanol.<ref name="Sch-4-89" /> Low-temperature foams, including AFFF types, have been developed for use at temperatures as low as −29 °C.<ref name="IS-B-64" />

== How AFFF extinguishes a fire ==
Foams are described in NUREG-1805 as a primary alternative to water, particularly for large fires.<ref name="IS-B-58">Iqbal and Salley (2004), p. B-58.</ref> As an [[extinguishing agent]] they are used mainly on flammable and combustible liquids, the Class B fuels in the [[fire classes]] used by NUREG-1805, and are also suitable for Class A fires of ordinary combustibles.<ref name="IS-B-58" /> Water alone is a poor agent on a burning liquid that is lighter than water and insoluble in it, because the liquid simply floats on the water and continues to burn.<ref name="IS-B-58" /> Foam excludes air from the fire, generates steam that dilutes the air and absorbs heat, penetrates crevices because of its low surface tension, and protects material that is not yet burning.<ref name="IS-B-58" />

The SFPE Handbook notes that the mechanisms of foam extinguishment of pool fires have not been developed from first principles.<ref name="Sch-4-89" /> Extinguishment is usually described as the stopping of fuel vaporisation at the fuel surface.<ref name="Sch-4-89" /> As the vapour supply falls, the combustion zone shrinks, and once the whole area is covered the fire goes out.<ref name="Sch-4-89" /> The floating foam forms an air-excluding layer of aqueous agent that suppresses combustion by halting vaporisation.<ref name="Sch-4-88" /> With AFFF, the solution that drains from the foam adds a film that spreads rapidly and is highly stable on the fuel.<ref name="Sch-4-89" /> NUREG-1805 adds that this kind of film re-forms rapidly if it is disturbed.<ref name="IS-B-63">Iqbal and Salley (2004), p. B-63.</ref>

[[File:fse-aqueous-film-forming-foam-v1.svg|thumb|upright=1.5|alt=Cross-section of a fuel pool: left part covered by a foam blanket over a thin aqueous film, right part still burning with flames above.|AFFF on a pool fire: the foam blanket and the aqueous film draining from it seal the fuel surface, while radiant heat and drainage consume the blanket. Based on Scheffey (2002), pp. 4-88 to 4-91.]]

The foam blanket is continuously lost while it works. Radiant heat from the fire evaporates water from the foam, and liquid drains out of the bubbles.<ref name="Sch-4-90">Scheffey (2002), p. 4-90.</ref> For radiant fluxes of 45 to 185 kW/m² from large pool fires, a simple heat balance gives evaporation rates of 18 to 72 g/(m²·s).<ref name="Sch-4-90" /> In laboratory tests on one AFFF, heat began to penetrate the blanket when it had thinned to about 50 mm.<ref name="Sch-4-91">Scheffey (2002), p. 4-91.</ref> When the fuel below reignited, almost all the foam had been lost, with less than 0.8 kg/m² remaining.<ref name="Sch-4-91" /> Foam must therefore be applied in enough volume and at a high enough rate to make up for these losses and keep a residual layer over the extinguished area.<ref name="IS-B-59">Iqbal and Salley (2004), p. B-59.</ref>

== Spreading coefficient ==
A film-forming foam is defined by the ability of the solution draining from it to spread spontaneously across a hydrocarbon fuel.<ref name="Sch-4-93" /> Whether this happens depends on three tensions: the surface tension of the fuel, the surface tension of the foam solution, and the interfacial tension between the two liquids.<ref name="Sch-4-93" /> The fundamental relation is the spreading coefficient:<ref name="Sch-4-93" />

[[File:fse-aqueous-film-forming-foam-v2.svg|thumb|upright=1.5|alt=Edge of a thin solution layer on fuel with three arrows: fuel surface tension pulling outward, solution surface and interfacial tension pulling back.|The spreading coefficient balances the surface tension of the fuel against the surface tension of the AFFF solution and the interfacial tension; a positive value allows a film to spread. Based on Scheffey (2002), p. 4-93.]]

<math display="block">S_{a/b} = \gamma_b - \gamma_a - \gamma_i</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>S_{a/b}</math> || spreading coefficient of the foam solution (a) on the fuel (b) || dynes/cm
|-
| <math>\gamma_b</math> || surface tension of the lower liquid, the hydrocarbon fuel || dynes/cm
|-
| <math>\gamma_a</math> || surface tension of the upper liquid, the AFFF solution || dynes/cm
|-
| <math>\gamma_i</math> || interfacial tension between liquids a and b || dynes/cm
|}

A positive spreading coefficient means that a film can form; a negative value means it cannot.<ref name="Sch-4-93" /><ref name="Sch-4-94">Scheffey (2002), p. 4-94.</ref> The tensions are measured with a ring tensiometer, and reagent-grade cyclohexane is the usual reference fuel.<ref name="Sch-4-93" /> As the fuel temperature rises, the surface tensions of both fuel and solution fall, and the spreading coefficient may drop to zero or below.<ref name="Sch-4-93" /> Because the surface tensions of most AFFFs are about the same, the interfacial tension is usually the deciding factor.<ref name="Sch-4-93" />

As a worked example, take the AFFF solution FC-194 tested by the US Naval Research Laboratory, with a surface tension of 15.5 dynes/cm at 25 °C.<ref name="Sch-4-94" /> On cyclohexane, with a surface tension of 24.2 dynes/cm and an interfacial tension of 4.3 dynes/cm, the spreading coefficient is 24.2 − 15.5 − 4.3 = 4.4 dynes/cm, and a film formed.<ref name="Sch-4-93" /><ref name="Sch-4-94" /> On certified n-heptane, with 19.8 and 5.5 dynes/cm, it is 19.8 − 15.5 − 5.5 = −1.2 dynes/cm, and no film formed.<ref name="Sch-4-93" /><ref name="Sch-4-94" /> On JP-5 jet fuel, the result is +5.2 dynes/cm.<ref name="Sch-4-94" /> Fuels with a low surface tension, such as avgas and n-heptane at 19 to 20 dynes/cm, are therefore the hardest for a film to cover.<ref name="Sch-4-93" />

[[File:fse-aqueous-film-forming-foam-v3.svg|thumb|upright=1.5|alt=Bar chart of spreading coefficients of one AFFF solution on seven fuels, from −1.2 for n-heptane to +5.2 for JP-5.|Spreading coefficient of AFFF solution FC-194 (15.5 dynes/cm) on seven hydrocarbon liquids at 25 °C; negative values gave no film. Based on Scheffey (2002), pp. 4-93 to 4-94.]]

The spreading coefficient test is used worldwide as a standard indicator of film-forming foams.<ref name="Sch-4-93" /> It does not, however, predict fire performance. Studies have shown no direct correlation between the spreading coefficient and fire extinguishment, so the coefficient alone cannot rank foams.<ref name="Sch-4-93" /> Formulators must also keep a balance: adding more fluorinated surfactant lowers the interfacial tension but can reduce the foam's ability to shed fuel.<ref name="Sch-4-93" />

== Fire tests and application rates ==
Since there is no theory of foam spreading, foam performance is measured with fire tests.<ref name="Sch-4-94" /> In the US standard UL 162, commercial n-heptane burns in a square pan of at least 4.6 m² for 60 s before foam is applied.<ref name="Sch-4-95">Scheffey (2002), p. 4-95.</ref> For hand-held nozzles, protein-based foams are tested at 2.5 L/(min·m²) for 5 minutes, while AFFF and FFFP are tested at 1.6 L/(min·m²) for 3 minutes.<ref name="Sch-4-95" /> After extinguishment, a burnback test checks that the foam blanket restricts a reignited area to no more than 0.9 m² for 5 minutes, or flows back over it.<ref name="Sch-4-95" />

Application rates in test standards are usually lower than those used in practice.<ref name="Sch-4-100" /> Practice adds a safety factor, while critical rates bring out the differences between agents.<ref name="Sch-4-100" /> Large aviation tests by Geyer with JP-4 pool fires of 21 to 43 m diameter illustrate this.<ref name="Sch-4-100" /> To control a fire within 60 s, AFFF needed about 1.6 to 2.4 L/(min·m²), while protein foam needed 3.3 to 4.1 L/(min·m²).<ref name="Sch-4-100" /> Above about 4.1 L/(min·m²) for AFFF and 8.2 L/(min·m²) for protein foam, higher rates hardly improved control times.<ref name="Sch-4-100" /> Below a critical application rate, a fire is not extinguished however long the agent is applied.<ref name="Sch-4-100" />

Aviation fire protection is built on these results. An aluminium fuselage exposed to a fuel spill fire burns through in about 1 minute, so rescue vehicles must control the fire within 1 minute of arriving.<ref name="Sch-4-103">Scheffey (2002), p. 4-103.</ref> The resulting design application rates are 5.5 L/(min·m²) for AFFF, 7.5 L/(min·m²) for fluoroprotein foam and 8.2 L/(min·m²) for protein foam.<ref name="Sch-4-103" /> Foam quality is also specified: under NFPA 412, air-aspirated AFFF must have an expansion ratio of at least 5:1 and a 25% drainage time of at least 2.25 minutes.<ref name="Sch-4-103" /> Longer drainage times generally give longer burnback times, but no direct correlation has been found between expansion, drainage and extinguishing performance.<ref name="Sch-4-104">Scheffey (2002), p. 4-104.</ref>

[[File:fse-aqueous-film-forming-foam-v4.svg|thumb|upright=1.5|alt=Bar chart of aviation design application rates: AFFF 5.5, fluoroprotein 7.5 and protein foam 8.2 litres per minute per square metre.|Design application rates for aircraft rescue and fire fighting by foam type. Based on Scheffey (2002), p. 4-103.]]

== Fixed systems and aircraft hangars ==
AFFF has a low viscosity and does not need elaborate foam-making devices, so it can be used in many water sprinkler and water spray systems.<ref name="IS-B-64" /> Overhead foam-water sprinkler systems serve two purposes: they control or suppress a fuel spill fire, and when the foam runs out they continue to cool with water.<ref name="Sch-4-112">Scheffey (2002), p. 4-112.</ref> In the US, NFPA 16 requires a minimum application rate of 6.6 L/(min·m²) for deluge and closed-head AFFF systems.<ref name="Sch-4-112" /> This rate includes a safety factor over the 4.1 L/(min·m²) at which AFFF from sprinklers is effective on pool fires.<ref name="Sch-4-112" />

Much of the test data for foam-water sprinkler systems was originally developed for aircraft hangars.<ref name="Sch-4-112" /> Hangars have traditionally been protected by deluge systems with open heads, activated by rapid-response detection.<ref name="Sch-4-106">Scheffey (2002), p. 4-106.</ref> Factory Mutual tested 83.6 m² JP-4 fires in an 18.3 m high space.<ref name="Sch-4-107">Scheffey (2002), p. 4-107.</ref> At 6.6 L/(min·m²), standard sprinklers discharging AFFF were 1.3 to 1.6 times as effective as air-aspirating foam-water sprinklers.<ref name="Sch-4-107" />

The explanation lies in the fire plume. It rose at up to 23.2 m/s, while foam particles fell at no more than about 9.1 m/s.<ref name="Sch-4-107" /> The denser droplets from standard sprinklers penetrated the plume better than the light, air-aspirated foam.<ref name="Sch-4-107" /> The US hangar standard NFPA 409 therefore allows 6.6 instead of 8.2 L/(min·m²) when standard sprinklers are used with AFFF, a 20% reduction in foam.<ref name="Sch-4-108">Scheffey (2002), p. 4-108.</ref>

Low-level systems avoid the plume problem altogether. Monitors, turret nozzles and floor or side-mounted nozzles deliver AFFF directly onto the fuel, at high local densities.<ref name="Sch-4-108" /> Such systems have been used for over two decades on US Navy ships that carry aircraft, protecting flight decks and special hazard areas.<ref name="Sch-4-108" />

== Limits and hazards ==
AFFF shares the general limits of foam. Low-expansion foams are effective on two-dimensional pool fires, but not on three-dimensional fires, especially those of low-flashpoint fuels.<ref name="Sch-4-88" /> Where running fuel or a pressurised spray is expected, an auxiliary agent such as dry chemical is used together with the foam.<ref name="Sch-4-88" /> Foam is an unstable mixture of air and water that can be broken down by mechanical forces, certain chemicals and some other extinguishing agents.<ref name="IS-B-58" /> Rupture of the foam blanket can lead to burnback, which may put firefighters at risk.<ref name="IS-B-60" />

Hot liquids are a particular hazard. If water penetrates an oil or fat that is well above the boiling point of water, it turns to steam below the surface and can cause an eruption of burning oil, known as [[boilover]].<ref name="IS-B-58" /> Because foam is water-based, its use on a liquid at 100 °C or above can cause a four-fold expansion of the foam, with boilover of the burning liquid.<ref name="IS-B-60" /> Foam should only be used on a horizontal liquid surface below its boiling point, and, like water, it is unsuitable for live electrical equipment or substances that react with water.<ref name="IS-B-60" /><ref name="IS-B-61">Iqbal and Salley (2004), p. B-61.</ref>

Engineering knowledge of AFFF is also limited. According to Scheffey, modelling of foam extinguishment cannot yet be performed, because too many parameters, such as foam loss rates and foam viscosity, have not been quantified.<ref name="Sch-4-92">Scheffey (2002), p. 4-92.</ref> Design therefore relies on rates from fire tests with added safety factors.<ref name="Sch-4-100" /><ref name="Sch-4-112" /> Environmental impact, together with reliability and false discharges of foam-water deluge systems, was among the reasons why the US Navy re-evaluated its approach to hangar fire protection.<ref name="Sch-4-110">Scheffey (2002), p. 4-110.</ref>

== See also ==
* [[Deluge system]]
* [[Extinguishing mechanisms]]
* [[Sprinkler system]]
* [[Fire suppression system]]

== Notes ==
<references />

== References ==
* DiNenno, P. J., Drysdale, D., Beyler, C. L., Walton, W. D., Custer, R. L. P., Hall, J. R., Jr., & Watts, J. M., Jr. (Eds.). (2002). ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association; Society of Fire Protection Engineers.
* Gottuk, D. T., & White, D. A. (2002). Liquid fuel fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-297–2-316). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Scheffey, J. L. (2002). Foam agents and AFFF system design considerations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-88–4-122). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Suppression]]

Stack effect

2026-11-02T17:14:42Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''stack effect''' is the movement of air or smoke through a tall building caused by pressure differences that arise from temperature differences.<ref name="IS-F61">Iqbal and Salley (2004), p. F-61.</ref> When the air inside a building is warmer than the air outside, it is lighter, rises through vertical shafts such as stairs and lift shafts, and leaves through openings high in the building. Cold outside air is drawn in low down. This is called the normal stack effect. When the outside air is warmer, the flow reverses and runs downwards, which is called the reverse stack effect.<ref name="KL-274">Klote (2002), p. 4-274.</ref> It is also called draught or chimney effect and is best known from chimneys.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref><ref name="EV-70">Evans et al. (2005), p. 70.</ref>

[[File:fse-stack-effect-v1.svg|thumb|upright=1.5|alt=Two building sections with a central shaft: left, air enters low, rises in the shaft and leaves high; right, air enters high, falls in the shaft and leaves low.|Normal stack effect in cold weather (left) and reverse stack effect in warm weather (right). The neutral plane is where the inside and outside pressures are equal. Based on Klote (2002), pp. 4-274 to 4-275.]]

The stack effect is one of the main driving forces of [[smoke movement]] in buildings, together with the buoyancy of hot smoke, expansion, wind and the ventilation system.<ref name="KL-274" /> It operates even without a fire and acts on the whole height of a building. Its strength grows with the height of the building and with the difference between inside and outside temperature.<ref name="KL-274" /> In a [[high-rise building]] in cold weather, it can carry smoke from a fire on a low floor to floors far above. During the 1993 bombing of the World Trade Center, it was a primary driver of rapid smoke spread.<ref name="KL-276">Klote (2002), p. 4-276.</ref><ref name="FS-xxxiv">Ferreira and Strege (2005), p. xxxiv.</ref> Engineers therefore account for the stack effect when they design smoke control systems, which must keep working whatever the weather.<ref name="KL-285">Klote (2002), p. 4-285.</ref>

== Physical principle ==
The stack effect is a form of [[buoyancy]]. When air is heated, its density falls, so a column of warm air weighs less than a column of cold air of the same height. The heavier cold air pushes the lighter warm air upwards.<ref name="VR-3.2">Van Rossum (2027), §3.2.</ref> The book names buoyancy as the driving force behind smoke plumes, the stack effect and natural smoke exhaust. The larger the difference in density and the greater the height, the larger the pressure difference in general.<ref name="VR-3.2" />

The density of air follows from the [[ideal gas law]]. At atmospheric pressure it is approximately 353 divided by the absolute temperature in kelvin, which gives about 1.2 kg/m³ at 20 °C.<ref name="VR-3.2" /> The pressure in still air increases with depth, by the weight of the air above. Inside and outside a building, the pressure therefore changes with height at slightly different rates, because the two air columns have different densities.<ref name="VR-3.2" />

At one height, the pressure inside equals the pressure outside. This level is called the [[neutral plane]].<ref name="KL-275">Klote (2002), p. 4-275.</ref> In the normal stack effect, the inside pressure is higher than outside above the neutral plane, so air flows out there. Below the neutral plane, the inside pressure is lower, so outside air flows in.<ref name="KL-275" /> The book describes the same pattern for a room with openings: air flows in below the neutral plane and smoke flows out above it.<ref name="VR-3.2" />

A chimney strengthens the effect because no horizontal mixing can take place inside it.<ref name="VR-6.6" /> In buildings, the stairs, lift shafts, service shafts and mail chutes act as chimneys. Klote notes that the normal stack effect matters most at low outside temperatures and in tall shafts, but it can even exist in a one-storey building.<ref name="KL-274" />

== Pressure difference ==
The pressure difference grows linearly with the distance from the neutral plane and with the difference between the reciprocals of the two absolute temperatures. Klote gives it for a shaft and the outside air as follows:<ref name="KL-274" /><ref name="KL-275" />

<math display="block">\Delta p = K_\mathrm{s} \left( \frac{1}{T_\mathrm{o}} - \frac{1}{T_\mathrm{i}} \right) h</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>\Delta p</math> || pressure difference between the shaft and the outside || Pa
|-
| <math>K_\mathrm{s}</math> || coefficient, 3460 at standard atmospheric pressure || Pa·K/m
|-
| <math>T_\mathrm{o}</math> || absolute temperature of the outside air || K
|-
| <math>T_\mathrm{i}</math> || absolute temperature of the air inside the shaft || K
|-
| <math>h</math> || height above the neutral plane (negative below it) || m
|}

The relation holds at standard atmospheric pressure and requires absolute temperatures.<ref name="KL-274" /> It follows from the density difference of the two air columns. The coefficient of 3460 is close to the gravitational acceleration multiplied by the constant 353 in the book's density relation.<ref name="VR-3.2" /> For the normal stack effect, the pressure difference is positive above the neutral plane and negative below it. For the reverse stack effect, the signs are swapped.<ref name="KL-275" />

The position of the neutral plane depends on where the building leaks. If the leakage paths are spread evenly over the height, the neutral plane lies near mid-height. Where leakage is uneven, for example in shafts vented at the top, it can lie much higher or lower.<ref name="KL-275" /><ref name="KL-276" /> The pressure difference therefore grows towards the top and the bottom of the building and is largest at the ends of the shafts.<ref name="KL-275" />

[[File:fse-stack-effect-v2.svg|thumb|upright=1.5|alt=Straight line of shaft-to-outside pressure difference against height in a 60 m building, from minus 54 Pa at the bottom to plus 54 Pa at the top, zero at 30 m.|Pressure difference between a shaft and the outside over the height of a 60 m building with the neutral plane at mid-height, at −18 °C outside and 21 °C inside. Based on Klote (2002), p. 4-275.]]

=== Worked example ===
Klote considers a building 60 m tall with the neutral plane at mid-height, so 30 m above the neutral plane at the top. The outside temperature is −18 °C (255 K) and the inside temperature is 21 °C (294 K).<ref name="KL-275" /> The pressure difference is then 3460 × (1/255 − 1/294) × 30 ≈ 54 Pa. Klote gives 55 Pa from the same calculation in US units.<ref name="KL-275" /> At the top, the shaft is about 55 Pa above the outside pressure; at the bottom, it is about 55 Pa below it.<ref name="KL-275" />

For the same temperatures, the pressure difference increases by about 1.8 Pa for every metre of height from the neutral plane. A 120 m building with its neutral plane at mid-height would therefore see about 108 Pa at its top and bottom. These values are large compared with the pressures used in smoke control. Klote suggests a minimum design pressure difference of 20 to 25 Pa across a barrier next to an intense fire.<ref name="KL-285" /> Codes for pressurisation in the United States require 12.5 Pa in sprinklered and 25 Pa in unsprinklered buildings.<ref name="EV-87">Evans et al. (2005), p. 87.</ref> The buoyancy of the fire gases themselves gives about 13 Pa at 1.52 m above the neutral plane of a fire room at 800 °C.<ref name="KL-276" />

[[File:fse-stack-effect-v3.svg|thumb|upright=1.5|alt=Falling line of pressure difference per metre against outside temperature from minus 40 to plus 40 °C, crossing zero at 21 °C.|Stack effect pressure difference per metre of height from the neutral plane as a function of outside temperature, for an inside temperature of 21 °C. Positive values mean normal, negative values reverse stack effect. Based on Klote (2002), pp. 4-274 to 4-275.]]

Very tall buildings reach much higher values. For the World Trade Center towers, NIST used the New York winter design temperature of −9.4 °C (15 °F). At that temperature, the stack effect pressures were of the order of 280 Pa at the bottom and top of the towers.<ref name="FS-14">Ferreira and Strege (2005), p. 14.</ref> NIST noted that this is an order of magnitude higher than the design pressure for unsprinklered buildings, so the stack effect can contribute significantly to smoke spread in tall buildings.<ref name="FS-14" /> The towers also influenced their surroundings. Under extreme weather conditions, air that entered the PATH railway tunnel in New Jersey left through the top of the towers.<ref name="FS-61">Ferreira and Strege (2005), p. 61.</ref>

== Effect on smoke movement ==
In a building with the normal stack effect, the existing air currents can move smoke far from the fire.<ref name="KL-276" /> If the fire is below the neutral plane, smoke enters the shafts with the building air and rises, helped by its own buoyancy. Above the neutral plane it flows out of the shafts into the upper floors. If there is little leakage between floors, the floors below the neutral plane, apart from the fire floor, stay relatively free of smoke for a time.<ref name="KL-276" /> If the fire is above the neutral plane, smoke is carried to the outside through openings in the facade. Where floors leak considerably, smoke also moves up to the floor above the fire.<ref name="KL-276" />

[[File:fse-stack-effect-v4.svg|thumb|upright=1.5|alt=Section of a tall building: fire on a low floor, smoke entering a shaft, rising and spreading onto the upper floors above the neutral plane.|Smoke from a fire below the neutral plane is drawn into the shafts by the normal stack effect and spreads onto floors above the neutral plane. Based on Klote (2002), p. 4-276.]]

The 1993 bombing of the World Trade Center shows how strong the effect can be. The explosion cut electrical power, the ventilation systems stopped, and the outside temperature was in the twenties degrees Fahrenheit, below freezing.<ref name="FS-xxxiv" /> Openings created in lift shafts on the lower levels let smoke rise rapidly to the upper parts of the tower. NIST concluded that the stack effect could be a primary driver of rapid smoke spread in high-rise buildings, and that openings in vertical shafts strengthen it.<ref name="FS-xxxiv" /><ref name="FS-xxxv">Ferreira and Strege (2005), p. xxxv.</ref> Stair doors held open by evacuating occupants increased the airflow up the stairs.<ref name="FS-xxxv" /> On 11 September 2001, fresh air drawn up from lower floors by the chimney effect also supplied oxygen to fires deep inside the towers.<ref name="MG-52">McGrattan et al. (2005), p. 52.</ref>

The reverse stack effect tends to move relatively cool smoke downwards. Hot smoke, however, can be buoyant enough to rise even under reverse stack effect conditions.<ref name="KL-276" /> In very airtight buildings with exterior stairs, it has been observed even at low outside temperatures, because the stair was colder than the rest of the building.<ref name="KL-275" /> New York fire service procedures note that the stack effect becomes noticeable in buildings over 60 feet (about 18 m) high and grows with height. Downward flow is most likely when the outside temperature is above 90 °F (about 32 °C).<ref name="LV-I">Lawson and Vettori (2005), Appendix I.</ref> The same procedures state that wind intensifies the effect.<ref name="LV-I" />

== Stack effect and smoke control ==
A smoke control system has to work against the stack effect that exists at the moment it operates. If the outside temperature falls below the winter design temperature, the system may not perform as intended.<ref name="KL-285" /> Klote points to the 99 % and 97.5 % winter design temperatures of the ASHRAE Handbook. In a normal winter, the temperature is at or below these values for about 22 and 54 hours respectively, and extremes can be much lower.<ref name="KL-285" /> Heating systems tolerate such short cold spells because building materials respond slowly. A smoke control system has no such time lag and must cope with the full stack effect at the moment it operates.<ref name="KL-285" />

[[Stair pressurisation]] is particularly sensitive. In the World Trade Center towers, studies after 1993 found that pressurised stairs needed relief venting at the top of each stair section. Without it, the stack effect would produce unacceptable door opening forces there.<ref name="FS-xxxvi">Ferreira and Strege (2005), p. xxxvi.</ref> Mechanically exhausting the stairs reduced smoke on the top floors. However, it increased the stack effect and drew more smoke into the stairs on the lower floors.<ref name="FS-xxxvi" />

NIST's analysis of alternative smoke management systems for the towers came to similar conclusions. Stack effect had a substantial impact on zoned smoke control combined with stair pressurisation, in some cases causing airflow from the fire floor into the core.<ref name="EV-90">Evans et al. (2005), p. 90.</ref> Fan speeds would probably have to be adjusted to measured pressure differences. Sandwich pressurisation, which exhausts the fire floor and pressurises the floors around it, remained effective even with the stack effect.<ref name="KL-274" /><ref name="EV-91">Evans et al. (2005), p. 91.</ref>

Fire services also use the stack effect deliberately. New York procedures describe supplying air with portable fans into the attack stair three floors below the fire, with the doors at top and bottom closed. This pressurises the stair near the fire floor and reduces the stack effect. With the top and bottom doors open, the same fans increase the stack effect and help to remove smoke through the stair.<ref name="LV-I" /> The procedures warn that using a stair for smoke removal can draw heat and flames from the fire floor towards that stair.<ref name="LV-I" />

== Calculation methods ==
Simple cases can be calculated by hand with the relation above.<ref name="KL-274" /> The International Fire Engineering Guidelines describe the stack effect as buoyancy-induced pressure differences caused by inside and outside temperature differences, and refer to Klote and Milke for the equations.<ref name="IFEG-2.5-6">National Research Council of Canada et al. (2005), p. 2.5-6.</ref> They also list the season, wind and the stack effect among the external conditions to consider when defining fire scenarios.<ref name="IFEG-2.2-3">National Research Council of Canada et al. (2005), p. 2.2-3.</ref>

Whole buildings are analysed with smoke movement models of the [[network model]] type. They represent a building as a network of spaces or nodes, each at its own pressure and temperature. Stairs and other shafts are modelled as a vertical series of spaces, one for each floor.<ref name="IS-B82">Iqbal and Salley (2004), p. B-82.</ref> Air flows through leakage paths from high to low pressure. The inputs include the inside and outside temperatures, the flow network and the flows of the ventilation or smoke control system.<ref name="IS-B82" />

NUREG-1805 notes that these models can represent the stack effect in tall buildings during extreme temperatures. ASCOS, CONTAMW, AIRNET and ASMET are examples, and NIST used CONTAM for the World Trade Center analyses.<ref name="IS-B82" /><ref name="EV-87" /> In zone models of single spaces, the position of openings relative to the neutral plane decides whether they act as inlets or outlets. Openings high in a facade are not necessarily outlets, and an opening that spans the neutral plane can act as both.<ref name="VR-9.3a">Van Rossum (2027), §9.3 (How a zone model works).</ref>

== Limitations and misconceptions ==
The relation for the pressure difference is a simplification. In real buildings, the shape and position of openings, flow resistance, wind and mechanical ventilation all influence the pressure differences and flows.<ref name="VR-3.2" /> The neutral plane is not fixed: it moves when doors or windows open or break.<ref name="KL-275" /><ref name="KL-276" /> Wind can strengthen or weaken the effect.<ref name="LV-I" />

A common misconception is that the stack effect matters only in skyscrapers. It increases with height, but it also exists in low buildings.<ref name="KL-274" /> Another is that it always moves smoke upwards. In hot weather, or with a cold exterior stair, the flow in the shafts can run downwards.<ref name="KL-275" />

The stack effect should also not be confused with the buoyancy of the fire gases themselves. Both rest on the same physics. The stack effect is driven by the normal temperature difference between the building and the outside air.<ref name="KL-274" /> Fire buoyancy is driven by the much higher smoke temperature and weakens with distance from the fire.<ref name="KL-276" />

== See also ==
* [[Smoke and heat exhaust ventilation]]
* [[Smoke barrier]]
* [[Smoke control]]

== Notes ==
<references />

== References ==
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Ferreira, M. J., & Strege, S. M. (2005). ''Smoke management systems'' (NIST NCSTAR 1-4D). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Lawson, J. R., & Vettori, R. L. (2005). ''The emergency response operations'' (NIST NCSTAR 1-8). National Institute of Standards and Technology.
* McGrattan, K. B., Bouldin, C., & Forney, G. P. (2005). ''Computer simulation of the fires in the World Trade Center towers'' (NIST NCSTAR 1-5F). National Institute of Standards and Technology.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Smoke Control]]

Protected escape route

2026-11-02T17:14:40Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''protected escape route''' is the part of an escape route that is separated from the rest of the building by fire- or smoke-resisting construction, or by open air. Occupants in it can reach a final exit without being exposed to the fire. In New Zealand, C/VM2 (second edition, 2025) calls these protected parts exitways and safe paths. In Australia they are known as fire-isolated exits, such as fire-isolated stairs and passageways.<ref name="MBIE25-63">Ministry of Business, Innovation and Employment (2025b), p. 63.</ref><ref name="MBIE25-69">Ministry of Business, Innovation and Employment (2025b), p. 69.</ref><ref name="ABCB-D1-246">Australian Building Codes Board (2020a), Data Sheet D1, p. 246.</ref> The protected part follows the open path through the room or floor where the fire may be, and it ends at a final exit to a safe place outside.<ref name="MBIE25-67">Ministry of Business, Innovation and Employment (2025b), p. 67.</ref>

Protected escape routes are a core element of the [[means of escape]] in multi-storey buildings. They rely on [[fire compartmentation]], on [[fire door]]s that close behind the occupants and, in many tall buildings, on [[stair pressurisation]] or other [[smoke control]] systems. The aim is that people who have left the open path need not walk through smoke, because smoke is what most often prevents escape.<ref name="VR-7">Van Rossum (2027), §7.</ref> Design rules therefore set requirements for the fire resistance of the enclosure, the doors, the linings, the separation between alternative routes and the [[travel distance]] to the protected route.<ref name="MBIE25-37">Ministry of Business, Innovation and Employment (2025b), p. 37.</ref><ref name="MBIE25-51">Ministry of Business, Innovation and Employment (2025b), p. 51.</ref>

== Definition and terminology ==
C/VM2 (second edition, 2025) defines an escape route as a continuous, unobstructed route from any occupied space to a final exit, so that occupants can reach a safe place.<ref name="MBIE25-63" /> An escape route consists of open paths, safe paths or both.<ref name="MBIE25-63" /> An open path is the part within a firecell where occupants may be exposed to fire or smoke while they escape.<ref name="MBIE25-67" /> The exitway is the part of the escape route protected by fire or smoke separations, or by distance where it is open to the outside air, and it ends at a final exit.<ref name="MBIE25-63" /> A safe path is the part of an exitway that is protected from the effects of fire by fire separations, external walls or distance in the open air.<ref name="MBIE25-69" />

[[File:fse-protected-escape-route-v1.svg|thumb|upright=1.5|alt=Floor plan: a dashed route crosses an office firecell (open path), enters a fire-separated corridor and stair (exitway) and ends at a final exit.|The parts of an escape route: the open path inside the firecell and the protected exitway that leads to the final exit. Based on Ministry of Business, Innovation and Employment (2025b), pp. 63, 67 and 69.]]

The final exit is the point where the escape route gives direct access to a safe place, such as a street or open space where people can disperse.<ref name="MBIE25-63" /><ref name="MBIE25-69" /> C/VM2 also allows a place of safety inside a building. Such a place must be enclosed by fire separations that withstand burnout, lie in a sprinklered building, hold the intended number of people and have its own means of escape to the outside.<ref name="MBIE25-68">Ministry of Business, Innovation and Employment (2025b), p. 68.</ref>

The Australian data sheets for the Fire Safety Verification Method use the term "protected exit", with a fire-isolated stair or passageway as examples.<ref name="ABCB-D1-246" /> The Building Code of Australia (BCA) states the matching performance requirement. As quoted in the International Fire Engineering Guidelines (2005), exits must be fire-isolated "to the degree necessary", depending on the storeys connected, the fire safety systems, the use of the building and fire brigade intervention.<ref name="IFEG-AU-0.2-5">National Research Council of Canada et al. (2005), Part 0 – Australia, p. 0.2-5.</ref> In the Netherlands, maintaining the escape routes for building users is one of the sub-goals of the fire safety rules in the Bbl, the Dutch building regulations.<ref name="VR-1.2">Van Rossum (2027), §1.2.</ref>

== Why escape routes are protected ==
In a fire, escape is usually prevented because hot smoke makes visibility poor, although temperature and toxicity can also play a part.<ref name="VR-7" /> Visibility inside a smoke layer can fall very quickly. In an example in the ''Introduction to Fire Safety Engineering'', a lit exit sign is visible from only 0.57 m two minutes after the fire starts. Escaping through such a layer is not a realistic option.<ref name="VR-6.5">Van Rossum (2027), §6.5.</ref> Several forces drive smoke into other spaces. The expanding gases push smoke through seams, gaps and holes, and stack effect, wind and building services spread it further.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref>

Smoke often travels far from the fire. Klote notes that stairs and lift shafts frequently become smoke-logged, which blocks or hinders evacuation.<ref name="Klote-274">Klote (2002), p. 4-274.</ref> Proulx describes how stairs in high-rise buildings become contaminated when occupants of the fire floor and of upper floors open the stair doors during a simultaneous evacuation.<ref name="Proulx-362">Proulx (2002), p. 3-362.</ref> This happens even if the stair is pressurised. Proulx concludes that the stairs may be the most dangerous place in a high-rise fire after the room of origin and the fire floor.<ref name="Proulx-362" /> In the 1993 bombing of the World Trade Center, 94 percent of the occupants of the first tower moved through smoke.<ref name="Proulx-362" />

A protected route is designed to remain usable for as long as occupants need it. In New Zealand, C/VM2 (second edition, 2025) requires tenability to be assessed in the escape routes over the period in which occupants are escaping.<ref name="MBIE25-57">Ministry of Business, Innovation and Employment (2025b), p. 57.</ref> The tenability criteria are the fractional effective dose of carbon monoxide, the fractional effective dose of heat and the visibility, all measured 2.0 m above the floor.<ref name="MBIE25-17">Ministry of Business, Innovation and Employment (2025b), p. 17.</ref> The explanatory notes to the Dutch Bbl give comparable criteria. For a non-enclosed space through which an escape route runs, they include a smoke-free height of 2.5 m above the route and at most 2 kW/m² of radiation below the smoke layer.<ref name="VR-7.7">Van Rossum (2027), §7.7.</ref>

== Enclosure and construction ==
The enclosure of a protected route is formed by fire separations and smoke separations. Klote explains that the effectiveness of such barriers depends on their leakage paths, such as cracks around doors and holes for pipes, and on the pressure difference across them.<ref name="Klote-278">Klote (2002), p. 4-278.</ref> For smoke modelling, C/VM2 assumes that fire and smoke separations have no leakage apart from their doors.<ref name="MBIE25-17" /> A smoke control door is assumed to leak through a 10 mm gap under the door. A fire door without smoke seals is modelled with a 10 mm gap over its full height.<ref name="MBIE25-17" />

The required fire resistance depends on the building and on the scenario. In the challenging fire scenario of C/VM2, the floors, stairs and walkways that form escape routes must keep their structural adequacy, integrity and insulation for the whole required safe egress time.<ref name="MBIE25-56">Ministry of Business, Innovation and Employment (2025b), p. 56.</ref> The fire and smoke separations that protect these routes must do the same.<ref name="MBIE25-56" /> When this construction is assessed, the effect of sprinklers in controlling the fire must be ignored.<ref name="MBIE25-56" /> For firefighting access, buildings with an escape height above 10 m need stairs designed as exitways, fire separated from the rest of the building and able to resist fire spread until burnout.<ref name="MBIE25-55">Ministry of Business, Innovation and Employment (2025b), p. 55.</ref>

The materials inside the route also matter. C/VM2 sets its strictest wall and ceiling lining class, Group 1-S, for exitways in unsprinklered buildings, the same as for sleeping areas where care is provided.<ref name="MBIE25-51" /> In most buildings, floor coverings in exitways must have a critical radiant flux of at least 2.2 kW/m².<ref name="MBIE25-52">Ministry of Business, Innovation and Employment (2025b), p. 52.</ref> The commentary explains that linings in exitways may be the largest fire load in those spaces, while the spaces themselves are critical paths for escape.<ref name="MBIE25-52" />

Openings in the enclosure are a weak point. The International Fire Engineering Guidelines give the example of a window in the wall between an apartment and an internal stair in a multi-storey residential building in Australia.<ref name="IFEG-AU-0.2-5" /> Such a window does not comply with the deemed-to-satisfy provisions, because the wall no longer has the required fire rating.<ref name="IFEG-AU-0.2-5" /> The structural strength of the enclosure can matter as well.

In the World Trade Center towers, the stairs were enclosed by cast gypsum panels covered with gypsum board.<ref name="NIST-12">National Institute of Standards and Technology (2005), p. 12.</ref> These enclosures were fire rated but not required to withstand accidental loads. On 11 September 2001, all stairs in the impact zone of the first tower were destroyed, and debris from damaged enclosures obstructed stairs in both towers.<ref name="NIST-xl">National Institute of Standards and Technology (2005), pp. xxxix–xl.</ref>

== Doors ==
Doors are the most frequent openings in a protected route and must close again after use. C/VM2 assumes that fire doors and smoke control doors with self-closers are closed unless occupants are using them.<ref name="MBIE25-15">Ministry of Business, Innovation and Employment (2025b), p. 15.</ref> When few people use a door, it is taken to be open for 3 s per occupant. When queuing is expected, the door is assumed open for the whole time of queuing.<ref name="MBIE25-15" /> If 20 people pass a door one by one, it is therefore open for 60 s in total. Open doors count as half their width in smoke flow calculations.<ref name="MBIE25-15" />

In practice, doors do not always work as intended. In a New Zealand survey quoted in the Australian data sheets, 4.8 percent of fire doors and 10.3 percent of smoke control doors had missing smoke seals.<ref name="ABCB-C7-215">Australian Building Codes Board (2020a), Data Sheet C7, p. 215.</ref> About 2 percent of the doors were wedged or blocked, and 2.9 percent did not close fully.<ref name="ABCB-C7-215" /> Estimated reliabilities in the same data sheet range from about 65 to 93 percent, with a mean of about 80 percent. The British draft DD240 gives 90 percent for a self-closing door to a protected stair.<ref name="ABCB-C7-215" />

[[File:fse-protected-escape-route-v4.svg|thumb|upright=1.5|alt=Bar chart of estimated fire door reliability from nine surveys, from 66 % in a UK study to 92.6 % in US nuclear data, mean about 80 %.|Estimated reliability of fire and smoke doors from surveys and guidance documents. Based on Australian Building Codes Board (2020a), Data Sheet C7, p. 215.]]

C/VM2 therefore treats fire doors and smoke control doors as key fire safety systems, whose failure must be checked in a separate design scenario. Doors on hold-open devices released by the fire alarm in sprinklered buildings are an exception.<ref name="MBIE25-58">Ministry of Business, Innovation and Employment (2025b), p. 58.</ref> In that scenario, each key system is assumed to fail in turn, and the design must still show that the available safe egress time exceeds the required time.<ref name="MBIE25-59">Ministry of Business, Innovation and Employment (2025b), p. 59.</ref>

== Pressurisation and smoke control ==
Smoke control combines barriers such as walls, floors and doors with airflows and pressure differences generated by fans.<ref name="Klote-278" /> The high-pressure side of a barrier can be a refuge area or an escape route. Air flowing through the cracks around a closed door then prevents smoke from entering.<ref name="Klote-278" /> When the door is opened, smoke can flow against a slow airflow into the escape route. A high enough velocity through the doorway prevents this backflow.<ref name="Klote-278" />

[[File:fse-protected-escape-route-v2.svg|thumb|upright=1.5|alt=Door details between a smoky floor and a stair: air leaks out through a closed door; through an open door, slow air lets smoke in, fast air keeps it out.|Pressurisation of an escape route: air leaking through a closed door, and smoke backflow through an open door at low and at high air velocity. Based on Klote (2002), p. 4-278.]]

Pressurised stairs are designed to provide a smoke-free escape route and, as a secondary aim, a smoke-free staging area for firefighters.<ref name="Klote-286">Klote (2002), p. 4-286.</ref> Klote estimates that stopping smoke backflow through an open door needs about 4 m/s for an intense fire of 2.4 MW, and about 1.5 m/s for a fire of 125 kW.<ref name="Klote-286" /> In a sprinklered building, where the smoke is cooled, 0.25 to 1.25 m/s may be used.<ref name="Klote-286" /> Australian deemed-to-satisfy provisions refer to AS 1668.1. It requires at least 1 m/s through each open doorway to the fire-affected compartment when those doors and the main discharge doors of the stairs are open.<ref name="ABCB-C4-127">Australian Building Codes Board (2020a), Data Sheet C4, p. 127.</ref>

Pressure creates its own problem: occupants must still be able to open the doors. For a door of 2.13 m by 0.91 m, a pressure difference of 75 Pa and a closer force of 53 N give a total opening force of 133 N.<ref name="Klote-282">Klote (2002), p. 4-282.</ref> Pressurisation systems also need relief openings, so that excessive pressures do not make doors hard to open or stop them from closing automatically.<ref name="ABCB-C4-127" /> The Australian data sheets give a typical reliability of only 50 percent for pressurisation of fire-isolated exits, with a range of 30 to 60 percent.<ref name="ABCB-C4-134">Australian Building Codes Board (2020a), Data Sheet C4, p. 134.</ref>

== Layout, distance and alternative routes ==
The travel to safety is often considered in two stages: the travel to a protected exit, and the travel through it.<ref name="ABCB-D1-246" /> The length of the first stage, on the open path, is the part most exposed to the fire. In an example in the Australian handbook for the Fire Safety Verification Method, the deemed-to-satisfy travel distance from an apartment door to a fire-isolated stair is 6 m. A proposed design extends it to 12 m.<ref name="ABCB-46">Australian Building Codes Board (2020b), p. 46.</ref>

Where two escape routes are required, they must not both be blocked by one fire. C/VM2 (second edition, 2025) requires alternative escape routes to remain separated until they reach a final exit.<ref name="MBIE25-37" /> They must diverge at an angle of at least 90° until they are 8 m apart, or 20 m apart when more than 250 people use only two routes. Smoke separations and smoke control doors are an alternative to these distances.<ref name="MBIE25-37" /> A route with a single direction of travel may be at most 50 m long for occupants familiar with the building and 40 m for others.<ref name="MBIE25-38">Ministry of Business, Innovation and Employment (2025b), p. 38.</ref>

[[File:fse-protected-escape-route-v3.svg|thumb|upright=1.5|alt=Plan of a room with two exits: routes leave a common point at an angle of at least 90 degrees and are 8 m or 20 m apart at the exits.|Alternative escape routes in C/VM2: they diverge at 90° or more until 8 m apart, or 20 m apart when more than 250 people use only two routes. Based on Ministry of Business, Innovation and Employment (2025b), p. 37.]]

The World Trade Center showed the importance of separation between protected stairs. On floors 83 and higher, two stairs were only about 21 m apart, while on floors 77 to 82 they were 54 to 63 m apart.<ref name="Averill-44">Averill et al. (2005), p. 44.</ref> In the first tower, where the stairs in the impact zone were clustered in the core, all were destroyed. In the second tower, where they were furthest apart, one stair remained marginally passable.<ref name="NIST-xl" /> NIST therefore recommended that egress systems maximise the remoteness of stairs and keep their functional integrity in foreseeable emergencies.<ref name="NIST-216">National Institute of Standards and Technology (2005), pp. 216–217.</ref>

== Use, maintenance and limits ==
A protected route only helps if people use it. The ''Introduction to Fire Safety Engineering'' notes that emergency doors and routes reserved for incidents are hardly used in practice. A practical approach is to integrate escape routes with the routes used every day.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> The International Fire Engineering Guidelines place the duty to keep egress paths clear, doors working and signs in good condition with building management.<ref name="IFEG-1.8-14">National Research Council of Canada et al. (2005), p. 1.8-14.</ref>

Protected routes can also become the place of a fire. The Australian handbook notes that fire-isolated stairs are sometimes used to store rubbish or furniture, and that combustibles may be brought in and ignited deliberately. It treats a fire at the base of a fire-isolated stair as a reference scenario for this reason.<ref name="ABCB-67">Australian Building Codes Board (2020b), p. 67.</ref> For sprinklered sleeping occupancies with a single vertical escape route, C/VM2 (second edition, 2025) requires visibility in that route of at least 5.0 m for the whole required safe egress time.<ref name="MBIE25-59" />

== See also ==
* [[Evacuation strategy]]
* [[Phased evacuation]]
* [[Exit capacity]]
* [[Emergency lighting]]
* [[Escape signage]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Fire water supply

2026-11-02T17:14:38Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''fire water supply''' is the source, storage and distribution of the water that sprinkler systems, hydrants, risers and hoses need to fight a fire. It can come from a public water main, from tanks on site, or from pumps that draw on either of them.<ref name="HO-21">Hopkins et al. (2005), p. 21.</ref><ref name="FL-83">Fleming (2002), p. 4-83.</ref> Its strength is described by how much water it can deliver, at what pressure, and for how long. Fleming calls the strength of the water supply the key to whether it will adequately serve a [[sprinkler system]].<ref name="FL-81">Fleming (2002), p. 4-81.</ref>

The supply serves two groups of users. Automatic systems such as sprinklers need water at once and without human action.<ref name="HO-21" /> The fire service needs water at the building for its hoses, from a [[fire hydrant]], from a riser inside the building or from its own pumping appliances.<ref name="EV-18">Evans et al. (2005), p. 18.</ref><ref name="MBIE-54">Ministry of Business, Innovation and Employment (2025b), p. 54.</ref>

In New Zealand, Building Code clause C5.5, as reproduced in C/VM2 (second edition, 2025), requires buildings to be provided with the means to deliver water for firefighting to all parts of the building.<ref name="MBIE-53">Ministry of Business, Innovation and Employment (2025b), p. 53.</ref> A water supply can also be a single point of failure. A closed valve, an empty tank or a pump that does not start can disable every system that depends on it.<ref name="IS-10-2">Iqbal and Salley (2004), p. 10-2.</ref> Redundant supply paths and independent sources are used to reduce this dependence.<ref name="EV-32">Evans et al. (2005), p. 32.</ref><ref name="MBIE-44">Ministry of Business, Innovation and Employment (2025b), p. 44.</ref>

== Sources of water ==
Water can come from several kinds of source. In the United States, a wet standpipe system needs a direct connection to an automatic water supply. Examples are a city water distribution system, a tank and pump, an elevated tank or a pressure tank.<ref name="HO-21" /> Sprinkler systems draw on the same kinds of source. NFPA 13, the US sprinkler standard of the National Fire Protection Association (NFPA), refers to separate standards for the supply components. These are NFPA 14 for standpipes, NFPA 20 for fire pumps, NFPA 22 for water tanks and NFPA 24 for private fire service mains.<ref name="FL-73">Fleming (2002), p. 4-73.</ref>

[[File:fse-fire-water-supply-v1.svg|thumb|upright=1.5|alt=Schematic of a building fed by a street main and a roof tank, with fire pump, riser, sprinklers, hose reel, street hydrant and fire service connection.|Typical parts of a fire water supply for a multistorey building. Based on Fleming (2002), pp. 4-82 to 4-83, Hopkins et al. (2005), p. 21 and Evans et al. (2005), pp. 14 and 18.]]

If not enough water is available at a suitable pressure, a stored water supply becomes necessary.<ref name="FL-82" /> The basic choice is between an elevated (gravity) tank, a pressure tank, or a suction tank combined with a pump.<ref name="FL-83" /> Without a pump, the energy for the system must come from the height of a gravity tank or from the air pressure in a pressure tank.<ref name="FL-83" />

Elevation adds or removes pressure at 0.433 psi per foot, or 0.099 bar per metre.<ref name="FL-80">Fleming (2002), p. 4-80.</ref> The pressure of a gravity tank must therefore be calculated from the lowest expected water level, which is normally the level at which the tank counts as empty.<ref name="FL-83" /> A pressure tank must contain enough air to push out the last water at an adequate pressure. A common rule of thumb has been one-third air at a minimum pressure of 75 psi (5.17 bar), but this rule does not hold for systems with high pressures.<ref name="FL-83" />

Fire pumps make up for a lack of pressure where enough water is available at a suitable suction pressure.<ref name="FL-82">Fleming (2002), p. 4-82.</ref> Plumbing codes sometimes set a minimum suction pressure of 10 to 20 psi (0.69 to 1.38 bar).<ref name="FL-82" /> Listed fire pumps have diesel or electric drivers and capacities from 25 to 5,000 gpm (95 to 18,927 L/min). The most common sizes lie between 250 and 2,500 gpm (946 to 9,463 L/min).<ref name="FL-82" /> NFPA 20 requires each horizontal fire pump to deliver at least 100 % of its rated pressure at its rated flow and at least 65 % of that pressure at 150 % of the flow. With no flow, the pressure must not exceed 140 % of the rated pressure.<ref name="FL-82" />

The World Trade Center complex in New York shows how these elements are combined in a large building. Its primary supply was a 12 in. fire main that looped the complex and was fed from the municipal supply by two redundant 16 in. connections.<ref name="EV-14">Evans et al. (2005), p. 14.</ref> Each tower had reserve tanks of 5,000 gal (about 19 m³) on several mechanical floors, and the tanks on the 41st floor had automatic fill lines from the domestic water system. Twelve fire pumps moved the water up the building.<ref name="EV-18" /> Fire department engines could also pump water from street hydrants into the buildings through 32 fire department connections at 14 locations.<ref name="EV-18" />

== Flow, pressure and duration ==
A water supply is assessed by testing it. Each test gives at least two values: the static pressure without flow and the residual pressure while a known flow is taken.<ref name="FL-81" /> The static pressure is read from a gauge on a hydrant. The residual pressure is read on the same gauge while water flows from another hydrant, preferably downstream.<ref name="FL-81" />

[[File:fse-fire-water-supply-v2.svg|thumb|upright=1.5|alt=Two hydrants on a water main: the first with a pressure gauge, the second, downstream, discharging water measured with a pitot tube.|Water supply test: the gauge on one hydrant reads the static pressure with no flow and the residual pressure while a downstream hydrant flows. Based on Fleming (2002), p. 4-81.]]

The static pressure reflects elevated tanks and running pumps, while the drop to the residual pressure reflects friction losses in the pipes.<ref name="FL-81" /> Because friction loss grows with the flow to the power 1.85, the two points can be drawn as a straight line on special graph paper. This line, the water supply curve, gives the available pressure at any flow.<ref name="FL-81" />

The required flow and pressure come from a [[hydraulic calculation]] of the system. For an adequate design, the system demand point, including the allowance for fire service hoses, must lie below the water supply curve.<ref name="FL-81" /> Outside the design area of a sprinkler system, no flow is added other than the hose stream allowances.<ref name="FL-75">Fleming (2002), p. 4-75.</ref> For standpipes, US practice based on NFPA 14 (2000) requires 500 gpm (1,893 L/min) at 100 psi (6.9 bar) at the hydraulically most remote standpipe. Each additional standpipe adds 250 gpm (946 L/min), up to a maximum of 1,250 gpm (4,732 L/min).<ref name="HO-20">Hopkins et al. (2005), p. 20.</ref>

Duration is the third requirement. Tanks must be sized to provide the minimum durations set by NFPA 13 or another applicable standard, and the required pressure must still be available when the tank is nearly empty.<ref name="FL-83" /> The stored volume follows from the flow and the duration:

<math display="block">V = Q \times t</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>V</math> || usable stored water volume || m³ (or L)
|-
| <math>Q</math> || water demand (flow) of the systems supplied || m³/min (or L/min)
|-
| <math>t</math> || required duration of supply || min
|}

The relation assumes a constant demand and no refill during the fire. Any automatic refill, for example from the public main, lengthens the actual duration. The volume is the usable volume above the level at which the tank counts as empty.<ref name="FL-83" /><ref name="EV-18" />

=== Worked example ===
Two 5,000 gal tanks and a 5,000 gal standpipe reserve tank on the 110th floor of each World Trade Center tower supplied the booster pump for the highest sprinkler zone. Their combined capacity was 15,000 gal (56.8 m³).<ref name="EV-18" /><ref name="EV-19">Evans et al. (2005), p. 19.</ref> The fire pump was rated at 500 gpm (1,893 L/min), and the maximum permissible flow was 150 % of that, 750 gpm (2,839 L/min).<ref name="EV-19" /> Rearranging the relation gives the duration of supply:

[[File:fse-fire-water-supply-v3.svg|thumb|upright=1.5|alt=Falling curve of supply duration against flow for a 56.8 m3 tank, with marked points 30 min at 1,893 L/min and 20 min at 2,839 L/min.|Duration of a 56.8 m³ (15,000 gal) stored supply as a function of the demand, t = V/Q. Based on Evans et al. (2005), pp. 18 to 19.]]

<math display="block">t = \frac{V}{Q} = \frac{56.8\ \mathrm{m^3}}{1.893\ \mathrm{m^3/min}} = 30\ \mathrm{min}</math>

At the maximum flow of 2.839 m³/min, the same tanks last 20 minutes. The National Institute of Standards and Technology (NIST) gives the same values: a minimum of 30 minutes at 500 gpm, without any supplemental water from the domestic system, and 20 minutes at 750 gpm.<ref name="EV-19" /> Fleming's example for a pressure tank shows the same logic. A 30-minute supply for a demand of 140 gpm (530 L/min) requires at least 4,200 gal (15.9 m³) of water, plus the air needed to keep the pressure up.<ref name="FL-83" />

Hydraulic calculations by NIST for the light hazard sprinkler systems of the towers gave delivered densities of 0.10 to 0.27 gpm/ft² with up to 16 open sprinklers. The primary water supply could maintain these flows for about 10 to 89 minutes, depending on the location and number of open sprinklers.<ref name="EV-32" /> NIST concluded that the storage tanks gave enough duration for normally expected fires to allow the fire department to respond and supplement the supply.<ref name="EV-40">Evans et al. (2005), p. 40.</ref>

== Water for the fire service ==
The fire service draws its water from hydrants, from risers and from its own vehicles. In a fire engineering analysis, the time to set up a water supply is one of the steps in the fire brigade intervention model of the International Fire Engineering Guidelines.<ref name="IFEG-2.9-5">National Research Council of Canada et al. (2005), p. 2.9-5.</ref> The amount of firefighting water available at the scene is one of the resources to be assessed.<ref name="IFEG-1.9-5">National Research Council of Canada et al. (2005), p. 1.9-5.</ref> If the water supply or the number of firefighters is insufficient for interior firefighting, the strategy may change from offensive, fighting the fire, to defensive, stopping its spread to neighbouring buildings.<ref name="IFEG-2.9-4">National Research Council of Canada et al. (2005), p. 2.9-4.</ref>

National rules set the facilities. In Australia, Table 2.8 of the Fire Safety Verification Method Standard 2022 requires external fire hydrants for all buildings in the fire brigade intervention scenario. Internal hydrants are required in sprinklered buildings if any point is more than 100 m away or the building has more than three levels, and in unsprinklered buildings from 70 m.<ref name="ABCB-16">Australian Building Codes Board (2022), p. 16.</ref> The scenario applies only to buildings within 50 km road travel of a fire station.<ref name="ABCB-15">Australian Building Codes Board (2022), p. 15.</ref> In New Zealand, C/VM2 (second edition, 2025) accepts water from a pumping appliance parked so that any point in the building lies within 75 m, about three hose lengths. Otherwise an internal hydrant to NZS 4510 is needed.<ref name="MBIE-54" />

Inside tall buildings, risers bring the water to the floors. Standpipe systems are classed as wet, damp or dry by their water supply. Wet systems are filled with water at all times, while manual [[Dry and wet risers|dry standpipes]] have only fire department connections and no automatic supply.<ref name="HO-21" /> Removing the need for long hose lays from the fire engine to an upper floor improves the initial attack.<ref name="HO-21" /> For Dutch indoor car parks, NIPV notes that a dry riser can shorten the attack time, because less hose has to be rolled out.<ref name="NIPV-66">Van der Graaf et al. (2021/2023), p. 66.</ref>

Occupants can use the same water. The International Fire Engineering Guidelines mention suppression by building occupants with extinguishers and a [[hose reel]].<ref name="IFEG-1.7-1">National Research Council of Canada et al. (2005), p. 1.7-1.</ref> The New York City building code had no requirement for fire extinguishers, because it required occupant hose reels.<ref name="LE-151">Lew et al. (2005), p. 151.</ref> In World Trade Center Building 7 (WTC 7), hose reels were located in the stair enclosures, with hose cabinets throughout the floors.<ref name="HO-128">Hopkins et al. (2005), p. 128.</ref>

== Reliability and redundancy ==
Several common causes of sprinkler failure lie in the water supply. The US Nuclear Regulatory Commission (NRC) lists a closed water supply valve, a broken supply header, an empty water tank and a fire pump that fails to start among the causes.<ref name="IS-10-2" /> An electric pump can fail through a power failure, and a diesel pump through poor maintenance, dead batteries or lack of fuel.<ref name="IS-10-2" /> The Australian data sheets summarise US statistics by Ahrens. A system that was shut off accounted for 59 % of the cases in which a sprinkler system failed to operate. When systems did operate but were ineffective, 30 % of the cases involved not enough water being discharged.<ref name="AH-79">Ahrens (2017), as cited in Australian Building Codes Board (2020a), Data Sheet C2, p. 79.</ref>

[[File:fse-fire-water-supply-v4.svg|thumb|upright=1.5|alt=Bar chart of why sprinklers failed to operate: shut off 59 %, manual intervention 17 %, maintenance 10 %, damage 7 %, wrong system 7 %.|Causes of sprinkler systems failing to operate in US fires, as summarised in the Australian data sheets. Based on Ahrens (2017), as cited in Australian Building Codes Board (2020a), Data Sheet C2, p. 79.]]

The [[reliability of fire protection systems]] therefore depends strongly on valves and their management. An Australian fault tree study of sprinkler systems in high-rise offices found reliabilities between 86.6 % and 97.9 %. Shutting off sprinkler zones during tenancy changes was one of the main factors in failure.<ref name="MT-81">Moinuddin and Thomas, as cited in Australian Building Codes Board (2020a), Data Sheet C2, p. 81.</ref> According to the fault tree analysis, a zone isolation valve for each floor could reduce the failure probability by 10 % to 63 % of its original value.<ref name="MT-81" />

Redundancy reduces the dependence on a single source. In New Zealand, C/VM2 (second edition, 2025) states that a sprinkler system with two independent supplies, of which only one may depend on a town's main, has a very high degree of reliability. Such a system, with a Class A or Class B2 water supply to NZS 4541, can alone satisfy clause C3.6, which limits the radiation received at the relevant boundary.<ref name="MBIE-43">Ministry of Business, Innovation and Employment (2025b), p. 43.</ref><ref name="MBIE-44" /> The World Trade Center towers were supplied from loops at two remote locations, with isolation valves that allowed either main to operate independently.<ref name="EV-30">Evans et al. (2005), p. 30.</ref> WTC 7, by contrast, was fed by two mains within 12 ft (3.7 m) of each other from the same city main.<ref name="EV-30" />

Redundant sources can still share a weak point, which then becomes a [[common cause failure]]. NIST found that the risers of WTC 1, 2 and 7 provided redundant supply. Each floor's sprinkler system, however, had a single connection to the riser, a single point of failure for the sprinklers on that floor.<ref name="EV-41">Evans et al. (2005), p. 41.</ref> The emergency fire pumps lacked remote redundancy in their power lines, which could have affected the sprinklers and standpipes once power was lost.<ref name="EV-42">Evans et al. (2005), p. 42.</ref>

At one chemical plant, the fire pump house flooded three times in five years, and the pump controllers were shut off each time. The pumps thus formed a common cause failure for numerous deluge systems. Barry gives target availabilities for such systems of 0.98 to 0.999.<ref name="BA-203">Barry (2002), p. 5-203.</ref>

== Limits and practical issues ==
A design water supply covers the expected fire, not every fire. In the World Trade Center, the secondary water supply for the high and mid-level zones depended on manually started fire pumps. NIST notes that an automatic supplemental supply is required by NFPA 14 and represents current best practice.<ref name="EV-41" /> The lower floors of WTC 7 were supplied directly from the city system, so there was no secondary source if that system failed.<ref name="EV-40" /><ref name="EV-41" /> A long run of 4 in. pipe from the 110th floor tanks restricted the flow to several floors, although it was sufficient for the design.<ref name="EV-41" />

New hazards can change the demand. NIPV reports that fires in electric vehicles in car parks need large amounts of water for a long time to cool effectively. The extinguishing water supply and water drainage should be designed for this, and contaminated cooling water needs attention.<ref name="NIPV-6">Van der Graaf et al. (2021/2023), p. 6.</ref> The operating time of a sprinkler system, typically 30 to 120 minutes depending on the standard, limits how long it can support the fire service.<ref name="NIPV-47">Van der Graaf et al. (2021/2023), p. 47.</ref> Firefighting water can also harm the environment. The International Fire Engineering Guidelines name run-off of firefighting water from a chemical warehouse as an example.<ref name="IFEG-2.9-3">National Research Council of Canada et al. (2005), p. 2.9-3.</ref>

Statistics on sprinkler performance can hide supply problems. Australian and New Zealand data in the International Fire Engineering Guidelines show that 65 % of fires were controlled by one sprinkler head. These data exclude fires in which sprinklers failed to operate or water supplies were not available, so they may not reflect the overall reliability of sprinkler systems.<ref name="IFEG-3.7-3">National Research Council of Canada et al. (2005), p. 3.7-3.</ref>

== See also ==
* [[Wet pipe sprinkler system]]
* [[Dry pipe sprinkler system]]
* [[Deluge system]]
* [[Sprinkler design density]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)

[[Category:Fire Suppression]]

Fire risk assessment

2026-11-02T17:14:35Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''fire risk assessment''' is a structured process for identifying fire hazards, estimating the likelihood and consequences of the resulting fires, and judging whether the risk is acceptable.<ref name="WH-3">Watts and Hall (2002), p. 5-3.</ref><ref name="WH-4">Watts and Hall (2002), p. 5-4.</ref> It can cover a building, a plant or an activity. Watts and Hall summarise the underlying questions as: what could happen, how bad would it be, and how likely is it to happen.<ref name="WH-1">Watts and Hall (2002), p. 5-1.</ref> A fire risk assessment builds on a fire hazard assessment, which identifies what could go wrong, and then weights each consequence by the likelihood of the scenario that causes it.<ref name="BU-231">Bukowski (2002), p. 5-231.</ref>

Methods range from simple checklists and narratives through risk indexes and risk matrices to fully probabilistic analyses with event trees.<ref name="WH-4" /><ref name="WH-5">Watts and Hall (2002), p. 5-5.</ref> The results are used to choose between designs, to show that a performance-based design is at least as safe as a compliant reference building, and to set priorities for protection in existing buildings and plants.<ref name="WH-6">Watts and Hall (2002), p. 5-6.</ref><ref name="HB-124">Australian Building Codes Board (2020b), p. 124.</ref> Every assessment ends in a value judgement, because zero risk is not achievable and no universally acceptable level of fire risk exists.<ref name="WH-4" />

== Definitions ==
The terminology of risk is not consistent. A committee of the Society for Risk Analysis identified 17 different definitions of risk, and the words analysis and assessment are often used interchangeably.<ref name="WH-3" /> In the SFPE Handbook, Watts and Hall define risk as the potential for unwanted, adverse consequences to life, health, property or the environment.<ref name="WH-3" /> When they speak of [[fire risk]], they mean the uncertainty of loss from fire.<ref name="WH-2">Watts and Hall (2002), p. 5-2.</ref> The risk of an event is usually estimated as the expected value of its probability multiplied by its consequence.<ref name="WH-3" /> The International Fire Engineering Guidelines (IFEG, 2005) use the same idea and define risk as the product of the probability and the consequence of an event.<ref name="IFEG-054">National Research Council of Canada et al. (2005), p. 0.5-4 (Part 0, Australia).</ref>

Watts and Hall distinguish several related activities. Risk assessment is the process of establishing information on acceptable levels of a risk, on actual levels of risk, or on both.<ref name="WH-3" /> Risk evaluation is the part of the assessment in which the significance and acceptability of a risk are judged. Risk analysis is the wider examination that also includes options for managing the risk.<ref name="WH-3" /> A fire hazard analysis, by contrast, looks at what could happen and how bad it would be, without analysing likelihood.<ref name="WH-2" />

Meacham describes risk as the possibility of an unwanted outcome in an uncertain situation. It depends on three factors: the loss of or harm to something that is valued, the hazard that may cause it, and a judgement about how likely the loss is.<ref name="ME-153">Meacham (2002), p. 5-153.</ref> In building fire risk analysis, the things that are valued are life safety, property, business continuity, the environment and heritage.<ref name="ME-154">Meacham (2002), p. 5-154.</ref>

== Steps of an assessment ==
Watts and Hall give a general sequence of five steps.<ref name="WH-4" /> The assessor first identifies the fire hazards, then quantifies the consequence and probability of each. Next, the options for controlling the hazards are identified and their effect on the risk is quantified. Finally, the appropriate protection is selected.<ref name="WH-4" /> The quantification can rest mainly on hard data and established science, or mainly on expert judgement.<ref name="WH-4" />

[[File:fse-fire-risk-assessment-v1.svg|thumb|upright=1.5|alt=Five boxes in sequence: identify hazards, quantify consequence and probability, identify control options, quantify their effect, select protection.|The five generic steps of a fire risk assessment. Based on Watts and Hall (2002), p. 5-4.]]

Hazard identification usually starts with a survey of the building or process. Meacham lists the information gathered: potential ignition sources, potential fuels, the arrangement of fuel packages, the configuration of the building and compartments, and the fire safety features present.<ref name="ME-155">Meacham (2002), p. 5-155.</ref> Historical loss data help to avoid focusing on unusual hazards while overlooking more common ones.<ref name="ME-155" /> The hazards are then expressed as [[fire scenario|fire scenarios]], each a sequence of events from ignition to an outcome.<ref name="WH-2" />

The consequence part asks what harm each scenario would cause. Meacham notes that valuing consequences is often less familiar to engineers than assessing the fire itself. Beyond death and injury, it may include reduced quality of life, smoke and water damage, and loss of market share.<ref name="ME-153" /> The likelihood part asks how often each scenario occurs. Meacham stresses the difference between frequency and probability. Twenty toaster fires per year is a frequency; with 20 million toaster starts per year, the probability of a fire per start is one in a million.<ref name="ME-159">Meacham (2002), p. 5-159.</ref>

== Methods ==
Watts and Hall classify fire risk methods into four groups: checklists, narratives, indexing and probabilistic methods.<ref name="WH-4" /> Checklists and narratives are not quantitative. They identify hazards, control options and protection, but skip the quantification of probability and of the effect of options.<ref name="WH-4" /> Risk indexing assigns values to selected positive and negative fire safety features based on judgement and experience. It combines them arithmetically into a single value that can be compared with a standard.<ref name="WH-5" /> ''Introduction to Fire Safety Engineering'' notes that qualitative assessments often use risk classes such as very low, low, medium, high and very high.<ref name="VR-1.4">Van Rossum (2027), §1.4.</ref>

Risk matrices sit between indexes and fully probabilistic methods.<ref name="WH-5" /> Meacham describes a matrix in which the largest realistic consequence of each type of loss is ranked as negligible, low, moderate or high.<ref name="ME-166">Meacham (2002), p. 5-166.</ref> The frequencies are ranked in bins as well.<ref name="ME-167">Meacham (2002), p. 5-167.</ref>

In Meacham's example, anticipated events occur more often than once per 100 years (more than 10−2 per year). Unlikely events occur between 10−4 and 10−2 per year, and extremely unlikely events between 10−6 and 10−4 per year.<ref name="ME-167" /> The consequence predictions should bound all outcomes at the 95th percentile or better.<ref name="ME-166" /> Frequencies should refer to exceeding a given loss rather than to single scenarios: 30 scenarios of 10−7 per year together give 3 × 10−6 per year.<ref name="ME-166" />

[[File:fse-fire-risk-assessment-v2.svg|thumb|upright=1.5|alt=Risk matrix: four consequence levels against four frequency bins; each cell marked high, moderate, low or negligible risk.|Example risk ranking matrix: each combination of consequence level and frequency bin is assigned a risk level. Based on Meacham (2002), pp. 5-166–5-167.]]

Probabilistic methods are the most informative, because their results can be traced through explicit assumptions, data and relationships.<ref name="WH-5" /> The most common tools are [[event tree analysis]], which follows the possible outcomes after an initiating event in time order, and fault trees, which trace the logical causes of a failure.<ref name="WH-5" /> Meacham explains that an event tree starts with an initiating event and asks, for each protection system or strategy, whether it succeeds or fails.<ref name="ME-156">Meacham (2002), p. 5-156.</ref> A full multi-scenario analysis of this kind is a form of [[quantitative risk assessment]], also called probabilistic risk assessment, which is widely used in the chemical process industry and for nuclear facilities.<ref name="ME-161">Meacham (2002), p. 5-161.</ref>

== Calculating risk from scenarios ==
In a quantitative assessment, the risk is built up scenario by scenario. For each scenario, the analyst estimates its probability and its consequence, multiplies them, and adds the results over all scenarios.<ref name="IFEG-235">National Research Council of Canada et al. (2005), p. 2.3-5.</ref><ref name="ME-161" /> The IFEG express this for life safety as the expected number of deaths, given that a fire has started:<ref name="IFEG-235" /><ref name="IFEG-234">National Research Council of Canada et al. (2005), p. 2.3-4.</ref>

[[File:fse-fire-risk-assessment-v3.svg|thumb|upright=1.5|alt=Two event trees: system A works 0.9 or fails 0.1, risk 0.2 persons; system B works 0.7 or fails 0.3, risk 0.6 persons.|Expected number of people exposed per fire for two systems with 90% and 70% reliability: the risk of system B is three times that of system A. Based on Australian Building Codes Board (2020b), pp. 118–119.]]

<math display="block">\mathit{END} = \sum_j P_j \, N_j</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\mathit{END}</math> || expected number of deaths for a fire, summed over all scenarios || persons
|-
| <math>j</math> || index of the fire scenario || –
|-
| <math>P_j</math> || probability that scenario ''j'' develops after ignition || –
|-
| <math>N_j</math> || number of occupants exposed to untenable conditions in scenario ''j'' || persons
|}

The relation is valid when the scenarios are distinct paths of one event tree, so that each possible outcome is counted once.<ref name="IFEG-232">National Research Council of Canada et al. (2005), p. 2.3-2.</ref> The consequences <math>N_j</math> come from deterministic analysis of each scenario, and the probabilities from reliability data or statistics.<ref name="IFEG-234" /> Meacham gives the same sum with the frequency of each scenario instead of its probability after ignition, which gives a risk per unit of time.<ref name="ME-161" /><ref name="ME-159" /> Dividing the expected number of deaths over the building's life by the number of occupants and the design life gives the expected risk to life (ERL) used in risk-cost models.<ref name="IFEG-235" />

'''Worked example.''' The Australian handbook for the Fire Safety Verification Method (FSVM) compares two fire safety systems.<ref name="HB-118">Australian Building Codes Board (2020b), p. 118.</ref> If either system works, no occupants are exposed to untenable conditions; if it fails, two people are exposed. System A, the reference, has a reliability of 90%, and system B, the proposal, 70%.<ref name="HB-118" /> For system A, the expected number exposed per fire is 0.9 × 0 + 0.1 × 2 = 0.2 persons. For system B it is 0.7 × 0 + 0.3 × 2 = 0.6 persons.<ref name="HB-119">Australian Building Codes Board (2020b), p. 119.</ref>

A purely deterministic analysis, which looks only at the case in which each system works, would find the two systems equivalent. The risk calculation shows that system B exposes occupants to three times the risk of system A.<ref name="HB-119" /> The example shows why system reliability must enter the comparison whenever it differs between the proposed and the reference design.<ref name="HB-118" />

== Use in design and regulation ==
The most common use of fire risk analysis is to support a choice, for example between two designs or two formulations of a code.<ref name="WH-6" /> In Australia, the FSVM handbook states that a risk assessment will generally be needed when the reliability of fire safety systems differs appreciably between a proposed design and the reference building.<ref name="HB-118" /> A detailed quantitative risk assessment is the most complete method, but it often needs substantial resources.<ref name="HB-119" /> For designs with minor differences from the reference building, simple event trees for the probability can be combined with deterministic analysis of the consequences.<ref name="HB-119" /> The proposed design must then show that, for each scenario, the risk to life is no greater than in the reference building.<ref name="HB-124" />

The handbook builds its event tree from the manage-fire branch of the [[fire safety concepts tree]], with added fire brigade intervention.<ref name="HB-120">Australian Building Codes Board (2020b), p. 120.</ref> It points out how the parts of a strategy depend on each other. For unassisted evacuation, detection and alarm must occur, occupants must move, and movement means and a safe destination must be provided. The person must also be able to leave without assistance; if any link fails, escape relies on fire brigade rescue.<ref name="HB-122">Australian Building Codes Board (2020b), p. 122.</ref> A failed detection system may also delay the call to the fire brigade and fail to start smoke control.<ref name="HB-122" />

[[File:fse-fire-risk-assessment-v4.svg|thumb|upright=1.5|alt=Event tree from fire ignition through occupant suppression, sprinklers and fire brigade to outcomes from minimal to catastrophic.|Simplified event tree derived from the manage-fire branch of the fire safety concepts tree: each later system acts only if the earlier ones fail. Based on Australian Building Codes Board (2020b), pp. 120–121.]]

The IFEG describe a probabilistic event tree approach that compares the expected risk to life of a design with an acceptance value. That value may be a fixed number, or the expected risk to life of a design that complies with the prescriptive provisions.<ref name="IFEG-1223">National Research Council of Canada et al. (2005), p. 1.2-23.</ref> They warn that probabilistic methods need much statistical data, can be time-consuming, and may be hard to validate.<ref name="IFEG-1222">National Research Council of Canada et al. (2005), p. 1.2-22.</ref> In the Netherlands, ''Introduction to Fire Safety Engineering'' notes that there are no risk-based limit values.<ref name="VR-1.4" /> Acceptable residual risk must therefore be agreed in advance, for example by comparison with a reference situation that complies with the regulations.<ref name="VR-1.2">Van Rossum (2027), §1.2.</ref>

Watts and Hall note that fire risk analysis is far more common in the chemical process industry and in nuclear power plants than in other fields.<ref name="WH-6" /> For US nuclear power plants, Siu describes how the core damage frequency due to fire is split into three terms. These are the frequency of the fire scenario, the probability of fire damage to critical equipment, and the probability of core damage given that damage.<ref name="SI-216">Siu (2002), p. 5-216.</ref> The three terms correspond to the layers of defence in depth for nuclear plants: fire prevention, fire detection and suppression, and fire mitigation.<ref name="SI-216" /> Qualitative and conservative quantitative screening removes areas that are not expected to contribute to risk, so effort can focus on the scenarios that matter most.<ref name="SI-216" />

== Existing buildings, operations and management ==
Fire risk assessment also applies to buildings and plants in use. Meacham describes safety reviews of existing facilities, carried out as scheduled visual inspections to identify conditions or procedures that could lead to accidents.<ref name="ME-155" /> "What if" analysis and hazard and operability (HAZOP) studies ask systematically what happens when something goes wrong.<ref name="ME-155" /> Checklists help inexperienced staff to focus on hazards, but users may ignore hazards that are not on the list.<ref name="ME-155" />

Operations and their control are central to such assessments. [[Hot work]] means activities that involve heat, sparks or open flame, such as cutting, welding and grinding.<ref name="IS-F37">Iqbal and Salley (2004), p. F-37.</ref> The IFEG list adherence to hot work procedures among the measures that control fire initiation.<ref name="IFEG-1210">National Research Council of Canada et al. (2005), p. 1.2-10.</ref> In US nuclear fire risk assessment, the frequency and nature of maintenance, hot work procedures and housekeeping affect the likelihood of transient fires. The number of work orders in an area is used to rank it.<ref name="NO-621">Nowlen et al. (2005b), p. 6-21.</ref> Where a fire watch is posted during hot work, a fire is likely to be detected quickly.<ref name="NO-1128">Nowlen et al. (2005b), p. 11-28.</ref>

The assessment depends on whether protection works when needed. Meacham distinguishes availability, when a system is in service, from reliability, when an available system actually functions. A detector covered with tape is available but not reliable.<ref name="ME-162">Meacham (2002), p. 5-162.</ref> Watts and Hall list typical questions: is the battery in the smoke alarm working, is the sprinkler valve open, and is the fire door working or blocked open.<ref name="WH-6" /> [[Fire safety management]], including maintenance and inspection, therefore directly changes the numbers in an assessment.

== Acceptability, uncertainty and limits ==
Deciding whether a risk is acceptable is a value judgement. Watts and Hall note that acceptable risk for a practising engineer is likely to be risk acceptable to the client. For an approving authority, it is risk acceptable to society or to a community.<ref name="WH-4" /> Zero risk is not an achievable goal, and reducing fire risk may increase other risks, as with the effect of halon on the ozone layer.<ref name="WH-4" /> Experts, the public, enforcers and managers may value the same measure, such as residential sprinklers, very differently.<ref name="WH-1" /><ref name="WH-2" />

Uncertainty is unavoidable. Meacham groups it into scientific uncertainty, uncertainty about human factors, uncertainty in perceptions and values, and uncertainty in decision-making.<ref name="ME-160">Meacham (2002), p. 5-160.</ref> Fire is a relatively rare event, so data are limited, and changes to buildings and contents make the past an imperfect guide to the future.<ref name="ME-159" /> Even the risk-cost models described by Meacham contain conservative simplifications. They should be used for comparing designs, not for absolute assessments of life risk.<ref name="ME-165">Meacham (2002), p. 5-165.</ref>

The number of scenarios is a practical limit. The IFEG note that an exhaustive measure of risk to life would require every possible fire scenario. Because the computational effort grows with the number of scenarios, the problem must first be simplified.<ref name="IFEG-235" /> A fire risk assessment is therefore a model of the risk, built from chosen scenarios, data and judgements. Its value lies in making those choices explicit, so that they can be checked and discussed.<ref name="WH-5" />

== See also ==
* [[Fire safety engineering]]
* [[Equivalency]]
* [[Reliability of fire protection systems]]
* [[Individual risk]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Bukowski, R. W. (2002). Fire hazard assessment for transportation vehicles. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-227–5-233). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Meacham, B. J. (2002). Building fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-153–5-175). National Fire Protection Association; Society of Fire Protection Engineers.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nowlen, S. P., Najafi, B., Wyant, F. J., Forester, J., Kazarians, M., Kolaczkowski, A., Joglar, F., Funk, D., Hannaman, G., & Anoba, R. (2005b). ''EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Vol. 2. Detailed methodology'' (NUREG/CR-6850; EPRI 1011989). Electric Power Research Institute; U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.
* Siu, N. (2002). Fire risk assessment for nuclear power plants. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-214–5-226). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Risk Analysis]]

Exit capacity

2026-11-02T17:14:33Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Exit capacity''' is the number of people that an exit, stair, door or corridor can pass in a given time, or the number of people for whom it is designed. It is usually expressed per metre of width, for example as persons per metre per minute or persons per second per metre of effective width.<ref name="NM-3-371">Nelson and Mowrer (2002), p. 3-371.</ref><ref name="VR-7.4">Van Rossum (2027), §7.4.</ref> Together with walking speed, exit capacity determines the evacuation time of a building. When many people reach an exit at once, the time they need to pass it, the flow time, usually decides how long the evacuation takes.<ref name="Proulx-3-347">Proulx (2002), p. 3-347.</ref>

Measured flows reach about one person per second through a typical office stair 1,120 mm wide, and up to about 1.3 persons per second per metre of effective width through doors and corridors.<ref name="Proulx-3-346">Proulx (2002), p. 3-346.</ref><ref name="NM-3-371" /> Building regulations translate this into simple rules. Examples are a fixed number of persons per metre of door or stair width in the Netherlands, or units of exit width in the United States.<ref name="VR-7.4" /><ref name="Averill-18">Averill et al. (2005), p. 18.</ref> Performance-based design calculates flow times directly and adds them to the other parts of the required safe egress time.<ref name="MBIE-32">Ministry of Business, Innovation and Employment (2025b), p. 32.</ref>

== Flow, speed and crowd density ==
The movement of a crowd through an exit is described by three quantities: speed, density and flow. Proulx gives the basic relation: the flow equals the speed times the density times the width of the walkway, with the speed measured along the slope for stairs.<ref name="Proulx-3-346" /> Speed depends on density. When there is less than about 0.5 person per m², people walk at about 1.25 m/s, their normal unrestricted speed. At higher densities they slow down, and at 4 or 5 persons per m², comparable to a crowded lift, movement stops.<ref name="Proulx-3-346" />

Flow is therefore small at both low and very high densities and reaches a peak at an intermediate density of around 2 persons per m².<ref name="Proulx-3-346" /> Nelson and Mowrer use a linear relation between speed and density for densities between 0.54 and 3.8 persons per m². Below that range people walk at their own pace, and above it nobody moves until the crowd has thinned out.<ref name="NM-3-370">Nelson and Mowrer (2002), p. 3-370.</ref> In their method the flow per metre of width reaches its maximum at a density of 1.9 persons per m².<ref name="NM-3-371" /> The book ''Introduction to Fire Safety Engineering'' shows the same pattern with data from the London Transport Board of 1958. The capacity of an escape route rises up to a density of about 4.5 persons per m² and then drops sharply as congestion sets in.<ref name="VR-7.4" />

[[File:fse-exit-capacity-v1.svg|thumb|upright=1.5|alt=Line chart: specific flow against crowd density, three arch-shaped curves peaking at 1.9 persons per square metre, highest for corridors and doors.|Specific flow against crowd density for corridors and doorways and for a steep and a shallow stair. Based on Nelson and Mowrer (2002), pp. 3-370 to 3-371.]]

Flow per unit of width is called [[specific flow]]. Nelson and Mowrer give maximum values of 1.3 persons per second per metre of effective width for corridors, aisles, ramps and doorways. For stairs the maximum ranges from 0.94 for steep stairs (190 mm risers, 254 mm treads) to 1.16 for shallow stairs (165 mm risers, 330 mm treads).<ref name="NM-3-371" /> Observations of total evacuations of office buildings gave an optimum on stairs of 1.18 persons per second per metre of effective width, at 2.0 persons per m² and 0.5 m/s along the stair.<ref name="Proulx-3-347" /> On a typical stair 1,120 mm wide this meant one person per second, each person occupying slightly less than two treads and descending one storey every 15 seconds.<ref name="Proulx-3-346" />

For design, Proulx refers to the levels of service of Fruin, which range from A, with little chance of congestion, to F, with much congestion. For emergency movement in limited space, levels C, D and E are recommended. These correspond to flows of about 0.5 to 1.4 persons per second per metre of effective width on level walkways and 0.4 to 0.9 on stairs.<ref name="Proulx-3-347" /> Ramps with slopes up to about 5 % do not reduce walking speed. An upward slope of 10 % reduces it by about 10 %, and an unusually steep upward slope of 20 % by about 25 %.<ref name="Proulx-3-347" />

== Effective width ==
People walking through an exit route keep a small clearance from walls, handrails and other obstacles, to allow for body sway and to keep their balance. The useful or effective width is therefore the clear width minus a boundary layer on each side.<ref name="NM-3-368">Nelson and Mowrer (2002), p. 3-368.</ref> The boundary layers are 15 cm at stair walls and doors, 20 cm at corridor walls, 9 cm at handrails, 10 cm at obstacles and 46 cm in wide concourses. Next to theatre chairs and stadium benches the boundary layer is zero.<ref name="NM-3-369">Nelson and Mowrer (2002), p. 3-369.</ref>

[[File:fse-exit-capacity-v2.svg|thumb|upright=1.5|alt=Three plan views of a door, a corridor and a stair, each with shaded strips along the edges and a narrower effective width in the middle.|Effective width of a door, a corridor and a stair: the clear width minus a boundary layer on each side. Based on Nelson and Mowrer (2002), pp. 3-368 to 3-369.]]

For stairs this means that the effective width is 300 mm less than the nominal width credited in building codes. A stair with a nominal width of 1,120 mm has an effective width of about 812 mm.<ref name="Proulx-3-347" /> The boundary layer concept explains why capacity is not simply proportional to the measured width. In New Zealand, C/VM2 (second edition, 2025) uses the same boundary layer values to calculate the effective width of doors, stairs and corridors.<ref name="MBIE-33">Ministry of Business, Innovation and Employment (2025b), p. 33.</ref>

== Calculating flow time ==
The time needed for a group of people to pass a point in an exit route follows directly from the flow rate. Nelson and Mowrer write it as the number of people divided by the calculated flow, which is the specific flow times the effective width.<ref name="NM-3-371" /> The book uses the same relation in step 6 of its hand calculation, with the flow capacity per metre of width, the width and the number of people.<ref name="VR-7.5">Van Rossum (2027), §7.5.</ref>

<math display="block">t_\text{flow} = \frac{P}{F_s \, W_e}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>t_\text{flow}</math> || flow time: time for the whole group to pass the exit || s
|-
| <math>P</math> || number of people assigned to the exit || persons
|-
| <math>F_s</math> || specific flow, the flow per metre of effective width || persons/(s·m)
|-
| <math>W_e</math> || effective width of the exit (clear width minus boundary layers) || m
|}

The relation is the core of the [[hydraulic evacuation model]]. It assumes that all people start at the same moment, that the flow is not interrupted by individual decisions, and that nearly everyone can keep up with the group.<ref name="NM-3-367">Nelson and Mowrer (2002), p. 3-367.</ref> Where a route narrows or two flows merge, the outgoing flow cannot exceed the maximum specific flow times the effective width of the next element. If more people arrive, a queue forms.<ref name="NM-3-372">Nelson and Mowrer (2002), p. 3-372.</ref> In a time-based analysis, the evacuation time of a space is the larger of the walking time to the exit and the flow time through it.<ref name="MBIE-32" /> The flow time then becomes part of the [[required safe egress time]].<ref name="MBIE-32" />

The book gives a worked example for 300 people leaving a space by four stairs: two of 2.4 m and two of 1.6 m wide. It uses the Dutch flow capacity of 45 persons per metre per minute for stairs, a walking speed of 1 m/s and an equal split of 75 people per stair.<ref name="VR-7.5" /> A stair 1.6 m wide then passes 1.6 × 45 = 72 persons per minute, or 1.2 persons per second, so the narrow stairs govern. The flow time is 75 / 1.2 = 63 s, rounded up. The first person needs 5 s to reach the stair and the last person 10 s from the stair to the exit, so the evacuation time is 5 + 63 + 10 = 78 s.<ref name="VR-7.5" />

[[File:fse-exit-capacity-v3.svg|thumb|upright=1.5|alt=Plan of a hall with 300 people and four stairs, 75 people each; a timeline adds 5 s walking, 63 s flow time and 10 s walking to 78 s.|Hand calculation of evacuation time for 300 people and four stairs: the 1.6 m stairs govern, giving 5 + 63 + 10 = 78 s. Based on Van Rossum (2027), §7.5.]]

The same stair can be checked with the values of Nelson and Mowrer. With a boundary layer of 0.15 m on each side, the effective width is 1.3 m. With maximum specific flows of 0.94 to 1.16 persons per second per metre for stairs, the flow is 1.2 to 1.5 persons per second.<ref name="NM-3-369" /><ref name="NM-3-371" /> The flow time for 75 people then lies between 50 and 62 s, close to the book's result. Both methods assume ideal conditions. Nelson and Mowrer stress that real evacuations are usually slower than the modelled time, for example because some exits are overloaded while others are hardly used.<ref name="NM-3-368" />

== Exit capacity in regulations ==
In the Netherlands, building regulations have set requirements for the width of escape doors since 1992. After the limit for the flow capacity of escape doors was questioned, TU Delft carried out laboratory experiments in 2009. In 13 of the 16 experiments the average measured capacity of a doorway was higher than 135 persons per minute.<ref name="VR-7.4" /> Lower values were found with an open door, with a wide opening of 275 cm, and with a group that included people who could not move unaided. With 5 % of such people, the capacity was 10 % below 135 persons per minute, the lowest result.<ref name="VR-7.4" />

The results were included in the Dutch building decree (Bbl), which expresses flow capacity in persons per metre of clear width. The most common value is 90 persons per metre for a doorway. A stair has 45 persons per metre when it bridges a height of more than 1 m and 90 persons per metre for 1 m or less. Other passages, including sliding doors, have 135 persons per metre.<ref name="VR-7.4" /> The book lists these rules as the simplest way to determine evacuation time. More detailed options are a calculation of holding and flow capacity, a hand calculation and evacuation simulation.<ref name="VR-7.5" />

In New Zealand, C/VM2 calculates the flow through a door with the same equation as Nelson and Mowrer, with a density of 1.9 persons per m². For a standard door leaf with a self-closer, the flow is capped at 50 persons per minute. This corresponds to a door 0.95 m wide with an effective width of 0.65 m.<ref name="MBIE-33" /><ref name="MBIE-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref> The cap reflects the way a self-closing door leaf gets in the way of the people passing through it. If the door is held open or has no self-closer, the full flow from the equation applies, and an automatic sliding door counts with its full effective width once it is open.<ref name="MBIE-34" /> A double door with self-closers on both leaves can pass at most 100 persons per minute.<ref name="MBIE-34" />

[[File:fse-exit-capacity-v4.svg|thumb|upright=1.5|alt=Line chart: door flow rising linearly with effective width, about 79 persons per minute per metre, with a horizontal cap at 50 persons per minute.|Flow through a doorway against effective width in C/VM2, with the cap of 50 persons per minute for a self-closing door leaf. Based on Ministry of Business, Innovation and Employment (2025b), pp. 33–34.]]

C/VM2 also limits how people may be assumed to spread over the exits. A main entrance must be designed for half of the occupants, and without a main entrance no exit may be given more than half of the occupants.<ref name="MBIE-32" /> Doors on escape routes used by more than 50 occupants must open in the direction of escape.<ref name="MBIE-32" />

In the United States, codes have long used units of exit width. The New York City code of 1938 allowed 30 persons per unit of exit width, and the code of 1968 allowed 60. That change effectively halved the required capacity of new buildings.<ref name="Averill-18" /> The International Building Code of 2003, as compared in the World Trade Center investigation, allows 0.005 m of exit width per person in fully sprinklered buildings. The New York code, the International Building Code and the Life Safety Code all allow the capacity of each stair to be doubled when two stairs are located in separated areas of refuge.<ref name="Averill-47">Averill et al. (2005), p. 47.</ref><ref name="Averill-48">Averill et al. (2005), p. 48.</ref> The required capacity is based on the [[occupant load]] of the area served, so capacity rules are always linked to occupant density assumptions.<ref name="Averill-47" />

Those assumptions can differ widely from reality. Studies in Canadian office buildings found that each occupant had on average about 25 m² of gross rentable area, whereas codes traditionally assume 9.3 m² per person. Code estimates of the number of occupants may therefore be wrong by a factor of more than two.<ref name="Proulx-3-355">Proulx (2002), p. 3-355.</ref>

== Experience from tall buildings ==
In tall buildings, exit capacity is a matter of time rather than of a single floor. NIST noted that the egress capacity required by building codes follows from calculations for a single floor, independent of the height of the building. It does not consider the time needed for a full building evacuation.<ref name="NIST-xxxix">National Institute of Standards and Technology (2005), p. xxxix.</ref> In the World Trade Center towers on 11 September 2001, the buildings were only a third to half occupied. The flow of people had slowed considerably some 20 minutes before each tower collapsed, which showed that the stair capacity was adequate that morning.<ref name="NIST-xxxviii">National Institute of Standards and Technology (2005), p. xxxviii.</ref><ref name="NIST-xxxix" />

NIST estimated that with 20,000 occupants per tower, evacuation by the stairs would have taken just over three hours. About 14,000 people might then have died, because the stairs could not have carried that many people in the available time.<ref name="NIST-xxxix" /> A fourth stair, which the New York code required because of the assembly spaces at the top of each tower, would probably have reduced this shortfall.<ref name="NIST-xxxix" /> During the bombing of 1993 the stairs met the code requirements for exit capacity, but they did not allow all occupants to leave quickly. Stair doors were held open by the crowd, which increased the airflow up the stairs caused by the stack effect.<ref name="FS-59">Ferreira and Strege (2005), p. 59.</ref>

== Limits of capacity calculations ==
Capacity values describe ideal flows. Nelson and Mowrer distinguish the modelled evacuation time from the actual time and introduce an apparent evacuation efficiency. It accounts for delays caused by wardens directing people, stopping and restarting at merging points, people delaying their own start, and unbalanced use of the exits.<ref name="NM-3-368" /> Drills confirm the last point. In 1972, a total evacuation of a 15-storey office building by four stairs took between 6.6 and 9.3 minutes per stair, mainly because the demand on the stairs was unequal.<ref name="Proulx-3-356">Proulx (2002), p. 3-356.</ref>

The way people are divided over the exits is uncertain. The book notes that both an equal split and a split in proportion to capacity are very optimistic assumptions.<ref name="VR-7.5" /> People prefer the route by which they entered the building and routes that look attractive, so some exits may be little used.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> Capacity also falls when the group includes people who cannot walk unaided.<ref name="VR-7.4" /> C/VM2 therefore requires the egress analysis to cover the whole length of the [[means of escape|escape route]], so that the flow is not restricted at some point closer to the final exit.<ref name="MBIE-33" />

== See also ==
* [[Travel distance]]
* [[Walking speed]]
* [[Protected escape route]]
* [[Evacuation modelling]]

== Notes ==
<references />

== References ==
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Ferreira, M. J., & Strege, S. M. (2005). ''Smoke management systems'' (NIST NCSTAR 1-4D). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* Nelson, H. E., & Mowrer, F. W. (2002). Emergency movement. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-367–3-380). National Fire Protection Association; Society of Fire Protection Engineers.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Hand calculations

2026-11-02T17:14:31Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Hand calculations''' in fire safety engineering are simple calculation models in which a few formulas give an estimate of a fire effect or an evacuation time. The work can be done literally by hand, in a spreadsheet or with a modest computer application.<ref name="VR8.2">Van Rossum (2027), §8.2.</ref> Typical examples are the height of a flame, the temperature of a smoke layer, the response time of a detector and the time needed to evacuate a room.<ref name="IS-iii">Iqbal and Salley (2004), p. iii.</ref><ref name="VR7.5">Van Rossum (2027), §7.5.</ref> They form the simplest level of [[fire modelling]], below the [[zone model]] and computational fluid dynamics.<ref name="HB21">Australian Building Codes Board (2020b), p. 21.</ref>

Hand calculations give quick insight and can be used to check a more complex calculation with many variables.<ref name="VR7.5" /> They rest on correlations that were mostly derived from experiments, so each one is valid only within the conditions of those experiments.<ref name="NUREG3-4">Najafi et al. (2007), p. 3-4.</ref> Walton notes that manual methods give good estimates of specific effects, such as the time to flashover. They are, however, poorly suited to analyses in which many processes interact over time.<ref name="Wa189">Walton (2002), p. 3-189.</ref> A well-known example is Heskestad's flame height correlation. It predicts a mean flame height of about 4.2 m for a stack of wood pallets burning at 2.6 MW.<ref name="He2-4">Heskestad (2002), p. 2-4.</ref>

== What hand calculations are ==
''Introduction to Fire Safety Engineering'' distinguishes three forms of model: real tests, simple calculation models and complex computer simulations. The simple models are those in which a hand calculation can be made with a few formulas.<ref name="VR8.2" /> The ''International Fire Engineering Guidelines'' (IFEG, 2005) list "formulas, equations and hand calculations" as the first of several forms of analysis. The others include spreadsheet calculations, statistical studies and experiments with physical scale models.<ref name="IFEG1223">National Research Council of Canada et al. (2005), p. 1.2-23.</ref>

Most hand calculations are closed-form algebraic expressions. They give a result directly from the inputs, without step-by-step numerical solution.<ref name="NUREG2-13">Najafi et al. (2007), p. 2-13.</ref> The US Nuclear Regulatory Commission (NRC) describes its library of such methods as "first-order" methods. These include traditional approaches, correlations, closed-form approximations, exact solutions and hazard models.<ref name="IS-xxxix">Iqbal and Salley (2004), p. xxxix.</ref> Many of the correlations come from the SFPE Handbook of Fire Protection Engineering and the NFPA Fire Protection Handbook.<ref name="IS-iii" />

Spreadsheets make such calculations faster and reduce errors. In the NRC spreadsheets, users enter data in marked input cells, and they can pick a material from a list so that its properties are filled in automatically. This avoids errors in typing property values by hand.<ref name="IS-xxxix" /> Each spreadsheet shows the calculation step by step, so that the user can follow the method.<ref name="IS-xxxix" />

A hand calculation is only as good as its input data. The NRC tools, for example, need thermal properties of walls and ceilings, such as thermal inertia, conductivity, specific heat and density. They also need fuel properties, such as the mass burning rate and the effective heat of combustion, and target properties, such as the ignition temperature and the critical heat flux for ignition.<ref name="DH2-2">Dreisbach and Hill (2007), p. 2-2.</ref> For this reason, each spreadsheet contains a table of properties of materials commonly found in nuclear power plants.<ref name="IS-iii" />

The term covers both fire and people. On the fire side, the IFEG name hand calculations for smoke layer height, smoke production and the response time of heat detectors.<ref name="IFEG254">National Research Council of Canada et al. (2005), p. 2.5-4.</ref><ref name="IFEG272">National Research Council of Canada et al. (2005), p. 2.7-2.</ref> On the people side, ''Introduction to Fire Safety Engineering'' gives an eight-step hand calculation of evacuation time.<ref name="VR7.5" /> The Australian FSVM handbook notes that evacuation models can be either simple hand calculations or more advanced simulation software.<ref name="HB109">Australian Building Codes Board (2020b), p. 109.</ref>

== Common fire calculations ==
The library of Fire Dynamics Tools (FDTs) published by the NRC in 2004 shows the range of fire hand calculations. It was written for inspectors of nuclear power plants, and its chapters cover the main topics.<ref name="IS-contents">Iqbal and Salley (2004), pp. v–viii.</ref> These include the hot gas layer temperature and smoke layer height in a room, the burning of liquid pool fires and flame heights. Further chapters treat radiant heat flux and ignition of a target, plume temperature, sprinkler and detector response, flashover and the fire resistance of steel members.<ref name="IS-contents" /> The spreadsheets need only a working knowledge of algebra.<ref name="IS-xxxvii">Iqbal and Salley (2004), p. xxxvii.</ref>

Several hand calculations have their own articles in this wiki. The [[Heskestad plume]] correlations come from the SFPE Handbook chapter on fire plumes, flame height and air entrainment, and include the flame height formula shown below.<ref name="He2-3">Heskestad (2002), p. 2-3.</ref> The [[Alpert correlations]] underlie hand calculations of the time at which heat detectors respond.<ref name="IFEG272" /> The [[MQH correlation]], named after McCaffrey, Quintiere and Harkleroad, approximates the temperature of the hot upper layer in a room from an energy balance and a fit to test data.<ref name="WT175">Walton and Thomas (2002), p. 3-175.</ref> Walton and Thomas collect such simplified techniques because compartment temperatures are used to predict the onset of hazardous conditions, damage, ignition of objects and flashover.<ref name="WT171">Walton and Thomas (2002), p. 3-171.</ref>

Walton and Thomas also explain why such methods exist. The energy equations of a room fire have no general explicit solution, so they are either solved approximately by hand, using a limiting set of assumptions, or more completely by a computer program.<ref name="WT175" /> Their hand methods predict average temperatures. They should not be used for local temperatures, such as at a detector or sprinkler or where flames touch a surface.<ref name="WT175" />

''Introduction to Fire Safety Engineering'' uses hand calculations throughout. Its heat transfer chapter assumes steady conditions, because changes over time are very hard to include in a hand calculation. The results are therefore mainly suited to finding the order of magnitude of a heat flow.<ref name="VR5.1">Van Rossum (2027), §5.1.</ref> Visibility in smoke can be predicted reasonably well with CFD, and to a lesser extent with hand calculations and zone models.<ref name="VR6.1">Van Rossum (2027), §6.1.</ref> For large malls and atria without smoke venting, the IFEG point to simple empirical correlations for the height of the smoke layer as it descends over time, for a given heat release rate.<ref name="IFEG254" />

== Example: flame height ==
Heskestad's correlation for the mean height of a flame shows how a hand calculation works. The flame height grows with the heat release rate of the fire, but a wider fire has a relatively lower flame. In its simplest form, for normal atmospheric conditions, the correlation reads:<ref name="He2-3" />

<math display="block">L = 0.235\,\dot{Q}^{2/5} - 1.02\,D</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>L</math> || mean flame height above the base of the fire || m
|-
| <math>\dot{Q}</math> || total heat release rate of the fire || kW
|-
| <math>D</math> || diameter of the fire, or the diameter of a circle with the same area || m
|}

For a small heat release rate, the formula gives a negative flame height. This is physically impossible, so the correlation is not valid there.<ref name="He2-3" /> The coefficient 0.235 is a typical value. For many gaseous and liquid fuels it lies between 0.226 and 0.240, but it is lower for some fuels, such as 0.200 for gasoline.<ref name="He2-3" />

[[File:fse-hand-calculations-v2.svg|thumb|upright=1.5|alt=Two rising curves of flame height against heat release rate for fire diameters of 0.5 m and 1.21 m; the larger fire starts below zero at small outputs.|Mean flame height from Heskestad's correlation for fires 0.5 m and 1.21 m across. Below about 63 kW the 1.21 m fire gives a negative height, where the correlation is not valid. Based on Heskestad (2002), p. 2-3.]]

The coefficient also depends on the state of the air. Heskestad shows this with a methyl alcohol pool fire 1.5 m across. Its mean flame height rises from 1.83 m under normal conditions to 2.23 m on a hot day at high altitude.<ref name="He2-4" />

'''Worked example.''' Heskestad gives the example of a 1.2 m high stack of wood pallets, 1.07 m by 1.07 m, burning at 2,600 kW.<ref name="He2-4" /> A circle with the same area has a diameter of 1.21 m. The flame height is then 0.235 × 26000.4 − 1.02 × 1.21 = 4.22 m above the base of the stack.<ref name="He2-4" /><ref name="Calc">Heskestad (2002), p. 2-4; recalculated with Equation 8 on p. 2-3.</ref>

[[File:fse-hand-calculations-v1.svg|thumb|upright=1.5|alt=Side view of a burning stack of wood pallets with a flame about three and a half times as high as the stack is wide; dimension lines mark D and L.|Worked example: a stack of wood pallets 1.07 m square, burning at 2,600 kW, has an equivalent diameter D of 1.21 m and a mean flame height L of about 4.2 m. Based on Heskestad (2002), pp. 2-3–2-4.]]

The flame is thus about 3.5 times as high as the fire is wide.<ref name="Calc" /> In ''Introduction to Fire Safety Engineering'', the same comparison of flame length and fire diameter decides which plume model is more reliable. When the flame is longer than the diameter, the Heskestad plume model applies; when it is shorter, the Thomas plume model is preferred.<ref name="VR6.7">Van Rossum (2027), §6.7.</ref>

== Example: evacuation time ==
''Introduction to Fire Safety Engineering'' sets out a hand calculation of evacuation time in eight steps.<ref name="VR7.5" /> The steps are: count the people per room, find the flow capacity of each door or stair, and divide the people over the exits. Next, find the governing exit and the walking time of the first person to it. Then add the time for everyone to pass that exit and the walking time of the last person to a safe place.<ref name="VR7.5" /> The method assumes a homogeneous group of people and a simple geometry.<ref name="VR7.5" />

In the book's example, 300 people are divided equally over four stairs, two of 2.4 m and two of 1.6 m width. The narrower stairs, with a capacity of 1.2 persons per second each, govern.<ref name="VR7.5" /> The first person walks 5 m at 1 m/s, the 75 people assigned to a narrow stair need 63 s to pass, and the last person walks a further 10 m.

[[File:fse-hand-calculations-v3.svg|thumb|upright=1.5|alt=Floor plan with 300 people and four stairs of two widths, and a time bar of 5 s walking, 63 s passing the stair and 10 s walking.|Evacuation hand calculation: 300 people divided over four stairs; the 1.6 m stairs govern, giving 5 s + 63 s + 10 s = 78 s. Based on Van Rossum (2027), §7.5.]]

The total is 5 + 63 + 10 = 78 s.<ref name="VR7.5" /> The book notes that even an equal division over the exits is a very optimistic assumption, because it is not known which exits people will actually choose.<ref name="VR7.5" /> Methods that treat people as a flow through doors and stairs in this way are known as hydraulic flow models; the IFEG describe them by the relation flow = speed × density × width.<ref name="IFEG283">National Research Council of Canada et al. (2005), p. 2.8-3.</ref> The [[hydraulic evacuation model]] article describes this approach in more detail.

== Use in design and regulation ==
Hand calculations are accepted in several regulatory frameworks, usually as one option among others. In Australia, the FSVM handbook asks the design team to agree, as part of the performance-based design brief, whether hand calculations, zone models or CFD models will be used. The choice is made for each enclosure where the time to untenable conditions must be found.<ref name="HB21" /> In New Zealand, C/VM2 (second edition, 2025) prescribes simple equations for travel time and door flow, which are themselves hand calculations.<ref name="VM2-32">Ministry of Business, Innovation and Employment (2025b), pp. 32–33.</ref> It adds that it does not give a complete guide to egress analysis.<ref name="VM2-32" />

Hand calculations also serve as a check on complex models. ''Introduction to Fire Safety Engineering'' notes that a simple approach can act as a check on a more complex one.<ref name="VR7.5" /> In the NIST investigation of the World Trade Center towers, the structural computer models were checked against hand calculations for typical floor sections. The steel stresses agreed within 4 % for the long-span trusses and 3 % for the short-span trusses. Deflections agreed within 5 to 15 %.<ref name="FG53">Faschan and Garlock (2005), p. 53.</ref>

[[File:fse-hand-calculations-v4.svg|thumb|upright=1.5|alt=Three stacked boxes for hand calculations, zone models and CFD, with arrows showing increasing detail and a return arrow labelled check.|Hand calculations, zone models and CFD as options for a design analysis; simple results are also used to check more complex models. Based on Australian Building Codes Board (2020b), p. 21, and Van Rossum (2027), §7.5.]]

In the same investigation, the failure loads of floor components, such as truss seats and shear connectors, were found either by hand calculations or by finite element analyses. Simplified models built on these results were then placed in the model of the whole floor.<ref name="GM66">Gross and McAllister (2005), p. 66.</ref> Agreement between methods still has limits. Walton notes that similar results from several models increase confidence but do not guarantee accuracy, because the models may share the same basic assumptions.<ref name="Wa190">Walton (2002), p. 3-190.</ref>

There are limits to this use. For complex buildings with large populations, the FSVM data sheets consider hand calculations of evacuation impractical, so computer models are needed. These may simply be automated hand calculations.<ref name="DS283">Australian Building Codes Board (2020a), p. 283.</ref> The IFEG note that hand calculations of smoke spread beyond the room of fire origin are more difficult and are rarely used. Their results also carry greater uncertainty.<ref name="IFEG256">National Research Council of Canada et al. (2005), p. 2.5-6.</ref>

== Accuracy and limits ==
The NRC verified and validated its library of hand calculations against experiments as part of the NUREG-1824 study.<ref name="DH1-2">Dreisbach and Hill (2007), p. 1-2.</ref> The study found that the libraries of engineering calculations had limited capabilities for the scenarios tested and lacked methods for many of the quantities of interest.<ref name="NUREG3-4" /> The correlations they do contain are based on conservation laws and widely accepted in fire protection engineering. Because they are empirical, they are subject to many limiting assumptions, and users must be cautious.<ref name="NUREG3-4" /> In the same study, the two-zone models and the CFD model reproduced most measured quantities within the experimental uncertainty.<ref name="NUREG3-4" /> The verification work also identified a number of corrections to the five models studied, ranging from inconsequential changes to errors that could lead to incorrect results.<ref name="NUREG3-4" />

The FDT spreadsheets need little or no computing time and give first-order results, either as a single value or as a series of values with a plot.<ref name="DH2-2" /> Validation for one application does not mean validation for a different scenario.<ref name="DH1-2" /> Iqbal and Salley therefore stress that the tools must be applied within the limits of their validity.<ref name="IS-xxxix" />

Precision is easily overstated. Iqbal and Salley advise users to focus on the magnitude of a result rather than its decimals. A component that is damaged at 700 °F (371 °C) is likely to be damaged at a calculated 699.99 °F as well.<ref name="IS-xl">Iqbal and Salley (2004), p. xl.</ref> They encourage bounding calculations, for example with doors open and closed or ventilation on and off, to find the range of credible results.<ref name="IS-xl" /> The IFEG ask for similar sensitivity studies, such as repeating a calculation with a fire that grows several times faster or with significantly lower walking speeds.<ref name="IFEG1224">National Research Council of Canada et al. (2005), p. 1.2-24.</ref>

''Introduction to Fire Safety Engineering'' gives similar advice on rounding. Only the final result should be rounded, inputs such as fire loads and walking speeds are often uncertain by tens of per cent, and two or at most three significant figures are realistic.<ref name="VR1.7">Van Rossum (2027), §1.7.</ref> When results are compared with limits, required times are rounded up and available times down.<ref name="VR1.7" />

== See also ==
* [[SFPE]]
* [[CFAST]]
* [[Computational fluid dynamics]]
* [[Sensitivity analysis]]
* [[Verification and validation]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Dreisbach, J., & Hill, K. (2007). ''Verification and validation of selected fire models for nuclear power plant applications: Vol. 3. Fire Dynamics Tools (FDTs)'' (NUREG-1824; EPRI 1011999). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* Faschan, W. J., & Garlock, R. B. (2005). ''Reference structural models and baseline performance analysis of the World Trade Center towers'' (NIST NCSTAR 1-2A). National Institute of Standards and Technology.
* Gross, J. L., & McAllister, T. P. (2005). ''Structural fire response and probable collapse sequence of the World Trade Center towers'' (NIST NCSTAR 1-6). National Institute of Standards and Technology.
* Heskestad, G. (2002). Fire plumes, flame height, and air entrainment. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-1–2-17). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* Najafi, B., Joglar, F., & Dreisbach, J. (2007). ''Verification and validation of selected fire models for nuclear power plant applications: Vol. 1. Main report'' (NUREG-1824; EPRI 1011999). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Walton, W. D. (2002). Zone computer fire models for enclosures. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-189–3-193). National Fire Protection Association; Society of Fire Protection Engineers.
* Walton, W. D., & Thomas, P. H. (2002). Estimating temperatures in compartment fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-171–3-188). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Modelling]]

Smoke barrier

2026-11-02T17:14:29Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''smoke barrier''' is a continuous surface, such as a wall, floor, ceiling assembly or damper in a duct, that is designed and built to restrict the movement of smoke. It may or may not also have a fire resistance rating, and it may contain protected openings such as doors.<ref name="IS-F50">Iqbal and Salley (2004), p. F-50.</ref> Smoke barriers divide a building into compartments, often called smoke zones, so that smoke from a fire stays in or near the zone where it starts. This approach is known as passive smoke management or compartmentation.<ref name="EV-69">Evans et al. (2005), p. 69.</ref>

No smoke barrier is completely tight. Smoke passes through cracks, gaps around doors and holes around pipes, and the amount depends on the size of these leakage paths and on the pressure difference across the barrier.<ref name="KL-277">Klote (2002), p. 4-277.</ref><ref name="KL-278">Klote (2002), p. 4-278.</ref> That pressure difference comes from the fire itself and from stack effect, wind and the ventilation system.<ref name="KL-278" /> Barriers are therefore often combined with fans that keep the protected side at a higher pressure, which is the basis of mechanical [[smoke control]].<ref name="KL-278" /> In performance-based design, a barrier is described by its leakage area or leakage rate at a given temperature. Engineers also have to allow for defects such as unsealed service penetrations and doors left open.<ref name="AB-147">Australian Building Codes Board (2020a), Data Sheet C5, p. 147.</ref><ref name="AB-156">Australian Building Codes Board (2020a), Data Sheet C5, p. 156.</ref>

== Purpose ==
Smoke is recognised as the major killer in fires. Klote notes that smoke often flows to places far from the fire, so that stairs and lift shafts become filled with smoke and evacuation is blocked.<ref name="KL-274">Klote (2002), p. 4-274.</ref> Smoke barriers limit that spread. In the Australian verification method, they form part of a smoke containment strategy that keeps smoke out of adjacent enclosures and escape routes long enough for evacuation and fire brigade intervention.<ref name="AB-147" />

Barriers are the oldest means of smoke management, together with smoke vents and smoke shafts.<ref name="KL-277" /> In the World Trade Center buildings, smoke barriers built into the architecture and the ductwork were one of two primary means of controlling smoke. The other was air movement equipment.<ref name="EV-69" /> The International Fire Engineering Guidelines list smoke barriers, together with natural smoke venting and mechanical smoke management, as protective measures in the sub-system for smoke development, spread and control.<ref name="IFEG-1.2-10">National Research Council of Canada et al. (2005), p. 1.2-10.</ref>

The choice of barriers affects the whole fire analysis. In a zone model, a fire in an adjacent room that is not separated from the main space by a fire- and smoke-resisting construction is often the governing scenario. Smoke from such a room mixes with more air on its way out and produces a much larger smoke volume.<ref name="VR-9.3a">Van Rossum (2027), §9.3 (How a zone model works).</ref> In large open spaces such as atria and covered malls, however, barriers and pressure differences are not practical, and other strategies are needed.<ref name="MI-292">Milke (2002b), p. 4-292.</ref>

== Types of smoke barrier ==
Walls and floors are the basic elements. In Australia, the deemed-to-satisfy provisions of the National Construction Code allow smoke-proof walls of non-combustible construction, which may include safety glass. The Code does not set an acceptable leakage limit.<ref name="AB-148">Australian Building Codes Board (2020a), Data Sheet C5, p. 148.</ref> In New Zealand, C/VM2 (second edition, 2025) speaks of smoke separations and requires the trial design to identify them together with fire separations and their closures.<ref name="MB-14">Ministry of Business, Innovation and Employment (2025b), p. 14.</ref>

Openings in a barrier need their own protection. Smoke doors must be built so that smoke does not pass from one side of the doorway to the other.<ref name="AB-151">Australian Building Codes Board (2020a), Data Sheet C5, p. 151.</ref> Where ducts pass through a barrier, smoke dampers are used. A smoke damper is designed to stop smoke when closed and is closed by a signal from a smoke detector or the building control system. A fire damper is designed to stop fire and is closed by heat that melts a fusible link.<ref name="EV-71">Evans et al. (2005), p. 71.</ref> Smoke dampers are typically designed to close if power to the actuator is lost.<ref name="EV-71" /> In the United States, smoke dampers are classified by leakage, from Class 1 (lowest) to Class 4, and by the elevated temperature at which they are tested.<ref name="IS-F19">Iqbal and Salley (2004), p. F-19.</ref>

Barriers do not always reach the floor. Draft curtains are barriers hung from the roof to limit the spread of smoke beneath it.<ref name="IS-F21">Iqbal and Salley (2004), p. F-21.</ref> They divide a large space into curtained sections, each forming a [[smoke reservoir]] in which a [[hot gas layer]] collects and from which it can be vented.<ref name="CO-220">Cooper (2002b), p. 3-220.</ref> If the layer drops below the bottom of the curtain, smoke flows out underneath it.<ref name="CO-229">Cooper (2002b), p. 3-229.</ref> NFPA 204, as cited by Cooper, recommends that such curtains reach down at least one fifth of the distance from the ceiling to the fire.<ref name="CO-234">Cooper (2002b), p. 3-234.</ref> Movable elements such as retractable screens, like doors and dampers, change position in a fire, and the guidelines ask for the delay of that change to be specified (see also [[fire curtain]]).<ref name="IFEG-1.5-5">National Research Council of Canada et al. (2005), p. 1.5-5.</ref>

== How smoke passes a barrier ==
Several forces drive [[smoke movement]] in a building: stack effect, buoyancy of the hot smoke, expansion of the heated gases, wind and the ventilation system.<ref name="KL-274" /> Each of these creates pressure differences across walls and floors. The hot gases in a fire room are lighter than the surrounding air. Above the neutral plane, where the pressures inside and outside are equal, this produces an overpressure that pushes smoke through any leakage path.<ref name="KL-276">Klote (2002), p. 4-276.</ref>

[[File:fse-smoke-barrier-v1.svg|thumb|upright=1.5|alt=Section through a wall between a smoke-filled room and a corridor, with small arrows of smoke through door gaps, a pipe hole and the wall-floor joint.|Leakage paths through a smoke barrier: gaps around the door, holes around penetrating pipes and cracks at the wall-floor joint. Smoke moves through them because of the pressure difference across the barrier. Based on Klote (2002), pp. 4-276 to 4-278.]]

The pressures involved are small. For a fire room at 800 °C, the buoyancy pressure 1.52 m above the neutral plane is about 13 Pa. In full-scale fire tests, the highest pressure difference measured across the wall of the burn room at the ceiling was 16 Pa.<ref name="KL-276" /> Buoyancy weakens as smoke moves away from the fire, because the smoke cools and is diluted.<ref name="KL-276" /> Even so, small pressure differences are enough to push smoke through the many small gaps in ordinary construction.<ref name="KL-278" />

Leakage depends mainly on workmanship rather than on the materials used.<ref name="KL-282">Klote (2002), p. 4-282.</ref> Klote gives typical leakage areas as a fraction of the wall or floor area, based on measurements by the National Research Council of Canada:<ref name="KL-282" />

{| class="wikitable"
|+ Typical leakage area ratios of walls and floors in commercial buildings<ref name="KL-282" />
! Construction element !! Tight !! Average !! Loose
|-
| Stairwell walls (without door gaps) || 0.14 × 10−4 || 1.1 × 10−4 || 3.5 × 10−4
|-
| Floors (including gaps around penetrations) || 0.066 × 10−4 || 0.52 × 10−4 || 1.7 × 10−4
|}

Doors add much more leakage. A door of 0.9 m by 2.1 m with an average gap of 3.2 mm already has a leakage area of 0.020 m².<ref name="KL-282" /> The Australian Building Codes Board notes that where smoke doors or dampers are present, leakage can be substantially greater than that of the barrier alone, even when they close properly.<ref name="AB-157">Australian Building Codes Board (2020a), Data Sheet C5, p. 157.</ref>

=== Leakage through a barrier ===
The flow of air or smoke through a crack or gap follows the orifice equation. The flow rises with the leakage area and with the square root of the pressure difference:<ref name="KL-280">Klote (2002), p. 4-280.</ref><ref name="AB-149">Australian Building Codes Board (2020a), Data Sheet C5, p. 149.</ref>

<math display="block">\dot{V} = C A \sqrt{\frac{2 \Delta p}{\rho}}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>\dot{V}</math> || volume flow through the leakage path || m³/s
|-
| <math>C</math> || flow coefficient, typically 0.6 to 0.7; 0.65 is commonly used || –
|-
| <math>A</math> || leakage area (flow area) || m²
|-
| <math>\Delta p</math> || pressure difference across the barrier || Pa
|-
| <math>\rho</math> || density of the gas passing through || kg/m³
|}

The equation applies to cracks and gaps of fixed geometry, for all but extremely narrow cracks.<ref name="KL-280" /> As an example, Klote gives a closed door with a crack area of 0.01 m². At 2.5 Pa it leaks about 0.013 m³/s of air, and at 75 Pa about 0.073 m³/s.<ref name="KL-281">Klote (2002), p. 4-281.</ref> For a new smoke-proof wall with service penetrations, the Board takes a leakage area ratio of 1.5 × 10−4. At 200 °C and 25 Pa this gives roughly 3 m³/h per square metre of wall.<ref name="AB-150">Australian Building Codes Board (2020a), Data Sheet C5, p. 150.</ref>

Consider a wall 10 m long and 3 m high, so 30 m². Its leakage area is 30 × 1.5 × 10−4 = 0.0045 m². Smoke at 200 °C, or 473 K, has a density of about 353/473 ≈ 0.75 kg/m³.<ref name="VR-5.3">Van Rossum (2027), §5.3.</ref> With C = 0.65 and a pressure difference of 25 Pa, the flow is about 0.024 m³/s, or 86 m³/h.<ref name="AB-149" /> A single-leaf smoke door meeting AS 6905 may leak up to 40 m³/h under similar conditions. The door alone therefore adds about half as much again as the whole wall.<ref name="AB-151" /><ref name="AB-152">Australian Building Codes Board (2020a), Data Sheet C5, p. 152.</ref>

[[File:fse-smoke-barrier-v2.svg|thumb|upright=1.5|alt=Bar chart of leakage in cubic metres per hour for a 30 square metre wall: tight 9, average 58, with services 86, loose 201, plus 40 for one smoke door.|Calculated smoke leakage through a 30 m² wall at 200 °C and 25 Pa for different tightness levels, compared with the permitted leakage of one single-leaf smoke door. Based on Australian Building Codes Board (2020a), Data Sheet C5, pp. 149 to 155.]]

== Performance requirements and testing ==
Smoke barriers are specified for three temperature conditions: ambient (20 ± 10 °C), medium (200 ± 20 °C) and high, representing a fully developed fire.<ref name="AB-147" /><ref name="AB-148" /> A wall that only meets the smoke-proof provisions is unlikely to remain an effective barrier at high temperatures. The Australian provisions are therefore read as aiming at ambient and medium conditions, unless a fire resistance level is also required.<ref name="AB-148" /> At high temperatures, leakage can be much higher, even when the fire resistance criteria are met, and the barrier itself may give off smoke.<ref name="AB-151" />

The Board expects a smoke-proof wall to keep a leakage area ratio of no more than 1.5 × 10−4. The exposure is a temperature rising linearly to 200 °C over 30 minutes and then held for another 30 minutes.<ref name="AB-150" /> Under the Australian standard AS 6905, a single-leaf smoke door may leak 40 m³/h and a two-leaf door 65 m³/h at medium temperature and 25 Pa. The test follows at least 30 minutes of exposure at 200 °C.<ref name="AB-151" /><ref name="AB-152" /> Smoke dampers under AS 1682.1 may leak no more than 360 m³/h per square metre at 300 Pa and must close within 30 seconds.<ref name="AB-152" />

In New Zealand, C/VM2 (second edition, 2025) sets fixed modelling assumptions. Fire and smoke separations other than doors are taken to have zero leakage.<ref name="MB-17">Ministry of Business, Innovation and Employment (2025b), p. 17.</ref> A smoke control door is modelled with a 10 mm gap under the door, and a fire door without smoke seals with a 10 mm gap over its full height. Unrated lined walls get a leakage area of 0.1 % of their surface.<ref name="MB-17" /> Smoke separations that comply with the relevant product specification remain in place until their rated temperature or until flashover, whichever comes first.<ref name="MB-14" /> A separation not designed to a recognised standard, such as plain toughened glass, is assumed to disappear once the average upper layer temperature reaches 200 °C.<ref name="MB-15">Ministry of Business, Innovation and Employment (2025b), p. 15.</ref>

== Barriers and pressurisation ==
Mechanical smoke control uses the barriers of traditional smoke management in combination with airflows and pressure differences created by fans. Air flowing outward through the small gaps around a closed door then prevents smoke from entering the protected side.<ref name="KL-278" /> Klote lists three advantages over barriers alone. Smoke control depends less on tight barriers. Stack effect, buoyancy and wind are less likely to overcome it. It can also prevent smoke flowing through a door that is open.<ref name="KL-279">Klote (2002), p. 4-279.</ref>

[[File:fse-smoke-barrier-v3.svg|thumb|upright=1.5|alt=Wall with a door between smoke and an escape route: left, smoke leaks through gaps; right, fan-driven air flows out through the gaps.|A smoke barrier alone (left) and a barrier with pressurisation (right): air flowing outward through the gaps keeps smoke on the fire side. Based on Klote (2002), pp. 4-278 to 4-279.]]

The pressure difference must be large enough to overcome the forces of the fire, but not so large that doors become hard to open. In the United States, NFPA 101 limits the force needed to open a door on an escape route to 133 N.<ref name="KL-285">Klote (2002), p. 4-285.</ref> For a fire close to a door in the barrier, Klote suggests a minimum pressure difference of 20 to 25 Pa. For smoke at 400 °C from a remote fire, 15 to 20 Pa is suggested, and 5 to 10 Pa where sprinklers cool the smoke.<ref name="KL-285" /> Field tests have shown pressure fluctuations of up to 5 Pa across partitions and closed doors, mostly attributed to wind.<ref name="KL-281" />

Dampers in ducts link barriers and ventilation systems. In the World Trade Center towers, combination fire and smoke dampers were not required. NIST found that installing them might have slowed the development of hazardous conditions on the upper floors. It would probably not have improved escape, because the stairs were impassable.<ref name="EV-xxxiv">Evans et al. (2005), p. xxxiv.</ref>

== Reliability and defects ==
Openings and defects largely determine how well a barrier performs.<ref name="AB-157" /> Among the factors that reduce reliability, the Board lists unprotected openings in fire and smoke barriers and fire and smoke doors that are wedged open.<ref name="AB-146">Australian Building Codes Board (2020a), Data Sheet C5, p. 146.</ref> For doors and dampers, reliability is the probability that they close when needed. For walls with service penetrations it is more complex, because an unsealed gap may or may not matter for the outcome.<ref name="AB-153">Australian Building Codes Board (2020a), Data Sheet C5, p. 153.</ref><ref name="AB-154">Australian Building Codes Board (2020a), Data Sheet C5, p. 154.</ref>

The Board therefore uses a distribution of leakage rather than a single reliability value. For a fully visible barrier without penetrations, it suggests a tight, average or loose wall with leakage area ratios of 0.15, 1.0 and 3.5 × 10−4 and probabilities of 25, 50 and 25 %.<ref name="AB-155">Australian Building Codes Board (2020a), Data Sheet C5, p. 155.</ref> Where the barrier is partly hidden above a false ceiling, defects become more likely. Based on inspections of more than 5,000 passive fire protection systems in New Zealand, 17 % of small penetrations such as single cables and 37 % of large penetrations such as cable trays were taken as unprotected.<ref name="AB-156" /><ref name="AB-158">Australian Building Codes Board (2020a), Data Sheet C5, p. 158.</ref> Each unprotected large penetration adds an open area of 100 × 10−4 m², more than twice the leakage area of the whole 30 m² wall in the example above.<ref name="AB-156" />

== Limitations and misconceptions ==
A common misconception is that a smoke barrier stops smoke completely. All barriers leak, and the leakage grows with the pressure difference and with temperature.<ref name="KL-278" /><ref name="AB-151" /> Smoke separations often have no fire rating. C/VM2 therefore treats a separation that does not meet a recognised standard as no barrier at all beyond flashover or 200 °C in the upper layer.<ref name="MB-15" />

Doors are a particular weak point. They are opened during evacuation and are sometimes left or propped open throughout a fire, and without mechanical smoke control, smoke flow through these doors is common.<ref name="KL-279" /> Purging, the supply and extraction of large amounts of air, cannot keep the fire space breathable or stop smoke movement on its own. It can, however, significantly limit the smoke level in spaces separated from the fire by smoke barriers.<ref name="KL-279" />

Finally, a barrier designed for one scenario may be bypassed in another. When fire scenarios are defined, whether doors and dampers are open or closed, whether smoke seals are fitted and whether leakage through barriers is controlled are treated as separate outcomes.<ref name="IFEG-2.2-3">National Research Council of Canada et al. (2005), p. 2.2-3.</ref> The performance of a barrier must also be confirmed by smoke modelling for the specific building, because its efficacy depends on the other fire protection systems and the building layout.<ref name="AB-153" />

== See also ==
* [[Smoke and heat exhaust ventilation]]
* [[Stack effect]]
* [[Stair pressurisation]]
* [[Atrium smoke control]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Cooper, L. Y. (2002b). Smoke and heat venting. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-219–3-242). National Fire Protection Association; Society of Fire Protection Engineers.
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Milke, J. A. (2002b). Smoke management in covered malls and atria. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-292–4-310). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Smoke Control]]

Pool fire

2026-11-02T17:14:26Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''pool fire''' is a fire burning on a horizontal, upward-facing layer of fuel, usually a flammable or combustible liquid in a container, a bund or a spill on the ground.<ref name="IS-3-5">Iqbal and Salley (2004), p. 3-5.</ref> The liquid itself does not burn: heat from the flames evaporates it, and the vapour burns above the surface.<ref name="VR-4.1">Van Rossum (2027), §4.1.</ref> The term can also cover flat slabs of solids that decompose much like liquids, such as polymethylmethacrylate and polyethylene.<ref name="IS-3-5" />

In United States nuclear power plants, for example, liquid spill fires are among the most commonly assumed fires of transient fuels.<ref name="IS-3-3" /> Such pool fires can result from leaking pump seals, oil spills from transformers, fuel spray from diesel generators, and liquids used for cleaning and painting.<ref name="IS-3-6">Iqbal and Salley (2004), p. 3-6.</ref> They can be large: a burning pool of gasoline of 1 m² releases about 2.5 MW, about the same as an average burning living-room sofa.<ref name="IS-3-2">Iqbal and Salley (2004), p. 3-2.</ref><ref name="VR-4.5">Van Rossum (2027), §4.5.</ref> The size of the fire follows mainly from the area of the pool and the burning rate of the fuel, and the main hazard to the surroundings is thermal radiation from the large, sooty flames.<ref name="GW-2-297">Gottuk and White (2002), p. 2-297.</ref><ref name="Bey-3-268">Beyler (2002a), p. 3-268.</ref> Engineers estimate the heat release rate, burning duration, flame height and radiant heat flux with simple correlations that rest on large-scale tests.<ref name="IS-3-3">Iqbal and Salley (2004), p. 3-3.</ref><ref name="Bey-3-269">Beyler (2002a), p. 3-269.</ref>

== Pools and spills ==
Gottuk and White distinguish two situations. A pool is a confined body of fuel, typically more than 1 cm deep, that forms when a release collects in a low spot such as a trench, or that exists in tanks and containers.<ref name="GW-2-297" /> A spill is a thin layer of fuel from an unconfined release, and its fire is highly variable, depending on the source, the surface it spreads over and the moment of ignition.<ref name="GW-2-297" /> Spill depths for liquid spill fires range from 0.7 mm to 4 mm, depending mainly on the quantity released.<ref name="GW-2-299">Gottuk and White (2002), p. 2-299.</ref>

The difference matters for the burning rate. In tests with jet fuel spilled on concrete, unconfined spill fires larger than about 1 m in diameter burned at roughly one fifth of the maximum burning rate of confined pool fires.<ref name="GW-2-311">Gottuk and White (2002), p. 2-311.</ref> The authors attribute this to greater heat loss from the thin fuel layer, about 1 mm deep, to the concrete.<ref name="GW-2-311" /> For a given amount of fuel, a spill with a large area burns with a high heat release rate for a short time. A spill with a small area burns with a lower rate for longer.<ref name="IS-3-5" />

In the event tree of Beyler for a release of flammable material, a pool fire follows when a spill accumulates and is ignited locally after a delay.<ref name="Bey-3-269" /> The geometry of the containment, or the lack of it, strongly influences its characteristics.<ref name="Bey-3-269" /> A pool fire can also follow a vapour cloud fire that burns back to the release point.<ref name="Bey-3-269" /> In quantitative risk assessment it is one of the radiant heat outcomes, defined as a fire involving a pool of confined or spreading flammable material.<ref name="Bar-5-184">Barry (2002), p. 5-184.</ref>

== How a pool fire burns ==
A liquid has to turn into vapour before it can burn.<ref name="VR-4.1" /> Part of the heat of the flames therefore has to return to the liquid surface to keep the evaporation going. The balance between that heat and the energy needed to evaporate the fuel controls how fast the fuel burns.<ref name="VR-4.3">Van Rossum (2027), §4.3.</ref> The [[burning rate]] is usually given as the mass of fuel burned per square metre of pool surface per second.<ref name="IS-3-3" />

[[File:fse-pool-fire-v1.svg|thumb|upright=1.5|alt=Cross-section of a burning liquid pool in a bund: vapour rises from the liquid, a tall flame burns above it with black smoke, arrows show heat back to the pool and radiation to a target.|A pool fire: heat fed back from the flame evaporates the liquid, the vapour burns above the pool, and the flame radiates heat to its surroundings. Based on Van Rossum (2027), §4.1 and §4.3, and Beyler (2002a), pp. 3-270 and 3-277.]]

The [[flash point]] is the lowest temperature at which a flammable mixture of vapour and air exists at the surface of a liquid.<ref name="IS-6-1">Iqbal and Salley (2004), p. 6-1.</ref> At that temperature the vapour pressure is just high enough to give a mixture at the lower flammability limit.<ref name="S-1-147">Simmons (2002), p. 1-147.</ref> In the NIST experiments for the World Trade Center investigation, for example, jet fuel with a flash point of 43 to 66 °C did not at first form a flammable vapour layer at room temperature.<ref name="Ohl-42">Ohlemiller et al. (2005), p. 42.</ref> The radiant heat of the burner first had to warm the fuel.<ref name="Ohl-42" /> Once ignited, flames spread over flammable and combustible liquids at about 10 cm/s to 2 m/s.<ref name="IS-3-6" />

The way a pool fire burns changes with its diameter. NUREG-1805, after Babrauskas, distinguishes four modes: convective and laminar below 0.05 m, convective and turbulent below 0.2 m, radiative and optically thin from 0.2 to 1.0 m, and radiative and optically thick above 1.0 m.<ref name="IS-3-5" /> In pools larger than about 1 m, radiation from the optically thick flame dominates the heat transfer to the surface.<ref name="IS-3-5" /> Above a diameter of about 0.2 m the burning rate per unit area increases with diameter, and it approaches a maximum value for very large pools.<ref name="GW-2-310">Gottuk and White (2002), p. 2-310.</ref> Fuel blends with components of very different volatility do not burn at a constant rate. The most volatile components burn first, and later the burning resembles that of the heavier components that remain.<ref name="GW-2-310" />

== Heat release rate ==
The heat release rate of a pool fire is found from the burning rate per unit area, the effective heat of combustion and the area of the pool.<ref name="IS-3-3" /> A correction factor accounts for the lower burning rate of smaller pools: it is close to zero for very small pools and approaches one for large ones.<ref name="IS-3-3" /><ref name="GW-2-310" /> NUREG-1805 gives the relation as follows.<ref name="IS-3-3" />

<math display="block">\dot Q = \dot m''\, \Delta H_\mathrm{c,eff}\, A_\mathrm{f} \left(1 - e^{-k\beta D}\right)</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\dot Q</math> || heat release rate of the pool fire || kW
|-
| <math>\dot m''</math> || mass burning rate per unit area of a large pool || kg/(m²·s)
|-
| <math>\Delta H_\mathrm{c,eff}</math> || effective heat of combustion || kJ/kg
|-
| <math>A_\mathrm{f}</math> || horizontal burning area of the fuel || m²
|-
| <math>k\beta</math> || empirical constant of the fuel || m−1
|-
| <math>D</math> || diameter of the burning area || m
|}

For a pool that is not circular, an equivalent diameter is used: the diameter of a circle with the same area.<ref name="GW-2-310" /> The relation applies to confined pools burning in the open, in still air, in a pan or tank without an excessive lip height.<ref name="GW-2-311" /> The fuel data come from large-pool tests compiled by Babrauskas. For gasoline the burning rate is 0.055 kg/(m²·s), the effective heat of combustion 43,700 kJ/kg and ''kβ'' 2.1 m−1. For transformer oil the values are 0.039 kg/(m²·s), 46,400 kJ/kg and 0.7 m−1.<ref name="IS-3-4">Iqbal and Salley (2004), p. 3-4.</ref> Where the constant ''kβ'' is unknown, NUREG-1805 advises a conservative value of 100 m−1, which sets the bracket to one.<ref name="IS-3-6" />

[[File:fse-pool-fire-v2.svg|thumb|upright=1.5|alt=Two rising curves of heat release rate per square metre against pool diameter; gasoline levels off near 2,400 kW/m² by 2 m, transformer oil rises more slowly.|Heat release rate per square metre of pool surface against pool diameter for gasoline and transformer oil, from the large-pool data and the diameter correction. Based on Iqbal and Salley (2004), pp. 3-3 to 3-4.]]

'''Worked example.''' NUREG-1805 considers a spill of 5 US gallons (18.9 L) of lubricating oil that covers 9.0 ft² (0.836 m²) of a concrete floor. It uses the properties of transformer oil, with an effective heat of combustion of 46,000 kJ/kg.<ref name="IS-3-6" /><ref name="IS-3-22">Iqbal and Salley (2004), p. 3-22.</ref> The equivalent diameter is 1.03 m, so the bracket is 1 − e−0.7 × 1.03 ≈ 0.514. Then:

<math display="block">\dot Q = 0.039 \times 46{,}000 \times 0.836 \times 0.514 \approx 771\ \mathrm{kW}</math>

This agrees with the 772 kW in the NUREG-1805 solution; the small difference comes from rounding.<ref name="IS-3-23">Iqbal and Salley (2004), p. 3-23.</ref> Without the diameter correction the result would be about 1.5 MW, so for a pool of this size the correction roughly halves the heat release rate. For a large gasoline pool the burning rate and heat of combustion give about 2.4 MW per square metre, consistent with the 2.5 MW quoted for a 1 m² pool.<ref name="IS-3-4" /><ref name="IS-3-2" /> Inside a room, the heat release rate can differ from these free-burn values. Radiation from the hot smoke layer and the heated ceiling and walls increases the burning rate, while small openings limit the oxygen supply and reduce it.<ref name="IS-3-5" />

== Burning duration and flame height ==
A fire that burns at a constant heat release rate consumes the fuel at a constant rate, so the burning duration follows from the amount of fuel and the burning rate.<ref name="IS-3-7">Iqbal and Salley (2004), p. 3-7.</ref> For a liquid pool, the fuel surface drops at the regression rate, which is the mass burning rate per unit area divided by the density of the liquid.<ref name="IS-3-8">Iqbal and Salley (2004), p. 3-8.</ref> In the lubricating oil example the regression rate is about 0.05 mm/s, and the 18.9 L spill burns for about 7.35 minutes.<ref name="IS-3-23" /> The same volume spread over a larger area would burn more intensely for a shorter time.<ref name="IS-3-8" />

The flame height matters because it shows whether flames reach objects above the pool and because it sets the size of the radiating flame.<ref name="Bey-3-270">Beyler (2002a), p. 3-270.</ref> Several correlations exist, based on the pool diameter and the burning rate or the heat release rate; the Heskestad correlation is used in both NUREG-1805 and the SFPE handbook.<ref name="Bey-3-270" /><ref name="Bey-3-274">Beyler (2002a), p. 3-274.</ref> For the lubricating oil fire it gives a flame height of 2.31 m, and the Thomas correlation gives 2.67 m.<ref name="IS-3-23" /> Both exceed the 8 ft (2.4 m) to a cable tray above the pool, so the example concludes that the flames impinge on the tray.<ref name="IS-3-23" /><ref name="IS-3-22" /> Pool fire flames fluctuate, so the flame tip moves well beyond the continuous luminous region, and flame height is defined in different ways in different studies.<ref name="Bey-3-270" /> Wind tilts the flame and changes its visible length.<ref name="Bey-3-270" /><ref name="Bey-3-271">Beyler (2002a), p. 3-271.</ref>

== Thermal radiation ==
Thermal radiation is the main way in which large open hydrocarbon fires injure people and damage property.<ref name="Bey-3-268" /> The [[flame radiation]] comes mainly from water vapour, carbon dioxide and soot.<ref name="Bey-3-270" /> Fires of flammable liquids typically have higher heat release rates per unit area than fires of ordinary combustibles, taller flames and more luminous soot. An observer near a liquid fire therefore feels more heat than near an ordinary fire of similar size.<ref name="IS-5-2">Iqbal and Salley (2004), p. 5-2.</ref>

Estimating the radiation involves three steps: describing the size of the fire, describing its radiative properties, and calculating the heat flux at the target.<ref name="Bey-3-269" /> The simplest method treats the flame as a point source at the centre of the fire, radiating a fraction of the heat release rate.<ref name="IS-5-3">Iqbal and Salley (2004), p. 5-3.</ref> That fraction ranges from about 0.15 for low-sooting fuels such as alcohol to 0.60 for high-sooting hydrocarbons, and cold soot around large flames can reduce it considerably.<ref name="IS-5-3" /> The point source method is accurate at large distances, but it is not a good choice where ignition of nearby combustibles must be assessed.<ref name="Bey-3-275">Beyler (2002a), p. 3-275.</ref>

The [[solid flame model]] represents the flame as a cylinder with the pool diameter and the flame height, whose surface radiates with an average emissive power.<ref name="Bey-3-276">Beyler (2002a), p. 3-276.</ref> The heat flux at a target equals that emissive power times the view factor between flame and target.<ref name="Bey-3-276" /> Shokri and Beyler found that the effective emissive power falls as the pool gets larger, because black smoke outside the flame increasingly hides the luminous flame.<ref name="Bey-3-277">Beyler (2002a), p. 3-277.</ref> Their correlation gives about 57 kW/m² for a 1 m pool, 48 kW/m² for a 10 m pool and 22 kW/m² for a 50 m pool.<ref name="Bey-3-277" /> Most of the uncertainty lies in the emissive power, not in the view factor.<ref name="Bey-3-277" />

[[File:fse-pool-fire-v3.svg|thumb|upright=1.5|alt=Falling curve of effective emissive power from about 57 kW/m² at 1 m pool diameter to about 22 kW/m² at 50 m.|Effective emissive power of large hydrocarbon pool fire flames against pool diameter, after the correlation of Shokri and Beyler. Based on Beyler (2002a), p. 3-277.]]

For quick screening, Shokri and Beyler also correlated the radiant heat flux to a vertical target at ground level with the distance from the pool centre divided by the pool diameter.<ref name="Bey-3-273">Beyler (2002a), p. 3-273.</ref> Above ground level the heat flux is expected to be higher than the correlation gives, because the radiation is strongest near the mid-height of the flame.<ref name="Bey-3-273" /> A safety factor of 2 is recommended for design use with this correlation, but not where a realistic estimate is needed.<ref name="Bey-3-273" />

== Protection and firefighting ==
Containment limits the size of a pool fire. A dike or curbing confines a spill to a known area, and the area, together with the burning rate, sets the heat release rate.<ref name="IS-3-5" /><ref name="IS-3-6" /><ref name="GW-2-297" /> A pool in a bund therefore burns longer than an unconfined spill of the same volume, but with a smaller fire.<ref name="IS-3-8" />

Water is often a poor extinguishing agent for these fires. A liquid that is lighter than water and does not mix with it floats on the water and keeps burning.<ref name="IS-B-58">Iqbal and Salley (2004), p. B-58.</ref> [[Firefighting foam]] is therefore widely used for fires of flammable and combustible liquids: it keeps air from the fuel, cools as the water drains from it, and protects surfaces that are not yet burning.<ref name="IS-B-58" /> It must be applied in sufficient volume and at a sufficient rate to make up for the foam destroyed by heat and flames.<ref name="IS-B-59">Iqbal and Salley (2004), p. B-59.</ref> Fixed foam systems are used in bunds and on spills, and subsurface injection of low-expansion foam into hydrocarbon storage tanks is a recognised design method in the NFPA standards.<ref name="IS-B-60">Iqbal and Salley (2004), p. B-60.</ref>

Tank fires carry a further hazard. When water penetrates a hot oil whose temperature is well above the boiling point of water, it turns into steam below the surface and causes an eruption of burning oil, called [[boilover]].<ref name="IS-B-58" /> NUREG-1805 describes boilover as a phenomenon of fires in open tanks containing blends such as crude oil, which requires water at the bottom of the tank.<ref name="IS-F-9">Iqbal and Salley (2004), p. F-9.</ref> Foam applied to a liquid at 100 °C or above can also cause boilover, because the steam formed expands the foam fourfold.<ref name="IS-B-60" />

Fire resistance tests for structures near pool fires need a different exposure. Most standard tests simulate a cellulosic fire, which grows more slowly than a hydrocarbon fire, although it may ultimately become hotter; specific hydrocarbon fire exposures for pool and jet fires were developed later.<ref name="IS-C-8">Iqbal and Salley (2004), p. C-8.</ref> In the United States, UL 1709 rates barriers for a hydrocarbon fire exposure, with H ratings for 120 minutes against the passage of flame and heat.<ref name="IS-C-9">Iqbal and Salley (2004), p. C-9.</ref> In Australia, the Fire Safety Verification Method data sheets (2020) refer to results of hydrocarbon fire tests and to a [[hydrocarbon fire curve|hydrocarbon heating regime]] alongside the standard one.<ref name="ABCB-233">Australian Building Codes Board (2020a), p. 233.</ref>

== Limits and uncertainty ==
The methods for pool fires are largely semi-empirical and rest on experimental data, so they carry uncertainties.<ref name="Bey-3-269" /> The burning rate correlations come from confined pools in the open and in still air. There are very few data for unconfined spills, and wind and lip height can change the burning rate in either direction.<ref name="GW-2-311" /><ref name="GW-2-313">Gottuk and White (2002), p. 2-313.</ref> The radiation methods assume a circular or nearly circular pool.<ref name="IS-5-17">Iqbal and Salley (2004), p. 5-17.</ref> The emissive power correlation was developed from fires of kerosene, fuel oil, gasoline, jet fuels and liquefied natural gas, with pool diameters from 1 to 50 m.<ref name="IS-5-17" />

Inside a building, the free-burn values are only a starting point. Feedback from the smoke layer and the walls, and the supply of air through openings, change the burning rate.<ref name="IS-3-5" /> The heat release rate is not a fundamental property of a fuel and is usually determined from tests.<ref name="IS-3-1">Iqbal and Salley (2004), p. 3-1.</ref>

== See also ==
* [[Aqueous film-forming foam]]
* [[BLEVE]]
* [[Jet fire]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
* Beyler, C. L. (2002a). Fire hazard calculations for large, open hydrocarbon fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-268–3-314). National Fire Protection Association; Society of Fire Protection Engineers.
* Gottuk, D. T., & White, D. A. (2002). Liquid fuel fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-297–2-316). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Ohlemiller, T. J., Mulholland, G. W., Maranghides, A., Filliben, J. J., & Gann, R. G. (2005). ''Fire tests of single office workstations'' (NIST NCSTAR 1-5C). National Institute of Standards and Technology.
* Simmons, R. F. (2002). Premixed burning. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-144–1-154). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Explosions & Industrial Fires]]

Voice alarm system

2026-11-02T17:14:24Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''voice alarm system''' is an [[evacuation alarm system]] that warns the occupants of a building with spoken messages through loudspeakers. The messages replace or accompany tones such as bells, horns or a slow whoop.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref><ref name="SMC-4-32">Schifiliti et al. (2002), p. 4-32.</ref> The messages can be prerecorded or spoken live by staff or the fire service from a control point.<ref name="Proulx-3-350">Proulx (2002), p. 3-350.</ref><ref name="KG-44">Keough and Grill (2005), p. 44.</ref> Larger systems divide the building into speaker zones, so that different floors can receive different messages.<ref name="KG-27">Keough and Grill (2005), p. 27.</ref><ref name="ABCB-97">Australian Building Codes Board (2020a), p. 97.</ref> In the sources the same function is also called an emergency voice communication system, a public address system or an emergency warning and intercommunication system.<ref name="KG-44" /><ref name="ABCB-97" />

Spoken messages address a known problem: people often hesitate when they hear a plain alarm signal.<ref name="Bryan-3-315">Bryan (2002), p. 3-315.</ref> Research summarised in the SFPE Handbook shows that occupants respond faster to clear spoken instructions than to bells or sounders, and fastest to informative live messages.<ref name="Bryan-3-316">Bryan (2002), p. 3-316.</ref><ref name="Proulx-3-350" /> The ''Introduction to Fire Safety Engineering'' likewise states that a slow whoop is less effective than the spoken word.<ref name="VR-7.3" /> Design methods reflect this: in New Zealand, C/VM2 (second edition, 2025) halves the pre-travel activity time for some occupancies when a voice alarm signal is used.<ref name="MBIE25-30">Ministry of Business, Innovation and Employment (2025b), p. 30.</ref> Voice systems also make [[phased evacuation]] possible, because occupants who stay in the building can be told what to do.<ref name="Lew-150">Lew et al. (2005), p. 150.</ref> The sources stress that messages must be audible and intelligible, and that they lose credibility when they conflict with what occupants see.<ref name="Proulx-3-350" /><ref name="Bryan-3-316" />

== How a voice alarm system works ==
A voice alarm system is one part of a [[fire alarm system]], whose basic purpose is to detect fires, notify occupants and summon emergency responders.<ref name="KG-44" /> The fire alerting part consists of audible and visual signals activated by the fire detection system.<ref name="SMC-4-32" /> The audible devices are usually horns, bells, chimes or speakers, and the visual indicators are usually strobe lights.<ref name="SMC-4-32" /> In a voice alarm system the audible devices are loudspeakers that can carry both tones and speech.<ref name="KG-44" /><ref name="ABCB-97" />

The Australian verification method data sheets describe a range of emergency warning systems.<ref name="ABCB-97" /> At the simple end, a single-zone system uses one tone generator and an amplifier to give alert and evacuation tones only.<ref name="ABCB-97" /> More complex systems add public address capability and visual alarms.<ref name="ABCB-97" /> Multi-zoned evacuation systems are used where a building is divided into zones for phased evacuation.<ref name="ABCB-97" /> Each zone is then controlled separately and has its own speakers for tones and public address, and warden intercom points connect each zone with the chief warden.<ref name="ABCB-97" /><ref name="ABCB-98">Australian Building Codes Board (2020a), p. 98.</ref>

The towers of the World Trade Center in New York had such a system. Its fire command station gave manual control of the emergency voice and alarm communication, by individual floor, by a group of floors or throughout the building.<ref name="KG-44" /> The alarm tones were silenced while the loudspeakers were in use and restarted automatically afterwards.<ref name="KG-44" /> Firefighters could make announcements over the same system, and floor wardens had a two-way telephone link to the command station.<ref name="KG-44" /> The earlier public address system had two sound channels, so that simultaneous and independent announcements could be made to selected speaker zones.<ref name="KG-27" />

Phased evacuation systems commonly use two tones.<ref name="ABCB-243">Australian Building Codes Board (2020a), p. 243.</ref> An alert tone tells occupants to prepare for evacuation and to await further instructions from wardens, security staff or speaker announcements.<ref name="ABCB-243" /><ref name="ABCB-244">Australian Building Codes Board (2020a), p. 244.</ref> An evacuation tone means that the building must be left immediately.<ref name="ABCB-244" /> In a two-stage system, a building-wide alert can sound first while the evacuation tone sounds on the fire floor, and the system then cascades through the other floors.<ref name="ABCB-254">Australian Building Codes Board (2020a), p. 254.</ref>

[[File:fse-voice-alarm-system-v1.svg|thumb|upright=1.5|alt=Section of a multi-storey building: the fire floor hears an evacuation message, other floors an alert tone, all controlled from a command station.|A zoned voice alarm system in a phased evacuation: evacuation signal and spoken instructions on the fire floor, alert tone on the other floors, controlled from the fire command station. Based on Australian Building Codes Board (2020a), pp. 97 and 243 to 254, and Keough and Grill (2005), p. 44.]]

== Why spoken messages are more effective ==
How people respond to a fire is a central question of [[human behaviour in fire]]. According to the ''Introduction to Fire Safety Engineering'', the reaction depends on a person's role, experience, culture, training and personality, on the perceived threat and on the actions of others.<ref name="VR-7.3" /> It also depends on the signals perceived, such as the evacuation alarm, smoke and heat.<ref name="VR-7.3" /> The means of alarm largely determines whether people take the signal seriously.<ref name="VR-7.3" />

Plain alarm signals have a weak record. A review of research in the United Kingdom, quoted by Bryan, found that the response to fire alarm bells and sounders tends to be less than optimal.<ref name="Bryan-3-315" /> People are often unsure whether the noise is a fire alarm at all, or merely a test or a drill.<ref name="Bryan-3-315" /> Proulx adds that occupants take longest to respond to the fire alarm signal if they have had no training or if the alarm often sounds for reasons other than fire.<ref name="Proulx-3-350" /> She regards three or more false alarms within six months as a large number.<ref name="Proulx-3-350" />

Spoken messages give occupants information that a tone cannot give. The quality, pitch and volume of a voice, and the content of the message, may reinforce the perceived threat.<ref name="Bryan-3-315" /> In evacuation drills in an underground station, passengers did not move at all after the fire alarm signal.<ref name="Proulx-3-349">Proulx (2002), p. 3-349.</ref> They started to move only 15 s after precise live messages that gave the type and location of the incident and told them what to do.<ref name="Proulx-3-349" /> Proulx concludes that messages through a voice communication system or directly from staff are the signals that occupants take most seriously.<ref name="Proulx-3-349" />

The type of message also matters. Proulx considers informative live messages probably the best way to start a fast response, while prerecorded messages are rarely as effective and contain less information.<ref name="Proulx-3-350" /> The ''Introduction to Fire Safety Engineering'' agrees that live announcements, whether through a public address system, in person or by telephone, are followed faster than automatic alarms.<ref name="VR-7.3" /> A well-trained evacuation organisation in the building makes it many times more likely that an automatic alarm is taken seriously.<ref name="VR-7.3" />

A message must also agree with what people see. If a spoken instruction conflicts with other cues, such as the smell or sight of smoke, its credibility may be questioned and the information ignored.<ref name="Bryan-3-316" /> In a fire in the South Tower of the World Trade Center on 17 April 1975, smoke spread to the 9th to 22nd floors.<ref name="Bryan-3-316" /> At 9:10 a.m. a message advised occupants in the core lobbies to stay calm and return to their offices.<ref name="Bryan-3-316" /> They stayed in the lobbies and grew more concerned, and an evacuation message followed at 9:16 a.m.<ref name="Bryan-3-316" />

== Effect on pre-movement time in design ==
The main design benefit of a voice alarm is a shorter [[pre-movement time]], the time between the warning and the start of purposeful movement to a place of safety.<ref name="VR-7.3" /> In New Zealand, C/VM2 (second edition, 2025) adds this time to the detection time, the notification time and the travel time to give the required safe egress time.<ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref> For standard evacuation strategies the notification time is taken as 30 s.<ref name="MBIE25-29">Ministry of Business, Innovation and Employment (2025b), p. 29.</ref>

C/VM2 gives fixed pre-travel activity times that depend on the occupants and, for some building types, on the alarm signal.<ref name="MBIE25-30" /> For retail shops, exhibition spaces and restaurants, it gives 60 s in the enclosure of origin and 120 s elsewhere with a standard alarm signal.<ref name="MBIE25-30" /> With a voice alarm signal these become 30 s and 60 s.<ref name="MBIE25-30" /> For hotels and motels, the time for occupants remote from the fire falls from 600 s with a standard signal to 300 s with a voice signal.<ref name="MBIE25-30" /> For offices, where occupants are awake and familiar with the building, the same values apply to any alarm signal.<ref name="MBIE25-30" /> C/VM2 lists the type of warning system among the factors that these values are meant to cover.<ref name="MBIE25-31">Ministry of Business, Innovation and Employment (2025b), p. 31.</ref>

[[File:fse-voice-alarm-system-v2.svg|thumb|upright=1.5|alt=Grouped bar chart of pre-travel times for retail and hotel occupants with standard and voice alarm signals; voice bars are half as long except one.|Pre-travel activity times in C/VM2 (second edition, 2025) with a standard alarm signal and with a voice alarm signal. Based on Ministry of Business, Innovation and Employment (2025b), p. 30.]]

Proulx reproduces a similar table from the British Draft for Development DD240 (1997).<ref name="Proulx-3-351">Proulx (2002), p. 3-351.</ref> It sets three levels of warning system.<ref name="Proulx-3-351" /> Level W1 is live directives through a voice communication system from a control room with closed-circuit television, or live directives with well-trained uniformed staff.<ref name="Proulx-3-351" /> Level W2 is non-directive prerecorded voice messages or an informative visual display, with trained staff, and level W3 is a fire alarm signal with untrained staff.<ref name="Proulx-3-351" /> For shops and other assembly buildings, the estimated delay before evacuation starts is less than 2 minutes with W1, 3 minutes with W2 and more than 6 minutes with W3.<ref name="Proulx-3-351" />

As a worked example, consider occupants of a retail shop who are remote from the enclosure of origin, under C/VM2.<ref name="MBIE25-30" /> With a standard alarm signal the pre-travel activity time is 120 s; with a voice alarm signal it is 60 s. If detection, notification and travel times are unchanged, the voice alarm therefore shortens the required safe egress time for these occupants by 60 s. For remote occupants of a hotel the same change saves 300 s, half of the 600 s that C/VM2 assigns with a standard signal.<ref name="MBIE25-30" /> The International Fire Engineering Guidelines (2005) note that there is no validated method for calculating pre-movement times and refer designers to the data in the SFPE Handbook.<ref name="IFEG-2.8-3">National Research Council of Canada et al. (2005), p. 2.8-3.</ref>

[[File:fse-voice-alarm-system-v3.svg|thumb|upright=1.5|alt=Two time bars of detection, notification, pre-travel and travel; the voice alarm bar has a pre-travel segment of 60 s instead of 120 s.|Required safe egress time for remote occupants of a retail shop with a standard alarm and with a voice alarm: only the pre-travel segment changes, from 120 s to 60 s. Based on Ministry of Business, Innovation and Employment (2025b), pp. 26, 29 and 30.]]

== Phased evacuation and zoned messages ==
Voice communication systems were developed in response to phased evacuation.<ref name="Lew-150" /> In a phased evacuation, some occupants are relocated within the building or held where they are, and they need instructions until they are told to leave.<ref name="Lew-150" /> Requirements for voice systems first appeared in US national standards in the mid-1980s, for example in the 1985 edition of NFPA 72F.<ref name="Lew-150" /> In the same period New York City adopted Local Law 16 of 1984.<ref name="Lew-150" />

The International Fire Engineering Guidelines describe the approach for tall buildings.<ref name="IFEG-2.8-2">National Research Council of Canada et al. (2005), p. 2.8-2.</ref> In high-rise buildings with suitable emergency warning and intercommunication systems, trained personnel may manage the evacuation.<ref name="IFEG-2.8-2" /> Occupants on floors furthest from the fire are first placed on alert and evacuated progressively, only if the fire continues to develop.<ref name="IFEG-2.8-2" /> The SFPE Handbook adds that stay-in-place and partial evacuation strategies depend on giving occupants information about the location and extent of the fire.<ref name="SMC-4-2">Schifiliti et al. (2002), p. 4-2.</ref>

Zoned messages need careful timing in the analysis. Where phased evacuation applies, the alarm reaches some areas later than others, and these delays must be included in the notification time.<ref name="ABCB-243" /> Delays for alarm investigation and verification must also be included, because they can have a significant effect on the required safe egress time.<ref name="ABCB-243" />

== Audibility and intelligibility ==
A voice message is useless if it is not heard and understood.<ref name="SMC-4-32" /> The SFPE Handbook gives sound levels that alarm signals need to reach. A level of 55 to 70 dBA is reported to wake a college-age person with normal hearing.<ref name="SMC-4-32" /> In the US, NFPA 72 requires signals 15 dBA above ambient in areas where people may sleep.<ref name="SMC-4-32" /> British standards require 65 dBA, or 5 dBA above ambient noise, where people are awake, and 75 dBA at the bedhead where people sleep.<ref name="SMC-4-32" />

To estimate these levels, the handbook starts from the sound power level of a sounder. This is the acoustic power that the device radiates, expressed in decibels. It follows from the output that the manufacturer states at a given distance, using a relation of Butler, Bowyer and Kew:<ref name="SMC-4-34">Schifiliti et al. (2002), p. 4-34.</ref>

<math display="block">L_W = L + 20\,\log_{10} r + 11~\mathrm{dB}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>L_W</math> || sound power level of the horn, bell, speaker or other sounder, referenced to 10−12 W || dB
|-
| <math>L</math> || output stated by the manufacturer at distance <math>r</math> || dBA
|-
| <math>r</math> || distance at which <math>L</math> is stated || m
|}

The handbook uses this relation for sounders in corridors, before corrections for the number of directions in which sound spreads, the wall and floor finishes, the distance and the sound insulation of walls and doors.<ref name="SMC-4-34" /> As a worked example, a typical compression-driver fire alarm speaker powered at 2 W gives 94 dBA at 3.05 m.<ref name="SMC-4-34" /> Its sound power level is then 94 + 20 × log10(3.05) + 11 ≈ 115 dB.<ref name="SMC-4-34" /> Doubling the power to 4 W raises the sound power level by only 3 dB, because decibels are logarithmic.<ref name="SMC-4-36">Schifiliti et al. (2002), p. 4-36.</ref>

Louder is not always better for speech. The handbook warns that, if the fire alarm system is also used for voice communication, a speaker tapped at 4 W in a small corridor might sound distorted and be unintelligible.<ref name="SMC-4-36" /> Proulx therefore advises assessing both audibility and intelligibility, so that occupants actually perceive the information.<ref name="Proulx-3-350" /> At the World Trade Center, the Port Authority analysed speaker intelligibility, meaning the clarity of the message, and audibility, meaning whether it is loud enough to be heard.<ref name="KG-49">Keough and Grill (2005), p. 49.</ref> The aim was a guideline for speaker locations that would ensure acceptable levels of both.<ref name="KG-49" /> Speakers were to be placed so that they would be heard clearly above the ambient noise.<ref name="KG-45">Keough and Grill (2005), p. 45.</ref>

== Reliability and lessons from the World Trade Center ==
The Australian data sheets list a default reliability of 0.9 for voice alarms, taken from the work of Zhang and colleagues, the same value as for central alarms.<ref name="ABCB-117">Australian Building Codes Board (2020a), p. 117.</ref> A local alarm has a value of 0.75 in that list.<ref name="ABCB-117" /> In the World Trade Center towers, the loudspeaker system was required to remain 50% operable throughout the building if the other 50% failed.<ref name="KG-45" />

On 11 September 2001, the fire alarm systems of the towers worked, but not all functions performed as intended.<ref name="KG-103">Keough and Grill (2005), p. 103.</ref> The alarm signal had to be activated manually, and in WTC 1 it was delayed by 12 minutes after the aircraft impact.<ref name="KG-106">Keough and Grill (2005), p. 106.</ref> There was no confirmation that the emergency voice communication system worked on any floor of WTC 1 after the impact, although alarm tones were heard below the impact floors.<ref name="KG-104">Keough and Grill (2005), p. 104.</ref> In WTC 2, voice messages and alarm tones were heard above the impact floors.<ref name="KG-104" /> Even with multiple communication risers, the systems degraded, and in WTC 1 all notification and communication functions above the impact floors appear to have been lost.<ref name="KG-106" />

The content of the messages mattered as much as the hardware. About three minutes before WTC 2 was hit, a building-wide announcement told its occupants that the fire was in WTC 1, that WTC 2 was secure and that they should return to their offices.<ref name="Averill-142">Averill et al. (2005), p. 142.</ref> Fifty-four percent of the occupants of WTC 2 reported receiving information from the public address system.<ref name="Averill-143">Averill et al. (2005), p. 143.</ref> Some people followed the instruction and returned to their offices, while many others continued to evacuate.<ref name="Averill-143" /> A second announcement at 9:02 a.m., one minute before the impact, said that occupants could start an orderly evacuation if conditions on their floor warranted it.<ref name="Averill-143" />

== See also ==
* [[Manual call point]]
* [[Unwanted fire alarms]]
* [[Multi-sensor detector]]
* [[Detector activation time]]
* [[Smoke detector]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Keough, R. J., & Grill, R. A. (2005). ''Fire alarm systems'' (NIST NCSTAR 1-4C). National Institute of Standards and Technology.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Detection & Alarm]]

Pre-movement time

2026-11-02T17:14:22Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''pre-movement time''' is the time between the moment occupants receive a first relevant warning or cue of a fire and the moment they begin purposeful movement towards a place of safety.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> It is part of the [[required safe egress time]], which can be divided roughly into the pre-movement time and the time needed to move.<ref name="VR-7.2">Van Rossum (2027), §7.2.</ref> During the pre-movement time, occupants recognise that something is wrong, decide what to do and carry out other actions before they set off, such as investigating, gathering belongings or finding family members.<ref name="ABCB-245">Australian Building Codes Board (2020a), Data Sheet D1, p. 245.</ref>

The pre-movement time is often the largest part of the evacuation time. In unannounced drills in Canadian apartment buildings with an audible alarm, occupants took on average 2 min 49 s to start, and three-quarters of the evacuation time was spent before moving.<ref name="Proulx-348">Proulx (2002), p. 3-348.</ref> It is also the part that is most strongly governed by [[human behaviour in fire]], one of the least predictable aspects of fire safety.<ref name="VR-7.3" /> Design guides therefore give default values that depend on the type of occupants, the alarm system and the management of the building. Such values range from 30 s for alert office workers in the room of fire origin to 600 s for sleeping hotel guests remote from the fire.<ref name="MBIE25-30">Ministry of Business, Innovation and Employment (2025b), p. 30.</ref>

== Definition and terminology ==
Several names are used for the same period. The ''Introduction to Fire Safety Engineering'' speaks of the pre-evacuation time.<ref name="VR-7.3" /> C/VM2 (second edition, 2025) uses "pre-travel activity time", the period after an alarm or fire cue is transmitted and before occupants first travel towards an exit.<ref name="MBIE25-68">Ministry of Business, Innovation and Employment (2025b), p. 68.</ref> The Australian data sheets for the Fire Safety Verification Method use both "pre-travel time" and "pre-movement time".<ref name="ABCB-244">Australian Building Codes Board (2020a), Data Sheet D1, p. 244.</ref><ref name="ABCB-294">Australian Building Codes Board (2020a), Data Sheet D1, p. 294.</ref> NIST, in its study of the World Trade Center evacuation, measured the "evacuation initiation delay".<ref name="Averill-131">Averill et al. (2005), p. 131.</ref>

In the SFPE Handbook, Proulx prefers the term "delay time to start". She notes that "premovement" is used particularly in the British literature.<ref name="Proulx-348" /> Her objection is that occupants often move around a great deal during this period, trying to find out what is happening.<ref name="Proulx-348" /> Nelson and Mowrer split the same period into two terms of their sum for the required safe egress time. The first runs from notification until occupants decide to act, the second from that decision until evacuation starts.<ref name="NM-367">Nelson and Mowrer (2002), p. 3-367.</ref>

The start of the period also varies between sources. In C/VM2 the pre-travel activity time begins at notification, after the detection time and a notification time of 30 s for standard evacuation strategies.<ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref><ref name="MBIE25-29">Ministry of Business, Innovation and Employment (2025b), p. 29.</ref> The International Fire Engineering Guidelines (2005) count from the occurrence of a cue, which can also be a fire cue or a warning by other people.<ref name="IFEG-1.8-1">National Research Council of Canada et al. (2005), p. 1.8-1.</ref> When values from different sources are combined, it must therefore be clear where each period begins.

== Components ==
Most sources divide the pre-movement time into a recognition part and a response part. The Australian data sheets define the recognition time as the period after an alarm or cue is evident but before an occupant begins to respond.<ref name="ABCB-245" /> During this time, people continue their normal activities, such as working, shopping or sleeping. The recognition time ends when they accept that they need to respond.<ref name="ABCB-245" /> The response time follows. Occupants stop their normal activities and do other things before they move towards an exit or a safer place.<ref name="ABCB-245" />

[[File:fse-pre-movement-time-v1.svg|thumb|upright=1.5|alt=Timeline from ignition: detection, alarm or cue, recognition, response and movement; a bracket marks recognition plus response as the pre-movement time.|The pre-movement time runs from the alarm or cue to the start of movement and consists of a recognition time and a response time. Based on Australian Building Codes Board (2020a), Data Sheet D1, pp. 244–245, and National Research Council of Canada et al. (2005), pp. 1.8-2 and 1.8-4.]]

The International Fire Engineering Guidelines use the same split under other names. The pre-movement phase is the sum of a response period, from the occurrence of a cue to its recognition, and a delay period, from recognition to the start of movement.<ref name="IFEG-1.8-2">National Research Council of Canada et al. (2005), p. 1.8-2.</ref><ref name="IFEG-1.8-4">National Research Council of Canada et al. (2005), p. 1.8-4.</ref> In the delay period, occupants may carry out a wide variety of actions, including no action at all.<ref name="IFEG-1.8-4" /> Proulx describes three subcomponents: perception of the alarm or cue, interpretation of the information, and action.<ref name="Proulx-348" /> The ''Introduction to Fire Safety Engineering'' uses three phases as well: awareness, decision-making and execution.<ref name="VR-7.3" />

The actions in the response part are varied. They include investigating the cue, shutting down equipment, finding friends and family, fighting the fire, alerting others, deciding whether to leave, and waiting for instructions.<ref name="ABCB-245" /> In residential drills, occupants got dressed and gathered children, pets, purses, wallets and keys before leaving.<ref name="Proulx-349">Proulx (2002), p. 3-349.</ref> Office workers finished phone calls, saved data, secured files and collected belongings, and many had to be prompted by their fire warden.<ref name="Proulx-349" /> In the World Trade Center towers in 2001, such pre-evacuation actions were the strongest predictor of how long people delayed their evacuation.<ref name="Averill-134">Averill et al. (2005), p. 134.</ref>

Interpreting ambiguous cues often takes the form of seeking information. NIST describes this as a typical human response in sudden emergencies, because people need to make sense of an event before they act on it.<ref name="Averill-133">Averill et al. (2005), p. 133.</ref> Its causal model for the towers linked environmental cues, the floor of the occupant, perceived risk and information seeking to the delay.<ref name="Averill-131" /> Such step-by-step descriptions of how people perceive cues, judge the risk and decide what to do are treated further under [[protective action decision model]].

== Factors that influence pre-movement time ==
The ''Introduction to Fire Safety Engineering'' lists many factors that shape the response to a fire. They include the person's assumed role, earlier experience, cultural background, training, personality and physical characteristics.<ref name="VR-7.3" /> Also important are the perceived threat, the available escape routes, the actions of others and the signals that are received.<ref name="VR-7.3" /> Proulx groups the factors into building characteristics and occupant characteristics.<ref name="Proulx-350">Proulx (2002), p. 3-350.</ref> Building characteristics include the warning system, the layout and wayfinding, visual access to other people, the presence of a focal point such as a stage, the level of training and the frequency of false alarms.<ref name="Proulx-350" /> In her experience, three or more false alarms in 6 months already count as a large number.<ref name="Proulx-350" />

The type of warning has a strong effect. An [[evacuation alarm system]] with a plain alarm signal is, according to Proulx, the cue that occupants take longest to respond to when they are untrained or used to false alarms.<ref name="Proulx-350" /> A [[voice alarm system]] with informative live messages is probably the best way to initiate a fast response, and prerecorded messages are rarely as effective.<ref name="Proulx-350" /> Bryan adds that a directive message loses credibility when it conflicts with what occupants see or smell.<ref name="Bryan-316">Bryan (2002), p. 3-316.</ref>

The ''Introduction to Fire Safety Engineering'' likewise states that a slow-whoop tone is less effective than the spoken word, and that live messages are followed faster than automatic alarms.<ref name="VR-7.3" /> In an underground station, passengers never started to evacuate on the fire alarm alone. When staff or precise live voice messages prompted them, they began to move about 15 s after the message.<ref name="Proulx-349" />

Occupant characteristics include familiarity with the building, a sense of responsibility, social affiliation and commitment to an activity.<ref name="Proulx-350" /><ref name="Proulx-351">Proulx (2002), p. 3-351.</ref> Families tend to gather before they leave, and people who are eating, gambling or queuing for a flight are slow to turn their attention to an alarm.<ref name="Proulx-350" /><ref name="Proulx-351" /> Sleeping occupants, and people with perceptual, physical or intellectual limitations or under the influence of drugs or alcohol, need longer.<ref name="Proulx-351" /> The Australian data sheets note that even blood alcohol levels below 0.05 can leave a significant share of residents unwoken by smoke alarms.<ref name="ABCB-245" />

Other people also matter. In experiments by Latané and Darley, students alone in a room reported smoke entering the room in 75 percent of cases. When two passive confederates sat with them, only 10 percent of the groups reported it.<ref name="Bryan-317">Latané and Darley (1968), as cited in Bryan (2002), p. 3-317.</ref> The ''Introduction to Fire Safety Engineering'' describes the same tendency to conform to the group and to follow authority. People may wait to leave until a manager has given permission.<ref name="VR-7.3" /> Group behaviour can also help: when enough people make the right decision, as in a well-practised evacuation organisation, the others tend to follow.<ref name="VR-7.3" />

[[Panic]] is not a major cause of delay. Proulx states that panic is very rare even in fires, and Bryan describes deliberate, purposeful behaviour as the most frequent response.<ref name="Proulx-344">Proulx (2002), p. 3-344.</ref><ref name="Bryan-315">Bryan (2002), p. 3-315.</ref> The ''Introduction to Fire Safety Engineering'' notes that a lack of awareness of the danger is usually the larger problem, because people find it hard to judge how fast a fire grows.<ref name="VR-7.3" />

== Measured values ==
Measured pre-movement times vary widely. In seven residential buildings in Canada, Proulx recorded the start times of more than 500 occupants during drills.<ref name="Proulx-348" /> In the three buildings with a clearly audible alarm, the mean delay was 2 min 49 s. In a winter drill, with coats and boots to put on, it was 5 min 19 s.<ref name="Proulx-348" /> In two buildings where more than 20 percent of residents found the alarm too quiet, some started only after 25 min, and the mean was 9 min.<ref name="Proulx-349" />

Office workers were much quicker. Video records of more than 1,000 occupants in three Canadian government office buildings gave a mean delay of 50 s.<ref name="Proulx-349" /> In an unannounced drill in a large retail store, customers started to move on average 25 s after the alarm, with a maximum of 55 s, mainly because staff responded fast.<ref name="Proulx-349" /> Real fires can be slower. Interviews after a high-rise office fire without a working alarm gave a mean of about 2 min 30 s. After a night-time residential fire, occupants took about 10 min to start.<ref name="Proulx-349" />

In the World Trade Center on 11 September 2001, the median time to initiate evacuation in the first tower was 3 minutes below floor 77. Near the impact region, on floors 77 to 91, it was 5 minutes.<ref name="Averill-xxxiii">Averill et al. (2005), p. xxxiii.</ref> The floor, environmental cues and seeking additional information best predicted the delay in that tower.<ref name="Averill-xxxiv">Averill et al. (2005), p. xxxiv.</ref>

[[File:fse-pre-movement-time-v2.svg|thumb|upright=1.5|alt=Bar chart of measured delays: retail 25 s, offices 50 s, apartments 169 s, office fire 150 s, winter drill 319 s, quiet alarm 540 s, night fire about 600 s.|Mean or median times to start evacuation from drills and fires: from under a minute in offices and shops to about 10 minutes at night. Based on Proulx (2002), pp. 3-348 to 3-349, and Averill et al. (2005), p. xxxiii.]]

== Design values ==
Because there is no validated method to calculate pre-movement times, design guides provide default values. The International Fire Engineering Guidelines (2005) recommend estimates based on published studies, and mention a probabilistic method that fits a Weibull distribution to observed times.<ref name="IFEG-2.8-3">National Research Council of Canada et al. (2005), p. 2.8-3.</ref> The commentary to C/VM2 notes that there are no widely accepted values. It names the British document PD 7974-6 as the most complete collection.<ref name="MBIE13-91">Ministry of Business, Innovation and Employment (2013), p. 91.</ref>

In New Zealand, C/VM2 (second edition, 2025) prescribes pre-travel activity times by occupant type, alarm type and location.<ref name="MBIE25-30" /> Occupants in the enclosure of fire origin get shorter times, because direct cues such as flames and smoke, or indirect cues from others, prompt a faster response.<ref name="MBIE25-31">Ministry of Business, Innovation and Employment (2025b), p. 31.</ref> Indirect cues may not be assumed in apartments, hotels or separate office tenancies.<ref name="MBIE25-31" /> The table lists a selection of the values.<ref name="MBIE25-30" />

[[File:fse-pre-movement-time-v3.svg|thumb|upright=1.5|alt=Grouped bar chart of C/VM2 pre-travel activity times in and remote from the enclosure of origin for offices, shops, apartments and hotels.|Pre-travel activity times in C/VM2 (second edition, 2025) for occupants in and remote from the enclosure of fire origin. Based on Ministry of Business, Innovation and Employment (2025b), p. 30.]]

{| class="wikitable"
! Occupants (example) !! Alarm signal !! In enclosure of origin (s) !! Remote from enclosure of origin (s)
|-
| Awake, alert and familiar (offices) || Any || 30 || 60
|-
| Awake, alert and unfamiliar (retail shops, restaurants) || Standard || 60 || 120
|-
| Awake, alert and unfamiliar (retail shops, restaurants) || Voice || 30 || 60
|-
| Sleeping and familiar (apartments) || Standard || 60 || 300
|-
| Sleeping and unfamiliar (hotels and motels) || Standard || 60 || 600
|-
| Sleeping and unfamiliar (hotels and motels) || Voice || 60 || 300
|-
| Awake, under care of trained staff (dental office, clinic) || Any || 60 || 120
|}

The commentary explains that these values were chosen from the most optimistic management options. New Zealand rules require evacuation schemes in most public non-residential buildings, which leads to a culture of prompt evacuation.<ref name="MBIE13-93">Ministry of Business, Innovation and Employment (2013), p. 93.</ref> C/VM2 also notes that the incipient phase of a fire is left out of its design fire, which gives an inherent safety factor on these times.<ref name="MBIE25-31" />

The Australian data sheets give distributions rather than single values. For staff in a simple workplace with a well-trained emergency organisation and an automatic alarm system, 10 percent start after 30 s, 80 percent after 60 s and 10 percent after 90 s.<ref name="ABCB-270">Australian Building Codes Board (2020a), Data Sheet D1, p. 270.</ref> With a minimal emergency organisation, much longer times are expected. A check of the design against failure of that organisation may assume that 10 percent do not respond and that the others take 15 to 30 minutes.<ref name="ABCB-270" /> The data sheets warn that a split in which 80 percent of occupants start at the same moment may be too crude.<ref name="ABCB-269">Australian Building Codes Board (2020a), Data Sheet D1, p. 269.</ref>

[[File:fse-pre-movement-time-v4.svg|thumb|upright=1.5|alt=Step chart of the share of staff that has started to move against time: with a trained organisation all start by 90 s, with a weaker one by 180 s.|Suggested default distributions of pre-movement time for staff in simple workplaces with two levels of emergency management. Based on Australian Building Codes Board (2020a), Data Sheet D1, p. 270.]]

Proulx reproduces values from the British draft DD240 that depend on the warning system. For offices and schools, the estimated delay is under 1 min with live directives, 3 min with prerecorded messages and trained staff, and more than 4 min with a plain alarm and untrained staff.<ref name="Proulx-351" /> For hotels the three values are under 2, 4 and more than 6 min, and for hospitals under 3, 5 and more than 8 min.<ref name="Proulx-351" /> For one and the same office building, the choice of warning system and staff training thus shifts the estimate by several minutes.<ref name="Proulx-351" />

== Use in fire safety design ==
In an analysis that compares the required with the available safe egress time, the pre-movement time is added after the detection and notification times.<ref name="MBIE25-26" /> Older calculations ignored it. Proulx notes that this neglect usually explains why calculated evacuation times were shorter than those observed in drills.<ref name="Proulx-351" /> Earlier editions of the SFPE Handbook chapter suggested doubling the calculated movement time. Many engineers found this too demanding and added only 15 to 30 s instead.<ref name="Proulx-348" />

Because occupants do not all start at once, the Australian data sheets recommend representing the population by a distribution or by several design occupant profiles.<ref name="ABCB-244" /> NIST simulated a phased evacuation of three floors in a World Trade Center tower. Random start delays of 0 to 10 minutes raised the time to clear the floors from about 4 to about 11 minutes.<ref name="Averill-160">Averill et al. (2005), p. 160.</ref> The pre-movement time can also be shortened by design and management. The International Fire Engineering Guidelines list additional and less ambiguous cues, more trained personnel, training programmes and better information as measures.<ref name="IFEG-1.8-13">National Research Council of Canada et al. (2005), p. 1.8-13.</ref>

== Limits and common misconceptions ==
A common assumption is that people start to evacuate as soon as they hear an alarm. Field studies show instead that people tend to ignore or play down initial cues, investigate and only then decide.<ref name="Proulx-344" /> Another assumption is that real fires produce faster responses than drills. Interviews with fire victims show the opposite: the first signs of a fire are usually ambiguous, which tends to lengthen the delay.<ref name="Proulx-351" />

Default values must be used with care. The Australian data sheets state that their suggested times must be checked against literature, drills and incident reports, and adjusted to the building.<ref name="ABCB-269" /> The International Fire Engineering Guidelines describe only "non-validated quasi-methodologies" for this phase.<ref name="IFEG-2.8-3" /> Any value based on engineering judgement must be justified in detail.<ref name="IFEG-1.8-4" /> For factories, time to shut down equipment may have to be added to the published values.<ref name="Proulx-352">Proulx (2002), p. 3-352.</ref>

== See also ==
* [[Unwanted fire alarms]]
* [[Evacuation]]
* [[Affiliative behaviour]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nelson, H. E., & Mowrer, F. W. (2002). Emergency movement. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-367–3-380). National Fire Protection Association; Society of Fire Protection Engineers.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Prescriptive design

2026-11-02T17:14:20Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Prescriptive design''' is fire safety design that follows explicit rules in regulations, standards or codes, which state the measures to be provided and their quantities instead of the outcome to be achieved. The International Fire Engineering Guidelines (IFEG, 2005) define prescriptive provisions as provisions that are expressed explicitly in quantitative form.<ref name="IFEG-054">National Research Council of Canada et al. (2005), p. 0.5-4 (Part 0, Australia).</ref> Typical rules limit the travel distance to an exit in metres, the fire resistance of separations in minutes and the floor area of a fire compartment in square metres.<ref name="VR-1.2">Van Rossum (2027), §1.2.</ref>

A prescriptive design is simple to apply and to check, because compliance can be confirmed by comparing the drawings with the rules.<ref name="VR-2.1">Van Rossum (2027), §2.1.</ref> In Australia such a design is called a deemed-to-satisfy (DTS) Solution and is described as following a set recipe.<ref name="HB-13">Australian Building Codes Board (2020b), p. 13.</ref> The rules are generic, however. They are only partly tailored to the building, its people and its surroundings, and they are often based on consensus rather than on a quantified goal.<ref name="VR-1">Van Rossum (2027), §1.</ref>

[[File:fse-prescriptive-design-v1.svg|thumb|upright=1.5|alt=Two panels: left, drawings checked against a rule book with a tick; right, objectives, analysis and criteria in a loop.|Prescriptive design (left) is checked by comparing the drawings with the rules; performance-based design (right) by analysing a trial design against agreed criteria. Based on Australian Building Codes Board (2020b), pp. 12–13 and Van Rossum (2027), §1 and §2.1.]]

Prescriptive codes also cannot cover emerging technologies or every combination of circumstances.<ref name="IFEG-021">National Research Council of Canada et al. (2005), p. 0.2-1 (Part 0, Australia).</ref> Performance-based codes such as the Australian National Construction Code therefore allow departures from the rules. The design must then be shown to meet the performance requirements or to be at least equivalent to the DTS provisions.<ref name="HB-15">Australian Building Codes Board (2020b), p. 15.</ref>

== Definition and terminology ==
Prescriptive rules are found in many forms, and the names differ between countries. In the performance-based code of Australia, the prescriptive part consists of the DTS provisions. The IFEG define these as the prescribed provisions of a code that are deemed to satisfy its performance requirements.<ref name="IFEG-052">National Research Council of Canada et al. (2005), p. 0.5-2 (Part 0, Australia).</ref> A DTS Solution uses materials, components, design factors and construction methods that are deemed to meet the performance requirements if they are used.<ref name="HB-13" /> In New Zealand, simple buildings are covered by the Acceptable Solutions C/AS1 to C/AS7, while Verification Method C/VM2 applies to complex buildings.<ref name="MBIE13-11">Ministry of Business, Innovation and Employment (2013), p. 11.</ref>

In the United States, the IFEG describe the codes in primary use as prescriptive codes, although they contain some elements of performance.<ref name="IFEG-US024">National Research Council of Canada et al. (2005), p. 0.2-4 (Part 0, United States).</ref> Model codes such as the International Building Code (IBC) are written by not-for-profit organisations and adopted by state and local governments as their [[building regulations]].<ref name="IFEG-US024" /> At the time of the IFEG, most US jurisdictions used a prescriptive building code, and the IBC was the predominant one.<ref name="IFEG-US022">National Research Council of Canada et al. (2005), p. 0.2-2 (Part 0, United States).</ref>

''Introduction to Fire Safety Engineering'' uses the term rule-based for this approach. Rule-based means that predetermined frameworks and limit values are met in order to achieve the fire safety goals.<ref name="VR-1" /> The book gives three examples of such rules: legislation such as the Dutch building regulations (Besluit bouwwerken leefomgeving, Bbl), the conditions of an insurer, and requirements that a large company imposes on all its sites.<ref name="VR-1" /> Its opposite is the risk-based approach, which looks at fire safety project by project and with more variables.<ref name="VR-1" />

== What prescriptive rules regulate ==
Prescriptive rules set measurable limits on the means of fire safety. ''Introduction to Fire Safety Engineering'' names metres of [[travel distance]], minutes of fire resistance against fire spread, and square metres of fire compartment as typical examples.<ref name="VR-1.2" /> Escape provisions are a clear case. The Australian handbook for the Fire Safety Verification Method (FSVM) notes that it is not practical to provide multiple escape paths from every point in a building.<ref name="HB-66">Australian Building Codes Board (2020b), p. 66.</ref> The DTS provisions therefore prescribe a maximum distance of travel to an exit, or to a point from which travel in two directions to two exits is possible.<ref name="HB-66" /> They also set minimum and maximum distances between exits and a minimum separation between paths of travel.<ref name="HB-66" />

[[File:fse-prescriptive-design-v2.svg|thumb|upright=1.5|alt=Floor plan with two exits, a dashed walking route to the nearest exit, a dead-end corridor and a compartment boundary, each with a dimension arrow.|Typical quantities controlled by prescriptive escape and compartment rules. Based on Australian Building Codes Board (2020b), p. 66 and Van Rossum (2027), §1.2.]]

The numbers can be small and specific. In one handbook example, the DTS provisions limit the travel distance from an apartment door to a fire-isolated stair to 6 m.<ref name="HB-46">Australian Building Codes Board (2020b), p. 46.</ref> In New Zealand, a single escape route has long been permitted to serve up to 50 people. The commentary to C/VM2 calls this limit largely historical and consistent with the earlier Acceptable Solution C/AS1 (2011).<ref name="MBIE13-123">Ministry of Business, Innovation and Employment (2013), p. 123.</ref>

The World Trade Center towers in New York show how a building was designed to such rules. Each office floor had three stairs, two 44 inches (1.12 m) wide and one 56 inches (1.42 m) wide.<ref name="LE-48">Lew et al. (2005), p. xlviii.</ref> Together they provided exactly the capacity required by the building code for 390 occupants per floor, based on 100 ft² (9.29 m²) of floor area per person.<ref name="LE-48" /> The layout met the code limits on travel distance of 200 ft (61 m) without sprinklers and 300 ft (91 m) with sprinklers.<ref name="LE-48" /> The code also traded one measure against another: dead ends could extend to 100 ft (30 m) instead of 50 ft (15 m) where corridor partitions were 2-hour rather than 1-hour fire rated.<ref name="LE-48" />

[[File:fse-prescriptive-design-v3.svg|thumb|upright=1.5|alt=Bar chart: travel distance 61 m without and 91 m with sprinklers; dead end 15 m with 1-hour and 30 m with 2-hour partitions.|Code limits that applied to the World Trade Center towers: the maximum travel distance depended on sprinklers and the maximum dead end on the fire rating of corridor partitions. Based on Lew et al. (2005), p. xlviii.]]

[[Fire compartmentation]] is often regulated through limits on compartment size. In the Netherlands, models for large fire compartments are often used to show that a large compartment performs as well as a compartment of, for example, 1,000 m².<ref name="VR-8.2">Van Rossum (2027), §8.2.</ref> In New Zealand, the previous Acceptable Solution C/AS1 (2011) set firecell floor area limits using a limiting fire load of 2 million MJ.<ref name="MBIE13-147">Ministry of Business, Innovation and Employment (2013), p. 147.</ref> Other rules prescribe the construction itself. For example, the Building Code of Australia of 2005 required a stair to be located in a fire-isolated shaft that discharges directly to the outside.<ref name="IFEG-1212">National Research Council of Canada et al. (2005), p. 1.2-12.</ref> In the United States, IBC section 803.5 limits the flame spread index of interior wall and ceiling finishes according to the occupancy group and location.<ref name="IFEG-US0210">National Research Council of Canada et al. (2005), p. 0.2-10 (Part 0, United States).</ref>

== Origins of prescriptive rules ==
Prescriptive rules grew out of experience. Watts and Hall describe narratives, series of recommendations on what to do and not do, as probably the earliest approach to fire risk.<ref name="WH-5">Watts and Hall (2002), p. 5-5.</ref> Over the years this narrative approach developed into the multi-volume set of NFPA National Fire Codes, which contain the bulk of present-day knowledge on fire safety.<ref name="WH-5" /> Checklists, which list hazards and recommended practices, are a related tool.<ref name="WH-4">Watts and Hall (2002), p. 5-4.</ref>

''Introduction to Fire Safety Engineering'' notes that rules sometimes originate in a risk assessment but are often the result of consensus.<ref name="VR-1" /> The book explains that Dutch regulations arose partly from consensus and partly from history, and that no quantified goal was pursued when they were written.<ref name="VR-1.2" /> The commentary to New Zealand's C/VM2 makes a similar observation: international codes contain a number of historic values that are commonly accepted but have no technical basis.<ref name="MBIE13-12">Ministry of Business, Innovation and Employment (2013), p. 12.</ref>

Prescriptive provisions are nevertheless refined over time. In Australia, the DTS provisions have been continuously improved since the first Building Code of Australia of 1988.<ref name="HB-14">Australian Building Codes Board (2020b), p. 14.</ref> Most technical changes went through extensive public consultation, often supported by detailed fire safety analyses and cost-benefit analyses.<ref name="HB-14" /> The 1988 code stated that its requirements should extend no further than necessary in the public interest. They should also be cost-effective, not needlessly onerous and easily understood.<ref name="HB-14" />

== Advantages ==
The main advantage of prescriptive design is simplicity. The Australian handbook describes the DTS route as generally having a simpler assessment process than a Performance Solution.<ref name="HB-12">Australian Building Codes Board (2020b), p. 12.</ref> ''Introduction to Fire Safety Engineering'' points out that in a rule-based approach, the approving authority can check the relevant aspects much more easily than in a fire safety engineering approach.<ref name="VR-2.1" /> A fire engineering justification, by contrast, can be so complex that the authority finds it hard or impossible to verify.<ref name="VR-2.1" />

Prescriptive rules also give a common benchmark. The Australian FSVM assumes that, in the absence of quantified performance requirements, acceptance criteria reflecting community expectations can be derived by comparison with the current DTS provisions.<ref name="HB-15" /> The DTS provisions thus define the level of safety that a performance-based design must at least match.<ref name="HB-15" />

Generic rules are also less sensitive to change during the life of a building. The IFEG note that a fire engineered design is specific to its building, occupants and site, so later changes may require a new evaluation.<ref name="IFEG-037">National Research Council of Canada et al. (2005), p. 0.3-7 (Part 0, Australia).</ref> Such a re-evaluation might not be needed if the broader DTS provisions had been adopted.<ref name="IFEG-037" />

== Limitations ==
The main limitation is that prescriptive rules regulate means rather than outcomes. The IFEG explain that a prescriptive code simply prescribes the required performance, without a method to determine or quantify the level of performance of a building.<ref name="IFEG-US026">National Research Council of Canada et al. (2005), p. 0.2-6 (Part 0, United States).</ref> Requirements on heights and areas, sprinklers and structure address the hazards and tolerable losses only indirectly, so the level of safety they provide is hard to measure.<ref name="IFEG-US026" /> For the same reason it is difficult to say what "equivalent" means when an alternative is proposed.<ref name="IFEG-US026" />

Because the rules are generic, they fit some buildings poorly. ''Introduction to Fire Safety Engineering'' describes them as very generic and inflexible measures that are only partly tailored to the building, the people and the environment.<ref name="VR-1" /> The IFEG add that a prescriptive code has only one value that is deemed reasonable. Communities must amend the code to reflect local needs, and such amendments are hard to justify when the code offers a single solution.<ref name="IFEG-US027">National Research Council of Canada et al. (2005), p. 0.2-7 (Part 0, United States).</ref> Prescriptive codes also cannot cover emerging technologies and every combination of circumstances.<ref name="IFEG-021" />

The resulting level of safety can be inconsistent. ''Introduction to Fire Safety Engineering'' notes large differences between occupancy types in actual performance on casualties and fire spread.<ref name="VR-1.2" /> Per hour of stay, the chance of dying in a fire is much greater in a residential building than in an assembly building, yet the Dutch detection requirements are stricter for assembly buildings.<ref name="VR-1.2" /> The rule-based limits also cannot be translated unambiguously into risk-based limits.<ref name="VR-1.4">Van Rossum (2027), §1.4.</ref>

Watts and Hall point to a further weakness of checklists and narratives. A risk is judged acceptable if it is addressed in line with the published recommendations, so the test is pass or fail and the residual risk is never quantified.<ref name="WH-5" /> Checklists also do not capture the interaction between fire safety factors. The relative value of hydrants, sprinklers and extinguishers, for example, depends on the form and use of the building.<ref name="WH-5" />

== Relationship with performance-based design ==
Most building codes combine prescriptive rules with a way to depart from them. In the United States, section 104.11 of the IBC states that the code does not prohibit a design that it does not specifically prescribe, provided the alternative is approved.<ref name="IFEG-US025">National Research Council of Canada et al. (2005), p. 0.2-5 (Part 0, United States).</ref> Approval requires that the alternative is at least equivalent in quality, strength, effectiveness, fire resistance, durability and safety.<ref name="IFEG-US025" /> In the Netherlands, article 4.7 of the Environment and Planning Act (Omgevingswet) allows a departure from the prescribed measures of the Bbl on the basis of [[equivalency]]. The alternative must reach at least the same level of fire safety as the original measure.<ref name="VR-1" />

[[File:fse-prescriptive-design-v4.svg|thumb|upright=1.5|alt=Flowchart: design meets all prescriptive provisions, yes leads to deemed to comply; no leads to analysis of the departures only.|A design that meets all prescriptive provisions is deemed to comply; departures are analysed for equivalence or against the performance requirements. Based on National Research Council of Canada et al. (2005), pp. 0.2-5 (Part 0, United States) and 1.2-12, and Australian Building Codes Board (2020b), p. 16.]]

In Australia, a design can use a DTS Solution, a Performance Solution, or a combination of both.<ref name="HB-12" /> The FSVM requires a Performance Solution to be compared with a reference building that complies fully with the DTS provisions.<ref name="HB-16">Australian Building Codes Board (2020b), p. 16.</ref> A design may mix complying and non-complying features. As the degree of compliance with the prescriptive provisions increases, the scope of the fire engineering evaluation generally decreases, because only the non-compliant features need to be analysed.<ref name="IFEG-1212" />

The two approaches therefore work together rather than as strict alternatives. Prescriptive rules cover most ordinary buildings, and [[performance-based design]] is used where the rules do not fit. In the Netherlands, fire safety engineering is applied mainly where the prescribed measures are not effective, or where a building falls outside the scope of the regulations.<ref name="VR-1.5">Van Rossum (2027), §1.5.</ref> The IFEG caution that a single prescriptive requirement often serves several intents. A change made to the fire safety strategy may therefore affect other aspects of the building, such as acoustics or structural integrity.<ref name="IFEG-US0210" />

== Common misconceptions ==
Compliance with prescriptive rules does not mean that a building is free of risk. The IFEG explain that the fire safety benchmark of a performance code cannot guarantee absolute safety or zero risk, because these are not possible.<ref name="IFEG-US028">National Research Council of Canada et al. (2005), p. 0.2-8 (Part 0, United States).</ref> The benchmark must reflect what a community expects and what the protection costs, and the IFEG add that the same is true of prescriptive codes.<ref name="IFEG-US028" />

Compliance also does not mean that the building itself is protected. Meacham notes that some owners may not realise that code compliance can be very different from protecting their building, processes, contents or other assets.<ref name="ME-155">Meacham (2002), p. 5-155.</ref> In the Netherlands, preserving the building is explicitly not a goal of the fire safety rules of the Bbl.<ref name="VR-1.2" /> In Australia, the NCC addresses the protection of the subject building and its contents only to a limited extent for some building classes.<ref name="HB-27">Australian Building Codes Board (2020b), p. 27.</ref>

Prescriptive rules are not simply less refined than engineered solutions. The IFEG record a concern that performance designs may remove redundancies that prescriptive codes contain. An example is relying on a single active fire protection system instead of a combination of passive compartmentation and active systems.<ref name="IFEG-US026" /> Nor do performance-based frameworks avoid prescriptive elements altogether. The commentary to C/VM2 describes its firefighting operations scenario as prescriptive in nature, simply because no design methodology is available for these issues.<ref name="MBIE13-196">Ministry of Business, Innovation and Employment (2013), p. 196.</ref>

== See also ==
* [[Fire safety strategy]]
* [[Fire scenario]]
* [[Ethics in fire safety engineering]]
* [[Fire safety objectives]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Meacham, B. J. (2002). Building fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-153–5-175). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Safety Design & Regulation]]

Evacuation strategy

2026-11-02T17:14:17Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

An '''evacuation strategy''' is the planned way in which the occupants of a building respond to a fire: who leaves, in what order, by which route and to which place of safety. The main types are simultaneous evacuation of the whole building, phased evacuation of the most threatened floors first, progressive horizontal evacuation to an adjacent fire compartment, and strategies in which occupants stay where they are.<ref name="IFEG-1.8-2">National Research Council of Canada et al. (2005), p. 1.8-2.</ref><ref name="IFEG-2.8-2">National Research Council of Canada et al. (2005), p. 2.8-2.</ref> The chosen strategy determines the alarm and communication systems a building needs, the capacity of its stairs and the role of staff during a fire.<ref name="ABCB-D1-247">Australian Building Codes Board (2020a), Data Sheet D1, p. 247.</ref>

Traditional practice is to evacuate everyone as soon as the alarm sounds. This is simple, but in tall buildings it can leave the people most at risk queuing behind everyone else.<ref name="IFEG-1.8-2" /><ref name="Proulx-3-356">Proulx (2002), p. 3-356.</ref> Managed strategies move those people first, but they rely on reliable detection, clear announcements and trained staff.<ref name="Proulx-3-356" /><ref name="ABCB-D1-250">Australian Building Codes Board (2020a), Data Sheet D1, p. 250.</ref> In Australia, the 2020 data sheets for the Fire Safety Verification Method require the evacuation management strategy to be specified as part of the fire safety strategy.<ref name="ABCB-D1-248">Australian Building Codes Board (2020a), Data Sheet D1, p. 248.</ref> Its expected performance must also be stated, because evacuation procedures can change during the life of a building.<ref name="ABCB-D1-249">Australian Building Codes Board (2020a), Data Sheet D1, p. 249.</ref>

== Types of strategy ==
The International Fire Engineering Guidelines (2005) describe the traditional practice as starting the [[evacuation]] of occupants in response to fire alarms, following an evacuation management plan.<ref name="IFEG-1.8-2" /> In this simultaneous or total evacuation, every occupant leaves at once. Proulx calls it uncontrolled total evacuation, because the order of movement depends mainly on how people give way to one another on the stairs.<ref name="Proulx-3-355">Proulx (2002), p. 3-355.</ref>

In high-rise buildings with an emergency warning and intercommunication system, trained personnel can manage the evacuation instead. Occupants on floors far from the fire are first placed on alert and evacuated progressively only if the fire continues to develop.<ref name="IFEG-1.8-2" /> This is known as [[phased evacuation]] or controlled selective evacuation. Its success depends on the type of control and on a voice communication system.<ref name="Proulx-3-355" /> In the World Trade Center investigation, phased evacuation meant that the occupants of the fire floor and the floors immediately around it went down to three floors below the fire floor. The report notes that this approach is also known as defend-in-place.<ref name="Averill-xxxv">Averill et al. (2005), p. xxxv.</ref>

In hospitals the first response is different. Nursing staff move patients horizontally to an adjacent smoke zone or fire compartment, and they are moved further, or other floors evacuated, only if the fire threat keeps developing.<ref name="IFEG-2.8-2" /> This [[progressive horizontal evacuation]] relies on staff, because in hospitals and aged care nearly all residents may need help to leave.<ref name="ABCB-D1-261">Australian Building Codes Board (2020a), Data Sheet D1, p. 261.</ref>

A fourth group of strategies keeps people where they are. In some buildings occupants go to a fire-safe refuge, a special fire compartment where they wait for rescue by the fire service. A more recent development is the protect-in-place concept, used particularly in fully sprinklered residential buildings. Occupants are encouraged to remain where they are rather than escape through corridors and stairs that may be full of smoke.<ref name="IFEG-2.8-2" /> Such approaches are also called a [[stay-put strategy]] or stay-in-place strategy. In these cases the detection system is used to inform occupants about the location and extent of the fire, and instructions are then given to the people concerned.<ref name="Schif-4-2">Schifiliti et al. (2002), p. 4-2.</ref>

[[File:fse-evacuation-strategy-v1.svg|thumb|upright=1.5|alt=Four building sketches: all leave at once; floors near the fire leave first; patients moved sideways; occupants stay in their flats.|Four types of evacuation strategy: simultaneous evacuation, phased evacuation, progressive horizontal evacuation and staying in place. Based on National Research Council of Canada et al. (2005), pp. 1.8-2 and 2.8-2.]]

New Zealand's C/VM2 (second edition, 2025) groups these under delayed evacuation strategies. They apply where occupants must stay in place or move to a place of safety inside the building. Examples are people who are detained or who are being treated in an operating theatre, a hyperbaric chamber or a dialysis unit. Such spaces must meet the definition of a place of safety. Because they are usually climate-controlled, smoke detection and the smoke control of the air handling system need special care.<ref name="MBIE-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref>

== Total versus phased evacuation ==
Proulx compares the two strategies for a hypothetical 15-storey office building with 70 people per floor and two stairs 1,120 mm wide. In a total evacuation, each stair discharges about one person per second, and people descend at about 3.6 floors per minute, or 0.5 m/s along the stair.<ref name="Proulx-3-355" /> In both cases about 9 minutes are needed to move everyone below a presumed fire on the tenth floor. The selective procedure takes about as long to bring 490 people below the ninth floor as the total evacuation takes to bring all 980 people to the ground. Starting and controlling a selective evacuation takes time.<ref name="Proulx-3-355" />

The comparison does not show that total evacuation is better. If there really were a fire on the tenth floor and everyone left at the same time, the people on the most threatened floors would queue near those floors for several minutes.<ref name="Proulx-3-355" /> With the stair doors open on the fire floor, smoke would reach all the upper floors, and people coming down from above would have to move through it.<ref name="Proulx-3-356" /> Uncontrolled total evacuation makes few demands on building management systems. It is, however, increasingly seen as too disruptive for occupants who are evacuated indiscriminately, if they can be persuaded to leave at all with ambiguous alarms. Selective evacuation needs more sophisticated systems and training, which are expensive and can delay the moment at which the right people are told to leave.<ref name="Proulx-3-356" />

Drills give practical numbers. In 1972, 1,453 people left a 15-storey office building by four stairs 1,140 mm wide in a total evacuation. The stairs emptied after 6.6 to 9.3 minutes, 7.8 minutes on average, against a predicted 7.0 minutes.<ref name="Proulx-3-356" /> In the same drill, 73 people who would have found the stairs too arduous, together with their assistants, used a fifth stair reserved for them.<ref name="Proulx-3-356" /> The time differences between the stairs came mainly from unequal numbers of people using each stair.<ref name="Proulx-3-356" />

For the 21-storey building of the Ottawa drill described below, Proulx estimates that an uncontrolled total evacuation of its 2,100 occupants by two stairs would take a minimum of nearly 20 minutes. If everyone stood at the stair doors at once, there would be only 0.17 m² of stair, about half a tread, per person.<ref name="Proulx-3-356" /> The selective procedure used in that drill could have been stopped after 6 minutes, once the presumed fire floor and the two adjacent floors were clear.<ref name="Proulx-3-356" />

The World Trade Center investigation simulated both strategies for one tower. A phased evacuation of 600 occupants from three floors took 3.5 to 4 minutes when everyone started at once, and 11 to 11.5 minutes with random delays of up to 10 minutes.<ref name="Averill-160">Averill et al. (2005), p. 160.</ref> A total evacuation of a fully occupied tower of 19,800 people would have taken at least 92 to 112 minutes.<ref name="Averill-xxxv" /> NIST concluded that phased evacuation was not appropriate for an emergency on the scale of 11 September 2001. Under certain circumstances, though, it moves the people most at risk to relative safety much more quickly and with less disruption than a full evacuation.<ref name="Averill-164">Averill et al. (2005), p. 164.</ref> It works well for routine emergencies that fall within the design envelope of the life safety systems.<ref name="Averill-160" />

[[File:fse-evacuation-strategy-v2.svg|thumb|upright=1.5|alt=Range bars: phased evacuation of 600 people in 4 to 11 minutes against full evacuations of 8,800 to 25,500 people taking 52 to 142 minutes.|Simulated times for a phased evacuation of three floors and for full evacuations of one World Trade Center tower. Based on Averill et al. (2005), pp. xxxv and 160.]]

== Alarm and communication ==
A managed strategy needs a way to tell different groups of people different things. In Australia, alarm systems for phased evacuation commonly use two tones. An alert tone means prepare for evacuation, for example by switching off equipment and waiting for instructions, and an evacuation tone means leave immediately.<ref name="ABCB-D1-243">Australian Building Codes Board (2020a), Data Sheet D1, pp. 243–244.</ref> Because the alarm then reaches some areas later, those delays must be included in the evacuation analysis.<ref name="ABCB-D1-243" /> On floors far from the fire, access to the stair may be delayed on purpose to give priority to the fire floor.<ref name="ABCB-D1-246">Australian Building Codes Board (2020a), Data Sheet D1, p. 246.</ref>

[[File:fse-evacuation-strategy-v3.svg|thumb|upright=1.5|alt=Section of a ten-storey building: fire floor and adjacent floors hear the evacuation tone and use the stair; other floors hear an alert tone and wait.|In a phased evacuation, floors at risk receive the evacuation tone while other floors receive an alert tone and wait for instructions. Based on Australian Building Codes Board (2020a), Data Sheet D1, pp. 243–244 and 246.]]

The way the message is given matters. The book ''Introduction to Fire Safety Engineering'' notes that a slow-whoop alarm signal is known to be less effective than the spoken word. An automatic [[evacuation alarm system]] is generally less effective than a live announcement, by public address, in person or by telephone.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> The effect of an automatic alarm depends strongly on the culture of the people in the building. Where a well-trained evacuation organisation is present, the chance that the alarm is taken seriously is many times greater.<ref name="VR-7.3" />

New Zealand's C/VM2 reflects this in its design values. For retail shops and restaurants, a voice alarm halves the pre-travel activity time from 60 to 30 s in the room of fire origin. For sleeping hotel guests away from the fire, it halves the time from 600 to 300 s.<ref name="MBIE-30">Ministry of Business, Innovation and Employment (2025b), p. 30.</ref>

[[File:fse-evacuation-strategy-v4.svg|thumb|upright=1.5|alt=Grouped bars of pre-travel times: shops 60 and 120 s with a standard alarm against 30 and 60 s with voice; hotels 600 s against 300 s for remote guests.|Design pre-travel activity times in C/VM2 for a standard alarm signal and a voice alarm signal. Based on Ministry of Business, Innovation and Employment (2025b), p. 30.]]

Announcements can also fail. In 1971, one of the first large controlled selective evacuation drills was held in a 21-storey office building in Ottawa, with the fire assumed on the third floor.<ref name="Proulx-3-356" /> Because of an incorrectly set switch, the first announcement came through only after nearly 3 minutes, and it was ambiguous. Up to 350 people left their floors too early, which delayed the floors that should have gone first.<ref name="Proulx-3-356" /> With only the alarm bell ringing, 17 % of the people surveyed thought there was a fire, nearly 60 % thought it was a fault and 16 % a drill.<ref name="Proulx-3-356" /> In the World Trade Center in 2001, an announcement in the second tower telling people to return to their offices was cited by many survivors as a constraint on their evacuation.<ref name="NIST-160">National Institute of Standards and Technology (2005), p. 160.</ref>

== Strategy and building design ==
The evacuation strategy and the physical fire safety measures must fit together. The Australian data sheets point out that the deemed-to-satisfy provisions of the National Construction Code assume typical strategies. Patient care areas in hospitals and aged care buildings must be divided into smoke compartments of at most 500 m². Horizontal exits there need a clear area of at least 2.5 m² per patient on the far side of the fire wall. These rules match a phased evacuation with staff assistance.<ref name="ABCB-D1-247" /> Emergency warning and intercommunication systems are required, among other cases, in buildings with an effective height of more than 25 m.<ref name="ABCB-D1-248" />

The strategy also affects other systems. The number of stair doors open at the same time, and therefore the design of [[Stair pressurisation|stair pressurisation]], depends on whether a staged evacuation or refuge concept is part of the emergency plan. Dividing a tall stair into pressurised compartments only works with a staged evacuation, not when the whole building leaves by the stairs at once.<ref name="Klote-4-286">Klote (2002), p. 4-286.</ref><ref name="Klote-4-287">Klote (2002), p. 4-287.</ref> For very tall buildings, the Australian handbook mentions protect-in-place or partial evacuation strategies and the use of lifts as possible enhancements.<ref name="ABCB-H-49">Australian Building Codes Board (2020b), p. 49.</ref>

The evacuation strategy is therefore a part of the wider [[fire safety strategy]]. The Australian data sheets state that the description of the fire safety strategy must include clear details of the evacuation management strategy and the performance expected from it.<ref name="ABCB-D1-249" /> They also warn that changes to the evacuation management strategy should not be used on their own to justify relaxing the prescriptive requirements. Both the proposed design and the reference building are expected to have an optimised strategy.<ref name="ABCB-D1-249" />

== Use in evacuation analysis ==
In a time-based analysis, the strategy changes the inputs. New Zealand's C/VM2 takes the notification time as 30 s for standard evacuation strategies. For non-standard strategies, such as management investigating the activation of a single detector before raising the alarm, the longer notification time must be included.<ref name="MBIE-29">Ministry of Business, Innovation and Employment (2025b), p. 29.</ref> In Australia, the data sheets describe levels of emergency management organisation. These range from long-term occupants with floor wardens and regular drills to small residential buildings with no permanent staff, where managing the evacuation falls to the fire brigade.<ref name="ABCB-D1-252">Australian Building Codes Board (2020a), Data Sheet D1, p. 252.</ref>

Training, drills and a reliable intercom system make occupant behaviour more uniform and predictable, as is common in health care, most workplaces and larger public buildings. With simple alarms and an informal organisation, human behaviour varies much more, which is typical of small and medium-sized residential buildings.<ref name="ABCB-D1-250" /> The analysis must therefore reflect the strategy that will actually be in place. Scenarios with a failed alarm or intercom system also need to be considered.<ref name="ABCB-D1-250" />

== Limits and failure modes ==
Every strategy rests on assumptions that can fail. A phased evacuation assumes that the fire stays on the floor of origin. The Australian data sheets warn that it may be sensitive to rapid fire spread over several floors via the façade, and that procedures for that case should be described.<ref name="ABCB-D1-250" /> At the World Trade Center in 2001, the scale of the event made a phased approach unsuitable.<ref name="Averill-164" /> In the first stages a full-building evacuation took place from all floors not badly affected by the impacts. Had stair pressurisation been installed, this would have lowered the pressure in the stairs below the required value.<ref name="FS-111">Ferreira and Strege (2005), p. 111.</ref>

Strategies in which people stay put depend on the protection around them and on reliable information about the fire.<ref name="Schif-4-2" /> NIST observed that current strategies generally required mobility-impaired occupants to shelter in place and wait for assistance. It recommended that new procedures should consider full building evacuation, so that these occupants have a way to leave on their own.<ref name="NIST-216">National Institute of Standards and Technology (2005), p. 216.</ref> The Australian data sheets state more generally that the fire safety strategy must show how the safety of all occupants is achieved, usually by combining physical fire safety features with evacuation management procedures.<ref name="ABCB-D1-251">Australian Building Codes Board (2020a), Data Sheet D1, p. 251.</ref>

== See also ==
* [[Fire safety management]]
* [[Evacuation of people with disabilities]]
* [[Evacuation lift]]
* [[Pre-movement time]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Ferreira, M. J., & Strege, S. M. (2005). ''Smoke management systems'' (NIST NCSTAR 1-4D). National Institute of Standards and Technology.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Fire modelling

2026-11-02T17:14:15Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Fire modelling''' is the simplified imitation of a fire, a room or a building, using relevant information and boundary conditions, to estimate what a fire will do and how it affects people and construction.<ref name="VR8.1">Van Rossum (2027), §8.1.</ref> A fire model can be anything from a few formulas worked out by hand to a computer simulation that divides a building into millions of small cells.<ref name="VR8.2">Van Rossum (2027), §8.2.</ref><ref name="FDS3">McGrattan et al. (2026), p. 3.</ref> Fire safety engineers use such models to estimate temperatures, smoke layer heights and visibility, and to compare the time available for escape with the time people need.<ref name="Wa190" /><ref name="VR6.1">Van Rossum (2027), §6.1.</ref><ref name="VM2-14">Ministry of Business, Innovation and Employment (2025b), p. 14.</ref>

The three main types are hand calculations, zone models and computational fluid dynamics (CFD), and they differ greatly in detail and computing time.<ref name="NUREG2-13">Najafi et al. (2007), p. 2-13.</ref> A zone model such as CFAST typically needs a few tens of seconds of computer time, while a CFD simulation may take hours, days or even weeks.<ref name="CFAST1">Peacock et al. (2026), p. 1.</ref><ref name="VR11.1">Van Rossum (2027), §11.1.</ref> Every model is a simplification, and its result depends on what the user puts in. ''Introduction to Fire Safety Engineering'' stresses that a fire simulation cannot predict how a fire will develop. It can only make visible the effects of a fire, given the assumptions of the modeller.<ref name="VR8.1" /> For that reason, guidance in several countries asks that models are verified, validated and used within their field of application.<ref name="NUREGxv">Najafi et al. (2007), p. xv.</ref><ref name="HB108">Australian Building Codes Board (2020b), p. 108.</ref>

== What a fire model is ==
A model always leaves information out. ''Introduction to Fire Safety Engineering'' states that the modeller must justify why leaving out certain information is acceptable. It also says the modeller must check whether the scenario falls within the field of application of the model.<ref name="VR8.1" /> The user should know how the model is built, which simplifications it contains, how accurate its parts are, and what it does and does not do.<ref name="VR8.1" /> However polished a simulation looks, it does not take over the thinking of the fire safety engineer.<ref name="VR8.1" />

Models take several forms. The book distinguishes real tests, such as a furnace test or a fire in a building due for demolition, from simple calculation models and complex computer simulations.<ref name="VR8.2" /> Phillips classifies models more generally as descriptive, physical and symbolic. Physical models include scale models, such as transparent plastic models that show how smoke flows through a building.<ref name="Ph112">Phillips (2002), pp. 5-112–5-113.</ref> Symbolic models describe a system with algebraic symbols and equations. Many problems in fire science lead to differential equations that have no simple solution and are solved numerically on a computer.<ref name="Ph113">Phillips (2002), p. 5-113.</ref>

Models also differ in purpose. Some are used to compare situations or products. A furnace test, for example, does not follow a real fire, but it is the same for every test so that products can be compared fairly.<ref name="VR8.2" /> Models for large fire compartments often compare a large compartment with one of, for example, 1,000 m².<ref name="VR8.2" /> Other models try to represent reality as closely as possible, such as a CFD simulation or an evacuation simulation. For these, leaving out information must be defensible.<ref name="VR8.2" />

== Types of fire models ==
Walton divides computer fire models into two classes. Stochastic or probabilistic models treat fire growth as a series of events, with probabilities assigned to each transition, for example from ignition to established burning. Deterministic models describe the processes in a compartment fire with equations based on physics and chemistry.<ref name="Wa189">Walton (2002), p. 3-189.</ref> In practice, three types of deterministic model are used. The US Nuclear Regulatory Commission, for example, evaluated two libraries of engineering calculations, two two-zone models and one field model.<ref name="NUREG2-13" />

[[File:fse-fire-modelling-v1.svg|thumb|upright=1.5|alt=Three views of the same room fire: a single formula result, a room split into two layers, and a room divided into a grid of cells.|The same room fire as seen by a hand calculation (one value from a formula), a zone model (two uniform layers) and a CFD model (a grid of cells). Based on Najafi et al. (2007), p. 2-13, and Walton (2002), p. 3-189.]]

[[Hand calculations]] are closed-form algebraic expressions, often programmed in spreadsheets. The Fire Dynamics Tools of the US Nuclear Regulatory Commission are an example of such a library.<ref name="NUREG2-13" /> The Australian Fire Safety Verification Method (FSVM) handbook links such equations to specific phenomena, such as fire plumes, smoke layers, ceiling jets, vent flows and flashover.<ref name="HB109">Australian Building Codes Board (2020b), p. 109.</ref> Walton notes that manual methods give good estimates of specific effects, such as the time to flashover. They are less suited to analyses in which many physical and chemical processes interact over time.<ref name="Wa189" />

A [[zone model]] divides each room into two layers: a hot upper layer of smoke and a cooler lower layer.<ref name="Qu162">Quintiere (2002a), p. 3-162.</ref> The fire acts as a source of energy and mass, and its plume carries air from the lower layer into the upper layer.<ref name="Qu162" /> Conservation equations for mass and energy are solved for each layer, and the properties within each layer are taken as uniform.<ref name="CFAST1" /> Before flashover, a zone model uses these two zones. When conditions approach flashover, the room may be treated as one well-mixed zone instead.<ref name="VR9.1">Van Rossum (2027), §9.1.</ref>

[[File:fse-fire-modelling-v2.svg|thumb|upright=1.5|alt=Room cross-section with a fire, a rising plume, a grey upper smoke layer, a clear lower layer and flows in and out of a door.|A two-zone model: the plume carries air from the lower layer into the hot upper layer, and gases flow out of the top of the opening while fresh air enters below. Based on Quintiere (2002a), pp. 3-162–3-163.]]

The approach emerged in the mid-1970s. By the time of the SFPE Handbook, a survey had counted 21 zone models in use around the world.<ref name="Qu162" /> In 1985, ASET-B became the first zone model written for personal computers.<ref name="Wa190" /> The most used zone models in the Netherlands include the TNO filling-time model, CFAST, OZone and the NEN 6093 method.<ref name="VR9.5">Van Rossum (2027), §9.5.</ref>

[[Computational fluid dynamics]] (CFD) divides the space into a grid of cells and solves the equations of fluid flow numerically in every cell.<ref name="VR11.1" /> Walton calls these "field models" and notes that they need substantial computer resources and are relatively complex to use.<ref name="Wa189" /> Unlike zone models, CFD was developed outside the fire community and imported into it.<ref name="CK194">Cox and Kumar (2002), p. 3-194.</ref>

Two main approaches exist within CFD for fires. Cox and Kumar describe most practical approaches as solving time-averaged equations, known as the Reynolds-averaged Navier-Stokes (RANS) equations.<ref name="CK194" /> The FDS technical guide explains that this averaging smooths out the large eddies typical of fire plumes. Large eddy simulation (LES) instead calculates the large eddies directly and approximates only the small ones.<ref name="FDS2">McGrattan et al. (2026), p. 2.</ref> The most widely used fire-specific CFD program is the Fire Dynamics Simulator (FDS) of the US National Institute of Standards and Technology, with Smokeview for visualising the results.<ref name="VR11.5">Van Rossum (2027), §11.5.</ref>

== Inputs ==
According to ''Introduction to Fire Safety Engineering'', the most important inputs are the heat release rate and the energy of the fire.<ref name="VR8.4">Van Rossum (2027), §8.4.</ref> CFAST does not predict fire growth on burning objects; the user specifies the heat release rate.<ref name="CFAST3">Peacock et al. (2026), p. 3.</ref> ''Introduction to Fire Safety Engineering'' says the same of CFD: the modeller must prescribe the fire scenario and the development of the heat release rate.<ref name="VR11.4">Van Rossum (2027), §11.4.</ref> This prescribed fire is the [[design fire]]. It is often a heat release rate curve, either for a specific object such as a car or a standard curve suited to the type of building.<ref name="VR8.6">Van Rossum (2027), §8.6.</ref> A common choice is the t-squared curve, with four growth rates defined by the time to reach 1 MW: 75 s (ultra-fast), 150 s (fast), 300 s (medium) and 600 s (slow).<ref name="VR8.6" />

[[File:fse-fire-modelling-v3.svg|thumb|upright=1.5|alt=Four rising curves of heat release rate against time, reaching 1 MW at 75, 150, 300 and 600 seconds.|Standard t-squared growth curves often imposed as the fire in zone and CFD models; each reaches 1 MW at its characteristic time. Based on Van Rossum (2027), §8.6.]]

Other inputs describe the fire and the building. Once oxygen runs short, the fire becomes ventilation-controlled, and most models handle this poorly or not at all. Many models therefore need to be told what happens. All gases may still burn in the room, the excess may burn outside the opening, or combustion may stop.<ref name="VR8.4" />

The building inputs include the dimensions and shape of rooms, material properties, the number, size and position of openings, and any mechanical ventilation.<ref name="VR8.4" /> The fire load mainly sets how long the fire lasts. It matters less for escape analyses, which concern the early phase of the fire, than for the thermal load on structures.<ref name="VR8.5">Van Rossum (2027), §8.5.</ref>

Regulations may fix some of these choices. In New Zealand, C/VM2 (second edition, 2025) requires the base of the design fire to be no more than 0.5 m above the floor.<ref name="VM2-17">Ministry of Business, Innovation and Employment (2025b), p. 17.</ref> It also prescribes how leakage through walls is represented. A zone model uses a vertical vent over the height of the space, and a CFD model uses two vents, at floor and ceiling level.<ref name="VM2-16">Ministry of Business, Innovation and Employment (2025b), p. 16.</ref>

== Use in design and regulation ==
In performance-based design, fire models often supply the time to untenable conditions.<ref name="HB21">Australian Building Codes Board (2020b), p. 21.</ref> The ''International Fire Engineering Guidelines'' (2005) note that smoke production and movement can be predicted by hand calculations or computer simulations, and that fire engineers increasingly use computerised fire models.<ref name="IFEG252">National Research Council of Canada et al. (2005), p. 2.5-2.</ref> In New Zealand, C/VM2 requires the available safe egress time to be found through fire modelling, following its rules and using a specified design fire. The model and the spaces to be modelled are agreed in the fire engineering brief.<ref name="VM2-14" /> The method is intended for professional fire engineers who are proficient in fire engineering modelling methods.<ref name="VM2-7">Ministry of Business, Innovation and Employment (2025b), p. 7.</ref>

The Australian FSVM does not nominate specific methods. It leaves the choice to the fire safety engineer, who must document the basis for selecting models and inputs.<ref name="HB107">Australian Building Codes Board (2020b), pp. 107–108.</ref> The choice between hand calculations, zone models and CFD for each enclosure is agreed as part of the performance-based design brief.<ref name="HB21" /> The handbook notes that zone and CFD models treat the separate phenomena "in a more holistic manner" and are therefore in common use.<ref name="HB109" />

Fire models are often combined with models of people. Phillips describes fire simulations as having a human part, the evacuation, and a physical part, the fire effects, which the model combines in their natural order.<ref name="Ph112" /> This is the link with [[evacuation modelling]]. ''Introduction to Fire Safety Engineering'' notes that modelling covers evacuation simulations, fire models or a combination of both.<ref name="VR8.1" />

== Verification, validation and sensitivity ==
Because so much depends on the model, guidance asks for evidence that it is fit for use. The NUREG-1824 study defines [[verification and validation]] as follows. Verification checks that a model correctly represents the developer's description, that is, whether the model was "built" correctly. Validation checks that the model is a suitable representation of the real world, that is, whether the "right model" was built.<ref name="NUREG1-2">Najafi et al. (2007), p. 1-2.</ref> In the United States, the standard NFPA 805 for nuclear power plants requires that fire models are verified and validated.<ref name="NUREGxv" />

[[File:fse-fire-modelling-v4.svg|thumb|upright=1.5|alt=Flowchart: design fire and building inputs, choice of model, simulation, sensitivity checks, comparison with criteria, with a loop back to inputs.|Steps in a fire modelling study. Model choice and inputs are agreed in the design brief; the model must be verified, validated and used within its field of application. Based on Australian Building Codes Board (2020b), pp. 107–108, and Walton (2002), p. 3-190.]]

The NUREG-1824 study compared five models with experiments, following the standard guide ASTM E 1355.<ref name="NUREGxv" /> The predictions sometimes agreed with the data, but they also showed both under- and over-prediction in some circumstances.<ref name="NUREGxvi">Najafi et al. (2007), p. xvi.</ref> The study graded each model for each output. A "green" grade means that the model can be used with confidence, but only within the range of the experiments. A "yellow" grade calls for caution.<ref name="NUREG2-33">Najafi et al. (2007), p. 2-33.</ref> In Australia, the FSVM handbook asks for a review that checks whether a method is used within its field of application with suitable inputs.<ref name="HB108" />

A [[sensitivity analysis]] tests how much the result changes when an input changes. Walton advises always testing the sensitivity of model outputs to changes in inputs. If small changes in input cause large changes in output, the inputs must be chosen with great care.<ref name="Wa190">Walton (2002), p. 3-190.</ref>

Experienced users often run more than one model. Similar results increase confidence, but they do not prove accuracy, because most zone models share the same basic assumptions.<ref name="Wa190" /> Notarianni notes that common practice for uncertainty analysis is a series of single-variable sensitivity studies. This becomes impractical for a full design with hundreds of variables.<ref name="No5-44">Notarianni (2002), p. 5-44.</ref>

== Limits ==
Each type of model has its own limits. CFAST, for example, is limited to strongly stratified compartments. Its empirical correlations were developed for relatively uncluttered rooms with flat ceilings, not for corridors or vertical shafts.<ref name="CFAST2">Peacock et al. (2026), p. 2.</ref> Its simple point-source model of fire radiation limits its accuracy within a few fire diameters.<ref name="CFAST3" />

In general, zone models assume a uniform smoke layer, a single fire with a free plume, a simple box-shaped geometry and a stable layer without mixing at the interface. Air flows from ventilation systems or wind can usually not be included.<ref name="VR9.2">Van Rossum (2027), §9.2.</ref> In a zone model, the plume reaches the ceiling instantly, so transport times are not accounted for.<ref name="Qu163">Quintiere (2002a), p. 3-163.</ref> The technical guide of FDS states that zone models are quite reliable as long as detailed spatial distributions are not required and the two-layer picture reasonably fits reality.<ref name="FDS1">McGrattan et al. (2026), pp. 1–2.</ref>

CFD has different limits. Turbulent flows, which are typical of fires, cannot be calculated exactly, only approximated, so exact CFD results must be interpreted with caution.<ref name="VR11.1" /> Capturing the chemical reaction zone of a fire would need a mesh finer than 1 mm.<ref name="CK194" /> The FDS guide notes that combustion takes place at length scales of 1 mm or less, while building fires extend over tens of metres. The range of scales to cover is therefore about 10,000 to 100,000, against about 100 for a desktop computer.<ref name="FDS3" /> Models of combustion, turbulence and water spray remain approximations of science that is still developing, and ventilation-controlled fires are modelled only in a limited way.<ref name="VR11.4" />

''Introduction to Fire Safety Engineering'' concludes that even with the best input and expert use, CFD models are mainly reliable for qualitative comparisons between situations. For quantitative analysis, comparison with laboratory experiments remains the best benchmark.<ref name="VR11.4" /> It also advises a sound background in physics and much practical experience before CFD is used.<ref name="VR11.4" /> Notarianni adds that uncertainties "hard-wired" into scientific tools often go unrecognised.<ref name="No5-44" />

== See also ==
* [[Fire safety engineering]]
* [[Available safe egress time]]
* [[CFAST]]
* [[Uncertainty analysis]]
* [[Monte Carlo simulation]]
* [[Froude scaling]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Cox, G., & Kumar, S. (2002). Modeling enclosure fires using CFD. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-194–3-218). National Fire Protection Association; Society of Fire Protection Engineers.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* Najafi, B., Joglar, F., & Dreisbach, J. (2007). ''Verification and validation of selected fire models for nuclear power plant applications: Vol. 1. Main report'' (NUREG-1824; EPRI 1011999). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Notarianni, K. A. (2002). Uncertainty. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-40–5-64). National Fire Protection Association; Society of Fire Protection Engineers.
* Peacock, R. D., McGrattan, K. B., Forney, G. P., & Reneke, P. A. (2026). ''CFAST – Consolidated Fire and Smoke Transport: Vol. 1. Technical reference guide'' (Version 7, NIST Technical Note 1889v1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.TN.1889v1
* Phillips, W. G. B. (2002). Computer simulation for fire protection engineering (D. K. Beller & R. F. Fahy, Rev.). In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-112–5-124). National Fire Protection Association; Society of Fire Protection Engineers.
* Quintiere, J. G. (2002a). Compartment fire modeling. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-162–3-170). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Walton, W. D. (2002). Zone computer fire models for enclosures. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-189–3-193). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Modelling]]

Extinguishing mechanisms

2026-11-02T17:14:13Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Extinguishing mechanisms''' are the physical and chemical ways in which a fire is put out or suppressed. Four mechanisms are distinguished, and usually several act at the same time. They are physically separating the fuel from the flame, removing or diluting the oxygen, lowering the temperature of the fuel or the flame, and adding chemicals that change the combustion chemistry.<ref name="IS-B15">Iqbal and Salley (2004), p. B-15.</ref> A glossary of the US Nuclear Regulatory Commission (NRC) lists the matching methods as cooling, oxygen depletion or removal, inhibition of the chemical reaction and flame removal, or blowout.<ref name="IS-F23">Iqbal and Salley (2004), p. F-23.</ref>

The mechanisms follow from the conditions that a fire needs. The [[fire triangle]] shows fuel, oxygen and heat; taking away any one of them stops the combustion.<ref name="IS-1-5">Iqbal and Salley (2004), p. 1-5.</ref> The [[fire tetrahedron]] adds a fourth element, the chain reaction of combustion, which explains why some agents work without cooling or smothering.<ref name="GR-152">Grant (2002), p. 4-152.</ref> Each [[extinguishing agent]] uses a mix of mechanisms. Water cools, foam excludes air from the fuel surface, inert gases act primarily by depleting the oxygen, and halons interrupt the chain reaction.<ref name="IS-10-1" /><ref name="SC-88">Scheffey (2002), p. 4-88.</ref><ref name="IS-B41" /><ref name="GR-153" /> The relative importance of each contribution varies with the circumstances.<ref name="IS-B16">Iqbal and Salley (2004), p. B-16.</ref>

== The fire triangle and the fire tetrahedron ==
''Introduction to Fire Safety Engineering'' describes combustion with three factors: oxygen, fuel and temperature.<ref name="VR-4.2">Van Rossum (2027), §4.2.</ref> Air contains 78 % nitrogen, 21 % oxygen and 1 % other gases by volume. If the oxygen level falls, a fire stops spreading and may eventually go out, depending on the exact oxygen concentration.<ref name="VR-4.2" /> A fuel mixed with oxygen cannot ignite unless enough heat is added, and the amount depends on the fuel.<ref name="VR-4.2" /> In the NRC guidance, the fire triangle requires fuel, an oxidising agent and a heat source in the same place at the same time. If any leg of the triangle is removed, the combustion is not sustained.<ref name="IS-1-5" />

The triangle leads directly to three classic mechanisms. Typical extinguishment removes the fuel from the fire, limits the oxygen (smothering), or removes the heat (quenching).<ref name="GR-152" /> The NRC adds that the triangle fits most scenarios, with the exception of extinguishment by dry chemicals and halons.<ref name="IS-1-5" />

[[File:fse-extinguishing-mechanisms-v1.svg|thumb|upright=1.5|alt=Fire triangle of heat, oxygen and fuel with cooling, smothering and fuel removal; tetrahedron adding the chain reaction and chemical inhibition.|The fire triangle (left) and the fire tetrahedron (right), with the extinguishing mechanism that acts on each element. Based on Grant (2002), pp. 4-152 to 4-153 and Iqbal and Salley (2004), p. 1-5.]]

Grant explains that the theory had to be modified as halons became better understood. Halons do not work in any of the three classic ways, but break up the uninhibited chain reaction of the combustion process.<ref name="GR-152" /> The resulting model is called the tetrahedron of fire, with temperature, oxygen, fuel and the chain reaction as its four faces.<ref name="GR-153">Grant (2002), p. 4-153.</ref> The NRC glossary defines it as a representation of fire in which fuel, oxidant, heat and chain reactions are the four corners of a tetrahedron.<ref name="IS-F29">Iqbal and Salley (2004), p. F-29.</ref>

The tetrahedron is also used in firefighting practice. The Netherlands Institute for Public Safety (NIPV) published guidance on fires in Dutch indoor car parks. For an interior attack, it advises crews to think in terms of the fuel, heat, oxygen and chemical chain reaction tetrahedron.<ref name="NIPV-63">Van der Graaf et al. (2021/2023), p. 63.</ref> The same guidance asks whether enough cooling capacity is available before an attack from inside or outside is chosen.<ref name="NIPV-63" />

== Cooling ==
Cooling removes heat from the fuel or the flame. Water is by far the most common cooling agent, mainly because it is widely available and inexpensive.<ref name="FL-72">Fleming (2002), p. 4-72.</ref> It also has a high [[specific heat capacity]] and a high latent heat of vaporisation.<ref name="FL-72" /> The specific heat of water at room temperature is 4,186 J/(kg·K), about four times that of air at 1,005 J/(kg·K).<ref name="VR-3.4">Van Rossum (2027), §3.4.</ref> When water evaporates, it takes extra energy from its surroundings, because every change from liquid to gas absorbs energy.<ref name="VR-3.5">Van Rossum (2027), §3.5.</ref>

The heat that a mass of water can absorb follows from the energy needed to heat it to its boiling point plus the energy needed to evaporate it. ''Introduction to Fire Safety Engineering'' gives the sensible part as mass times specific heat times temperature rise, and notes that the energy of a phase change must be added.<ref name="VR-3.4" /> Combined, the heat absorbed is:

<math display="block">E = m \left( c_p \, \Delta T + h_v \right)</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>E</math> || heat absorbed by the water || kJ
|-
| <math>m</math> || mass of water || kg
|-
| <math>c_p</math> || specific heat of liquid water || kJ/(kg·K)
|-
| <math>\Delta T</math> || temperature rise from the initial temperature to the boiling point || K
|-
| <math>h_v</math> || heat of vaporisation of water at the boiling point || kJ/kg
|}

The relation assumes a constant specific heat and that all of the water evaporates. Water that runs off or drains away absorbs only the sensible part, or less.<ref name="VR-3.4" /> Tewarson gives the property values: a specific heat of 0.0042 kJ/(g·K) and a heat of vaporisation of 2.26 kJ/g at 373 K.<ref name="TE-100">Tewarson (2002), p. 3-100.</ref>

=== Worked example ===
Tewarson calculates the heat needed to gasify water that starts at 298 K (25 °C). Heating the water to 373 K takes 0.0042 × 75 = 0.32 kJ/g, and evaporating it takes 2.26 kJ/g, so the total is 2.58 kJ/g.<ref name="TE-100" /> One kilogram of water, about one litre, therefore absorbs about 2.58 MJ. Almost 88 % of this is absorbed by evaporation rather than by heating. Fleming gives 9,280 Btu for one US gallon of water heated from 21 °C to steam at 100 °C, which is equivalent to about 2.59 MJ per litre.<ref name="FL-72" />

[[File:fse-extinguishing-mechanisms-v2.svg|thumb|upright=1.5|alt=Stacked bar for 1 kg of water: 0.32 MJ to heat it from 25 to 100 °C and 2.26 MJ to evaporate it, total 2.58 MJ.|Heat absorbed by 1 kg of water that is heated from 25 °C and evaporated. Based on Tewarson (2002), p. 3-100.]]

An average living room sofa can burn at about 2.5 MW.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref> To absorb all of that heat, water would have to evaporate at 2,500 / 2,580 ≈ 0.98 kg/s, or about 59 L/min. If the water were only heated to 100 °C without evaporating, the same task would need about eight times as much water. In practice, much less than the full heat release has to be absorbed, as the section on flame cooling below explains. The calculation does show why evaporation, not heating, gives water most of its cooling power. In ''Introduction to Fire Safety Engineering'', the heat release rate of such an object is set against the cooling capacity that a fire service can deliver at a given flow rate.<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref>

=== Cooling the fuel ===
Cooling the fuel surface slows down the pyrolysis that feeds the flame with gaseous fuel. When water is applied to a burning solid in air, the solid is cooled by contact with the water, so its rate of pyrolysis falls.<ref name="IS-B15" /> For fuels that give no flammable vapour at room temperature, such as solids and liquids with a high flash point, wetting and cooling of the surface is in many cases the dominant mechanism.<ref name="MB-314">Mawhinney and Back (2002), p. 4-314.</ref> If the vapour and air mixture above the surface falls below the lower flammability limit, the flame goes out.<ref name="MB-314" /> The NRC describes sprinklers in the same terms. The water vaporises and cools the burning materials below their ignition temperature, and nearby materials can no longer heat up to it.<ref name="IS-10-1">Iqbal and Salley (2004), p. 10-1.</ref>

Tewarson links this mechanism to the fire properties of the fuel. He calculates a [[critical water application rate]] for [[flame extinction]], expressed in grams of water per square metre of burning surface per second.<ref name="TE-148">Tewarson (2002), p. 3-148.</ref> The calculated values assume no water puddle on the surface and perfect efficiency. They are 2.1 to 2.5 g/(m²·s) for polyoxymethylene, polymethyl methacrylate (PMMA) and polyethylene with 25 % chlorine. Polyethylene and polypropylene need 3.5 to 4.1 g/(m²·s), and polystyrene 5.1 g/(m²·s).<ref name="TE-148" /> Materials that do not burn in normal air without an external heat flux have a critical rate of zero.<ref name="TE-148" />

[[File:fse-extinguishing-mechanisms-v3.svg|thumb|upright=1.5|alt=Range bars of critical water application rate: POM, PMMA and chlorinated PE 2.1 to 2.5, PE and PP 3.5 to 4.1, polystyrene 5.1 g/(m2 s).|Calculated critical water application rates for flame extinction of plastics, with no water puddle and perfect efficiency. Based on Tewarson (2002), p. 3-148.]]

=== Cooling the flame ===
Water can also act in the gas phase. Gas phase cooling is the removal of heat from the flame and hot gases by evaporating water, and fine droplets evaporate faster.<ref name="MB-313">Mawhinney and Back (2002), p. 4-313.</ref> If the flame temperature falls below the limiting adiabatic flame temperature, about 1,600 K (1,326 °C) for diffusion flames, the flame goes out.<ref name="MB-313" /> Only enough heat has to be removed to bring the flame down to this limit, not to ambient temperature.<ref name="MB-313" /> In Wighus's experiments, the heat absorbed by evaporating water needed to be only 0.3 times the heat release rate for an unconfined propane flame under optimum conditions. For more realistic machinery spaces, ratios of about 0.6 were noted.<ref name="MB-313" />

A cooler flame also radiates less heat back to the fuel, which further reduces the pyrolysis rate.<ref name="MB-313" /><ref name="IS-B15" /> In laboratory counterflow methane flames, water mist and vapour were more effective on a mass basis than Halon 1301, provided they were delivered with nearly 100 % efficiency. In full-scale compartment fires the efficiency is much lower.<ref name="MB-313" />

== Removing or diluting the oxygen ==
A flame needs a minimum oxygen concentration. For most hydrocarbon fuels, this limiting oxygen concentration is about 13 %.<ref name="MB-314" /> Smothering means limiting the oxygen supply to the fire.<ref name="GR-152" /> In an interior attack on a building fire, NIPV advises anti-ventilation, keeping the structure closed where possible.<ref name="NIPV-63" /> ''Introduction to Fire Safety Engineering'' notes that a fire stops spreading and may go out as the oxygen level falls.<ref name="VR-4.2" />

Water can dilute the oxygen as steam. When water droplets evaporate, the volume of the water increases by more than three orders of magnitude.<ref name="MB-314" /> In the flame, this expansion can disrupt the entrainment of air. In a compartment, steam from water hitting the flame, hot gases or hot surfaces can significantly reduce the oxygen concentration.<ref name="MB-314" /> The NRC notes that steam may keep oxygen away from the fire under some confined conditions.<ref name="IS-B15" />

Inert gases work in the same way, but they also act thermally. They lower the flame temperature below the threshold for the combustion reactions, both by reducing the oxygen concentration and by raising the heat capacity of the atmosphere.<ref name="DN-174">DiNenno (2002), p. 4-174.</ref> Adding enough inert gas to reduce the oxygen concentration below 12 % extinguishes flaming fires.<ref name="DN-174" /> Carbon dioxide displaces the normal atmosphere and reduces the oxygen below the 15 % that the NRC report gives as necessary for a diffusion flame.<ref name="IS-B46">Iqbal and Salley (2004), p. B-46.</ref>

The NRC report describes the action of carbon dioxide as mainly thermophysical. The reacting gases are kept below the temperature needed to maintain the free radicals that sustain the flame chemistry.<ref name="IS-B46" /> For inert gases, the extinguishing concentration is linearly related to the heat capacity of the agent and air mixture.<ref name="IS-B46" />

== Separating the fuel from the flame ==
The third route breaks the contact between fuel, heat and air. Physically separating the combustible substance from the flame is one of the four mechanisms in the NRC list.<ref name="IS-B15" /> The simplest form is removing the fuel from the fire.<ref name="GR-152" /> Flame removal, or blowout, is listed in the NRC glossary as a further method of extinguishment.<ref name="IS-F23" />

Firefighting foam is the best known separating agent. It consists of air-filled bubbles formed from a solution of foam concentrate in water, typically 1, 3 or 6 % concentrate.<ref name="SC-88" /> The foam floats on the fuel and forms an air-excluding layer that halts the vaporisation of fuel at the surface. If the entire surface is covered, the fuel vapour is completely suppressed and the fire goes out.<ref name="SC-88" /> Film-forming foams can also produce a vapour-sealing film of surface-active water solution on most hydrocarbon fuels.<ref name="SC-88" /> Low-expansion foams are effective on flat pool fires, but not on three-dimensional fires such as running fuel or pressurised sprays.<ref name="SC-88" />

Foam combines several mechanisms on a burning pool. It stops radiant heat from the fire reaching the liquid surface and supplying the heat of vaporisation. If the fire point of the liquid is above the foam temperature, the liquid is cooled and its vapour pressure falls.<ref name="IS-B15" /> A water-soluble fuel such as alcohol is also diluted by water from the foam.<ref name="IS-B15" /> Water alone cannot separate a hydrocarbon fuel from the air. It is denser than most hydrocarbon fuels and does not mix with them, so the fuel floats on the water and may continue to burn and spread.<ref name="FL-72" />

== Chemical inhibition ==
Chemical inhibition interrupts the chain reaction in the flame rather than removing a leg of the triangle. Halogenated agents act in this way, although the mechanism is not completely understood.<ref name="GR-153" /> The halogen atoms remove the active chemical species that carry the flame chain reaction.<ref name="GR-153" />

Bromine is much more effective than chlorine or fluorine.<ref name="GR-153" /> In Halon 1301, which is 54 % bromine by weight, it is the bromine radical that acts as the inhibitor.<ref name="GR-153" /> The NRC report states that Halon 1301 suppresses a fire chemically and very quickly, unlike agents that work by removing heat or oxygen.<ref name="IS-B42">Iqbal and Salley (2004), p. B-42.</ref>

The replacements for halon show that the boundary between chemical and physical action is not sharp. Halocarbon clean agents extinguish fires by a combination of chemical and physical mechanisms, depending on the compound.<ref name="DN-174" /> Compounds with bromine or iodine scavenge flame radicals in the same way as Halon 1301. Other compounds work mainly by extracting heat from the flame zone, through their heat of vaporisation, their heat capacity and the energy absorbed when they decompose.<ref name="DN-174" /> Because these compounds have little chemical inhibition, they need higher extinguishing concentrations than Halon 1301.<ref name="DN-174" /> The NRC report likewise describes halocarbon agents as acting largely by absorbing heat, with some chemical effect on the flame reactions.<ref name="IS-B41">Iqbal and Salley (2004), p. B-41.</ref>

Dry chemical agents also combine mechanisms. They can interact chemically with the flame, coat the fuel surface, cool the flame and block the transfer of radiant heat.<ref name="IS-B15" /> The NRC also treats extinguishment by dry chemicals, like that by halons, as an exception to the simple fire triangle.<ref name="IS-1-5" />

== Combined action of common agents ==
In practice, no agent relies on a single mechanism, and the relative importance of each mechanism varies with the circumstances.<ref name="IS-B16" /> The table summarises the mechanisms that the sources describe for the most common agents. It is based on the descriptions of the individual agents.<ref name="IS-B15" /><ref name="IS-B41" /><ref name="IS-B46" /><ref name="MB-313" /><ref name="MB-314" /><ref name="SC-88" /><ref name="DN-174" /><ref name="GR-153" />

{| class="wikitable"
|+ Main extinguishing mechanisms of common agents
! Agent !! Cooling !! Oxygen removal or dilution !! Separation of fuel and flame !! Chemical inhibition
|-
| Water spray || fuel and flame || steam, in confined conditions || – || –
|-
| Water mist || flame and fuel || steam, locally and in the compartment || – || –
|-
| Foam || cools the liquid || – || floating, air-excluding layer || –
|-
| Dry chemical || flame || – || coats the surface || yes
|-
| Inert gases, carbon dioxide || raised heat capacity || primary action || – || –
|-
| Halon 1301 and bromine or iodine clean agents || – || – || – || primary action
|-
| Other halocarbon clean agents || heat extraction (primary) || important role || – || minor
|}

Water mist shows how mechanisms interact. Mawhinney and Back name gas phase cooling, oxygen depletion with dilution of the fuel vapour, and wetting and cooling of the fuel surface as primary mechanisms. Radiation attenuation and kinetic effects are secondary.<ref name="MB-313" /> All of these act to some degree, with different weights for liquid and solid fuels.<ref name="MB-313" /> Water vapour between the fuel and the flame absorbs radiant energy, which helps to prevent fire spread to unburnt fuel.<ref name="MB-314" />

[[File:fse-extinguishing-mechanisms-v4.svg|thumb|upright=1.5|alt=Pool fire under water mist with labels for gas phase cooling in the flame, steam diluting oxygen, wetting of the fuel surface and radiation attenuation.|Where the extinguishing mechanisms of water mist act on a fire. Based on Mawhinney and Back (2002), pp. 4-313 to 4-314.]]

The choice of agent also depends on the class of fire. The NRC table matches agents to fire classes. Water, dry chemicals, foam and some halons are listed for ordinary combustibles (Class A), and carbon dioxide, dry chemicals, halons and foam for flammable liquids and gases (Class B).<ref name="IS-B16" /> Water and water-based extinguishers must never be used on burning metals (Class D).<ref name="IS-B16" />

== Limits and misconceptions ==
A common misconception is that the fire triangle explains every extinguishment. The NRC notes that it does not apply to dry chemicals and halons, which act on the chain reaction.<ref name="IS-1-5" /> Another is that a flame that has gone out means the fire is extinguished. Halon 1301 at normal concentrations of 5 to 6 % instantly extinguishes visible flames on Class A materials. Glowing combustion inside the material may continue and reignite when the enclosure is ventilated.<ref name="IS-B43">Iqbal and Salley (2004), p. B-43.</ref>

Some fuels resist the usual mechanisms. Reactive metals such as magnesium, sodium and titanium burn so intensely that they overpower halon.<ref name="GR-152" /> Fuels that contain their own oxidiser, such as gunpowder and cellulose nitrate, also burn freely in a halon atmosphere.<ref name="GR-152" /> Water can be inappropriate for water-reactive materials, because it may produce heat, flammable or toxic gases, or explosions.<ref name="FL-72" />

Water can even make a flame worse at first. Flare-up of a flame on first contact with water mist has been measured, possibly because the rapid evaporation adds turbulence and speeds up burning.<ref name="MB-314" /> Finally, the mechanisms are understood far better than they can be calculated. The ways in which water suppresses fire have not been characterised well enough for mathematical modelling, so sprinkler design remains empirical.<ref name="FL-73">Fleming (2002), p. 4-73.</ref> No fundamental explanation of foam performance based on first principles has been developed either.<ref name="SC-88" /> Even where water cannot suppress a fire, its cooling can still protect the structure by containing the fire until other means put it out.<ref name="FL-72" />

== See also ==
* [[Firefighting foam]]
* [[Firefighting]]
* [[Water mist system]]

== Notes ==
<references />

== References ==
* DiNenno, P. J. (2002). Halon replacement clean agent total flooding systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-173–4-200). National Fire Protection Association; Society of Fire Protection Engineers.
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* Grant, C. C. (2002). Halon design calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-149–4-172). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Mawhinney, J. R., & Back, G. G., III. (2002). Water mist fire suppression systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-311–4-337). National Fire Protection Association; Society of Fire Protection Engineers.
* Scheffey, J. L. (2002). Foam agents and AFFF system design considerations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-88–4-122). National Fire Protection Association; Society of Fire Protection Engineers.
* Tewarson, A. (2002). Generation of heat and chemical compounds in fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-82–3-161). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Fire Suppression]]

Detonation

2026-11-02T17:14:10Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''detonation''' is the propagation of a combustion zone through a flammable mixture at a speed greater than the speed of sound in the unreacted mixture.<ref name="IS-F-21">Iqbal and Salley (2004), p. F-21.</ref> It is led by a shock wave that is driven and sustained by the energy released in the chemical reaction, and the shock and the reaction zone travel together at roughly 1,500 to 2,000 m/s.<ref name="IS-16-9">Iqbal and Salley (2004), p. 16-9.</ref> The shock compresses the unburned mixture ahead of the reaction, which is why the reaction can propagate so fast.<ref name="IS-15-3">Iqbal and Salley (2004), p. 15-3.</ref>

A detonation differs from a [[deflagration]], the much more common subsonic form of [[explosion]], in the way it loads its surroundings. In a deflagration the pressure rises almost evenly through an enclosure; in a detonation it rises suddenly and unevenly as the shock passes.<ref name="Z-3-410">Zalosh (2002), p. 3-410.</ref> The peak pressure is about twice that of a closed vessel deflagration, roughly 16 to 20 atmospheres for gas-air mixtures that start at atmospheric pressure.<ref name="Z-3-411">Zalosh (2002), p. 3-411.</ref> Because the pressure peak arrives with the shock, explosion vents and explosion suppression systems cannot protect against a detonation.<ref name="Z-3-411" /><ref name="Z-3-414">Zalosh (2002), p. 3-414.</ref> Accidental detonations are rare, and in practice they develop from a deflagration that accelerates rather than from direct initiation.<ref name="Z-3-412">Zalosh (2002), p. 3-412.</ref><ref name="Z-3-413">Zalosh (2002), p. 3-413.</ref><ref name="IS-15-12">Iqbal and Salley (2004), p. 15-12.</ref>

== Definition and distinction from deflagration ==
NUREG-1805 distinguishes detonations and deflagrations mainly by the speed of the combustion wave.<ref name="IS-15-3" /> In a detonation the wave propagates through the reactants at supersonic speed, of the order of 2,000 m/s.<ref name="IS-15-3" /> In a deflagration the speed stays below that of sound in air at 20 °C, approximately 330 m/s.<ref name="IS-15-3" /> Supersonic propagation implies a shock wave, which compresses the reactants and makes them react faster.<ref name="IS-15-3" />

The shock front brings an abrupt increase in pressure, temperature and density of the gas, and moves the gas particles forward.<ref name="IS-16-9" /> Zalosh describes the consequence for an enclosure. In a deflagration the flames move well below the speed of sound, about 340 m/s in air at room temperature, so the pressure rises virtually uniformly throughout the enclosure.<ref name="Z-3-410" /> In a detonation the pressure rise is highly non-uniform and happens virtually instantaneously as the shock passes through the mixture.<ref name="Z-3-410" />

Between the two lies the quasi-detonation. If the flame speed is slightly less than the speed of sound, the pressure rise is non-uniform, but no shock waves occur.<ref name="Z-3-410" /> Like a detonation, a quasi-detonation rules out the use of explosion venting and explosion suppression.<ref name="Z-3-411" />

The definitions in the sources apply to combustion waves in gas mixtures. NUREG-1805 also lists condensed phase explosions of high explosives and propellants as a separate class of chemical explosion.<ref name="IS-15-2">Iqbal and Salley (2004), p. 15-2.</ref> The blast of such explosives is the reference for the [[TNT equivalence]] method, which expresses a gas explosion as an equivalent mass of TNT. The method is used because the blast effects of TNT have been studied extensively as a function of charge mass and distance.<ref name="IS-15-10">Iqbal and Salley (2004), p. 15-10.</ref>

== Structure of a detonation wave ==
The idealised theory of the detonation front is the Chapman-Jouguet theory, which combines thermochemical equilibrium with the conservation equations of gas dynamics across the front.<ref name="Z-3-411" /> It gives the peak pressure and the speed of the wave, and its results can be calculated with chemical equilibrium programs such as STANJAN and CET89.<ref name="Z-3-411" /> The speed equals the Chapman-Jouguet Mach number times the speed of sound in the unburned mixture.<ref name="Z-3-411" />

For stoichiometric gas-air mixtures, the Chapman-Jouguet Mach numbers lie in the narrow range 4.9 to 5.5.<ref name="Z-3-411" /> The speed of sound in most stoichiometric hydrocarbon-air mixtures is 330 to 350 m/s, and in hydrogen-air it is 403 m/s.<ref name="Z-3-411" /> With the Mach number of 5.38 for propane-air, the detonation therefore travels at roughly 1,780 to 1,880 m/s.<ref name="Z-3-411" /><ref name="Z-3-412" /> For hydrogen-air, with a Mach number of 4.89, it is close to 2,000 m/s.<ref name="Z-3-411" /><ref name="Z-3-412" /> Simmons gives a typical range of 1,500 to 3,000 m/s for detonations that develop in ducts.<ref name="S-1-147">Simmons (2002), p. 1-147.</ref>

Behind the front, the pressure falls to a plateau of about 0.35 times the Chapman-Jouguet pressure when the wave runs away from the ignition point in both directions in a tube.<ref name="Z-3-411" /> If the wave meets a rigid closed end, it is reflected with a pressure that is about 2.76 times the Chapman-Jouguet pressure for many gas-air mixtures.<ref name="Z-3-411" /> If the end is open, the wave is reflected as a rarefaction wave, which lowers the pressure.<ref name="Z-3-411" /> Simmons quotes up to 20 bar behind a detonation in a gas that starts at atmospheric pressure, and up to 100 bar when the wave is reflected from the end of a duct.<ref name="S-1-147" />

[[File:fse-detonation-v1.svg|thumb|upright=1.5|alt=Tube with a detonation front and a pressure graph: sharp peak at the front, lower plateau behind, higher reflected peak at a closed end.|Pressure in a tube during a detonation: a sharp peak at the front, a plateau of about 0.35 times the peak behind it, and about 2.76 times the peak when the wave reflects from a rigid closed end. Based on Zalosh (2002), p. 3-411.]]

A real detonation front is not flat. It consists of a network of curved shock segments that move across the direction of travel.<ref name="Z-3-412" /> Where they meet, they trace a diamond or fish-scale pattern on a smoke foil placed on the inner wall of a test tube.<ref name="Z-3-412" /> The width of one such cell, the detonation cell width, is the length scale that describes how easily a mixture detonates.<ref name="Z-3-412" />

== Peak pressure ==
The Chapman-Jouguet calculation needs a chemical equilibrium program. Burgess and co-workers suggested a much simpler approximation, quoted by Zalosh: the detonation pressure is about twice the pressure at the end of a closed vessel deflagration of the same mixture.<ref name="Z-3-411" /> In words: if the same mixture would reach a certain absolute pressure when it burns slowly in a closed vessel, its detonation reaches about double that pressure at the front.

<math display="block">P_\mathrm{CJ} \approx 2\, P_\mathrm{m}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>P_\mathrm{CJ}</math> || Chapman-Jouguet detonation pressure (absolute) || bar
|-
| <math>P_\mathrm{m}</math> || pressure at the end of a closed vessel deflagration of the same mixture (absolute) || bar
|}

The approximation gives the pressure at the detonation front itself in a gas-air mixture.<ref name="Z-3-411" /> It does not include reflection, which raises the pressure much further at closed ends and walls.<ref name="Z-3-411" /> Calculated values for near-stoichiometric mixtures at atmospheric initial pressure lie between 16 and 20 atmospheres.<ref name="Z-3-411" />

'''Worked example.''' For propane in air, the measured maximum closed vessel deflagration pressure is 7.9 bar gauge.<ref name="Z-3-407">Zalosh (2002), p. 3-407.</ref> With an atmospheric pressure of 1.013 bar this is about 8.9 bar absolute, so:

<math display="block">P_\mathrm{CJ} \approx 2 \times 8.9\ \mathrm{bar} \approx 18\ \mathrm{bar\ (absolute)}</math>

The result lies within the range of 16 to 20 atmospheres given for near-stoichiometric mixtures.<ref name="Z-3-411" /> At a rigid closed end, the reflected pressure would be about 2.76 × 17.8 ≈ 50 bar. Behind a freely running wave the pressure falls to about 0.35 × 17.8 ≈ 6.3 bar (values rounded up, since they are loads).<ref name="Z-3-411" /> For methane, with a measured deflagration maximum of 7.1 bar gauge, the same approximation gives about 16 bar absolute.<ref name="Z-3-407" />

The comparison with other values in the sources shows the scale. NUREG-1805 gives a detonation pressure of up to 20 atmospheres (284 psi) and estimates detonation pressure rises at 2 to 4 times those of a deflagration.<ref name="IS-15-3" /><ref name="IS-16-8">Iqbal and Salley (2004), p. 16-8.</ref> It also lists pressure effects on people: about 35 kPa is the limit for eardrum rupture and 180 kPa corresponds to 1 % mortality, far below the pressures inside a detonation.<ref name="IS-15-12" />

== How a detonation starts ==
A detonation can start in two ways: by direct initiation with a very strong ignition source, or by acceleration of a deflagration.<ref name="Z-3-412" /> The minimum energy for direct initiation ranges from about 5 kJ for acetylene and hydrogen in air to an estimated 93,000 kJ for methane.<ref name="Z-3-412" /> These energies are many orders of magnitude above those of accidental ignition sources, so direct initiation can be ruled out in almost all accident scenarios.<ref name="Z-3-412" />

The usual route is the [[deflagration-to-detonation transition]]. When the hot products of a flame in a duct cannot vent, they push the flame and the unburned gas forward with increasing speed, which creates turbulence ahead of the flame.<ref name="S-1-147" /> In a long duct the flame front then acts as an accelerating piston, a shock wave forms ahead of it, and the flame becomes a detonation.<ref name="S-1-147" /> The run-up distance is shorter for more reactive mixtures, rough walls or obstructions, larger diameters, higher initial pressure and temperature, and stronger initial turbulence.<ref name="Z-3-412" /><ref name="Z-3-413" />

Smooth pipes need long run-up distances. In tests in a smooth 5 cm pipe, alkanes in air did not undergo transition within 120 pipe diameters.<ref name="Z-3-413" /> Flessner and Bjorkland saw no transition with gasoline-air mixtures in a smooth pipe 15 cm in diameter and 16 m long. When they placed an expanded metal liner at the pipe entrance to simulate a rough wall, transition occurred after 74 pipe diameters.<ref name="Z-3-413" /> NUREG-1805 adds that the ability for such a transition depends on both the mixture and its environment, and that detonable limits, unlike flammability limits, are also influenced by the environment.<ref name="IS-B-26">Iqbal and Salley (2004), p. B-26.</ref> In process plants, a flammable mixture in a vessel or pipework may explode either as a deflagration or as a detonation.<ref name="IS-15-5">Iqbal and Salley (2004), p. 15-5.</ref>

== Detonability and cell width ==
A sustained detonation is only possible if the space filled with the mixture is larger than some multiple of the cell width.<ref name="Z-3-412" /> In a pipe, a detonation will not propagate if the pipe diameter is less than about one third of the cell width.<ref name="Z-3-412" /> It will not pass from the open end of a pipe into the surrounding mixture if the pipe diameter is less than 13 cell widths.<ref name="Z-3-412" />

[[File:fse-detonation-v2.svg|thumb|upright=1.5|alt=A detonation cell and three pipes: a thin pipe stops the detonation, a medium pipe carries it, a wide pipe passes it into a room.|Critical sizes for a detonation in a pipe, expressed in the cell width Sc of the mixture. Based on Zalosh (2002), p. 3-412.]]

Measured or calculated cell widths for stoichiometric gas-air mixtures range from about 1 cm for acetylene to about 28 cm for methane.<ref name="Z-3-412" /> Hydrogen has 1.5 cm, ethylene 2.8 cm and propane 6.9 cm.<ref name="Z-3-412" /> These values apply at atmospheric pressure and room temperature; higher initial pressure or temperature gives smaller cells, and some fuel-rich mixtures have cells up to 40 % smaller than the tabulated values.<ref name="Z-3-412" /> A small cell width therefore means a mixture that can detonate in smaller pipes and spaces.<ref name="Z-3-412" />

[[File:fse-detonation-v3.svg|thumb|upright=1.5|alt=Bar chart of detonation cell widths: acetylene about 1 cm, hydrogen 1.5 cm, ethylene 2.8 cm, propane 6.9 cm, methane 28 cm.|Detonation cell widths of stoichiometric gas-air mixtures at atmospheric pressure and room temperature; a smaller cell means a mixture that can detonate in smaller spaces. Based on Zalosh (2002), p. 3-412, Table 3-16.4.]]

Zalosh gives an example of how the criterion is used. A process oven is fed by a fuel line 1 cm in diameter and 5 m long that contains a stoichiometric propane-air mixture.<ref name="Z-3-413" /> With a cell width of 6.9 cm, a detonation needs a pipe of at least 2.3 cm, so it cannot run through the line.<ref name="Z-3-413" /> With hydrogen-air the limit is 0.50 cm, so a detonation in the line would be possible, but it could only pass into the oven through a pipe wider than about 20 cm.<ref name="Z-3-413" /> Explosion venting can therefore still protect the oven.<ref name="Z-3-413" />

== Effects on structures ==
Detonations load structures in a different way from deflagrations. Their pressure loads have a high peak and a short duration, which calls for special consideration when the structural resistance is evaluated.<ref name="Z-3-411" /> The load is described by the impulse, the integral of the overpressure over time, which in a pipe grows with the distance from the point of initiation.<ref name="Z-3-411" /> Pipeline detonations often cause periodic ruptures along the length of the pipe: each break relieves the pressure, and the load builds up again over the next length.<ref name="Z-3-411" />

The remains after an explosion can show which kind it was. Detonative loads tend to produce fragmented debris from brittle failure, whereas structural steel loaded by deflagration pressures beyond its yield point bulges and fails in a more ductile way.<ref name="Z-3-411" /><ref name="Z-3-412" /> In the hydrogen fuel-line example, the average pressure on the pipe wall during the passage of the detonation was about 530 kPa, about one third of the Chapman-Jouguet pressure.<ref name="Z-3-414" />

Well-documented accidental detonations are rare, since most accidental explosions are deflagrations.<ref name="Z-3-413" /> One example is a refinery accident described by Jacobs and co-workers. A mixture of 3 % naphtha, 19 % oxygen and 78 % inert gas at 105 psig, about 720 kPa above atmospheric, entered several large items of process equipment.<ref name="Z-3-413" /> The equipment was connected by more than 1,000 ft (about 300 m) of piping, and pressures of 3,000 to 4,000 psi, about 21 to 28 MPa, developed at some locations.<ref name="Z-3-413" />

== Detonations in open clouds and assessment ==
Outside equipment, the related hazard is the [[vapour cloud explosion]]: a large cloud of flammable gas or vapour ignited in an obstructed area such as a chemical plant.<ref name="Z-3-420">Zalosh (2002), p. 3-420.</ref> The strength of its blast depends on the flame speed that develops and on the degree of confinement.<ref name="Z-3-420" /> NUREG-1805 notes that a premixed gas-air mixture ignited inside a building gives a maximum pressure of about 8 bar. An unconfined vapour cloud explosion gives pressures of the same order, and a detonation even higher pressures.<ref name="IS-15-12" /> It adds that detonations are very rare.<ref name="IS-15-12" />

No exact criteria exist to decide whether a given situation will produce a detonation rather than a deflagration.<ref name="Z-3-412" /> Sherman and Berman developed a semi-quantitative method that classifies mixtures by cell size and geometries by size, confinement, obstructions and venting, and extrapolates the test data subjectively.<ref name="Z-3-413" /> The assessment matters because the usual explosion protection for deflagrations does not work: vents open only after the explosion has started, while the peak pressure of a detonation arrives at once with the shock front.<ref name="Z-3-414" /> Protection against detonation therefore relies on preventing the flammable mixture, on geometries below the critical size, or on structures designed for the detonation loads.<ref name="Z-3-402">Zalosh (2002), p. 3-402.</ref><ref name="Z-3-411" /><ref name="Z-3-412" />

== See also ==
* [[Explosion venting]]
* [[Hydrogen safety]]
* [[Inerting]]
* [[Deflagration index]]

== Notes ==
<references />

== References ==
* 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, Office of Nuclear Reactor Regulation.
* Simmons, R. F. (2002). Premixed burning. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-144–1-154). National Fire Protection Association; Society of Fire Protection Engineers.
* Zalosh, R. (2002). Explosion protection. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-402–3-421). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Explosions & Industrial Fires]]

Smoke and heat exhaust ventilation

2026-11-02T17:14:08Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Smoke and heat exhaust ventilation''' ('''SHEV''') is the removal of hot, smoky gases from the upper part of a space in a fire, combined with the supply of cool, clean air into its lower part. The smoke can be driven out by its own buoyancy through roof or high-level vents, by fans, or by a combination of both.<ref name="CO-219">Cooper (2002b), p. 3-219.</ref> Natural ventilation and extraction, the two forms of SHEV, are two of the five techniques of [[smoke control]] listed in the International Fire Engineering Guidelines, next to zone pressurisation, dilution and containment.<ref name="IFEG-2.5-7">National Research Council of Canada et al. (2005), p. 2.5-7.</ref>

The aim is to keep a layer of clear air under the smoke for long enough. Cooper names three objectives. SHEV helps people escape by limiting the spread of smoke into escape routes, helps firefighters to enter and see the seat of the fire, and reduces damage from smoke and hot gases.<ref name="CO-219" /> The principle is a mass balance. If the system removes as much smoke as the fire plume feeds into the layer, the layer stops descending.<ref name="VR-6.7">Van Rossum (2027), §6.7.</ref>

Natural systems are typical for single-storey halls such as warehouses and for atria, while fans are used where wind, cool smoke or the building layout make natural venting impractical.<ref name="CO-219" /><ref name="CO-236">Cooper (2002b), p. 3-236.</ref> How well SHEV works depends on enough replacement air, on the temperature of the smoke and on the wind.<ref name="CO-223">Cooper (2002b), p. 3-223.</ref><ref name="CO-224">Cooper (2002b), p. 3-224.</ref><ref name="CO-226">Cooper (2002b), p. 3-226.</ref>

== Purpose and applications ==
Venting may happen by chance, for example when smoke escapes through windows broken by the fire. A designed SHEV system instead uses dedicated openings or fans, such as ceiling vents that open automatically when a fusible link melts.<ref name="CO-219" /> Cooper describes two typical building types. The first is a large, open hall with ceiling vents, often divided by curtains hung from the ceiling into separate sections. The second is an atrium with shops at ground and mezzanine level that open onto it.<ref name="CO-219" /><ref name="CO-220">Cooper (2002b), p. 3-220.</ref>

Large-volume spaces need this approach because smoke spreads through them freely. Barriers and pressure differences, the methods used in compartmented buildings, are not practical there.<ref name="MI-292">Milke (2002b), p. 4-292.</ref> The high ceiling also has a benefit: smoke becomes diluted and cooled as it rises and spreads, which lowers the hazard it poses.<ref name="MI-292" />

The Australian verification method distinguishes two systems. Automatic smoke exhaust systems extract smoke from the hot layer near the roof with fans. Automatic smoke-and-heat vents work on the same principle but rely on the buoyancy of the hot gases.<ref name="AB-C4-128">Australian Building Codes Board (2020a), Data Sheet C4, p. 128.</ref> For both, an adequate supply of make-up air at low level is an integral part of the system. The objective is to keep tenable conditions in the space and escape routes long enough for evacuation and to support fire brigade intervention.<ref name="AB-C4-128" />

== How it works ==
A fire produces a rising plume of hot gases. As the plume rises, it draws in large amounts of surrounding air, a process called [[entrainment]]. Most of what is seen as smoke is therefore entrained air.<ref name="MI-292" /> Entrainment increases the mass flow in the plume and the amount of smoke, while it also cools and dilutes the combustion products.<ref name="MI-293">Milke (2002b), p. 4-293.</ref> The plume feeds a hot layer under the ceiling, and the higher the plume has to rise before reaching the layer, the more smoke it delivers.<ref name="VR-6.7" />

The layer grows unless smoke is removed. The book gives a simple example: if a plume supplies 10 kg/s to the layer and the exhaust removes 8 kg/s, the mass of the layer grows by 2 kg every second.<ref name="VR-3.7">Van Rossum (2027), §3.7.</ref> In Zukoski's experiments, a hood above a fire kept the smoke layer at a constant depth by extracting exactly the mass flow that the plume delivered at the layer interface. The book compares this with a SHEV system that holds a stationary smoke layer.<ref name="VR-6.7" />

[[File:fse-smoke-and-heat-exhaust-ventilation-v1.svg|thumb|upright=1.5|alt=Hall cross-section: fire plume rising into a smoke layer under the roof, smoke leaving through roof vents, fresh air entering through low inlets.|Principle of smoke and heat exhaust ventilation: the smoke removed through the roof vents balances the mass flow that the plume delivers to the layer, while replacement air enters low down. Based on Cooper (2002b), pp. 3-219 and 3-223, and Van Rossum (2027), §6.7.]]

Ceilings are often divided into [[smoke reservoir|smoke reservoirs]] by curtains hanging from the roof, so that the smoke stays in the curtained section above the fire. NFPA 204, as cited by Cooper, recommends that these curtains reach down at least one fifth of the height between the fire and the ceiling. The ceiling jet below the ceiling is about half that deep, and the layer must be thicker than the jet.<ref name="CO-234">Cooper (2002b), p. 3-234.</ref> If the smoke layer drops below the bottom of the curtains, smoke flows into the neighbouring sections.<ref name="CO-229">Cooper (2002b), p. 3-229.</ref>

For smoke to leave at the top, the same amount of air must enter at the bottom. Cooper's design method assumes large inlets near the floor, with a total area at least 1.5 to 2 times the area of the open ceiling vents.<ref name="CO-223" /> This [[replacement air]] should be clean, enter below the smoke layer and arrive slowly, at about 1 m/s near the plume. Fast inflow can deflect the plume, increase entrainment and even make the fire burn faster.<ref name="MI-306">Milke (2002b), p. 4-306.</ref> In mechanical systems the make-up air supply must also be smaller than the exhaust. Otherwise the space is pressurised and smoke is pushed through openings into adjacent spaces.<ref name="MI-306" />

== Natural and mechanical systems ==
Natural SHEV uses the pressure created by a hot layer. Inside the layer the pressure at the ceiling is higher than outside, so smoke flows out through open roof vents. The deeper and hotter the layer, the larger this pressure difference.<ref name="CO-223" /> Cooper gives an example of a 3 m deep layer that is 100 K warmer than ambient air at 293 K. The pressure difference is then 9.0 Pa, the outflow velocity 2.7 m/s and the mass flow 2.4 kg/s per square metre of vent area.<ref name="CO-224" />

Hotter smoke does not always vent better. The flow through a vent of given size reaches its maximum when the absolute temperature of the layer is twice that of the ambient air, about 586 K. Above 1.3 times the ambient temperature, a rise of about 90 K, the flow stays within 80 % of this maximum, but cooler layers give much lower flows.<ref name="CO-224" /> The effective size of an opening is smaller than its geometric size, because the flow contracts as it passes through. Cooper and the International Fire Engineering Guidelines use a coefficient of 0.61. Manufacturers give their vents a similar factor, for example 0.65, and the product is the [[aerodynamic free area]].<ref name="CO-222">Cooper (2002b), p. 3-222.</ref><ref name="IFEG-2.5-7" /><ref name="VR-6.8">Van Rossum (2027), §6.8.</ref>

[[File:fse-smoke-and-heat-exhaust-ventilation-v3.svg|thumb|upright=1.5|alt=Curve of relative vent mass flow against layer-to-ambient temperature ratio, rising steeply to a maximum of 1 at a ratio of 2 and falling slowly after.|Relative mass flow through a natural roof vent as a function of the ratio of layer to ambient absolute temperature. Above a ratio of 1.3 the flow is within 80 % of its maximum. Based on Cooper (2002b), p. 3-224.]]

Wind can help or hinder natural vents. Flat roofs tend to have negative wind pressures whatever the wind direction. If the open inlets are on the windward side, the raised pressure inside increases the flow through the roof vents. If they are on the leeward side, the lowered pressure inside can disrupt the venting completely and even reverse the flow.<ref name="CO-226" /> For a low, square building, identical inlets on all four sides would avoid the problem.<ref name="CO-226" /> In Dutch practice, a natural system is required to take its air supply from two façades that are at least 90° apart.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref>

Mechanical SHEV uses fans to extract the smoke. Cooper recommends it where wind could seriously reduce the efficiency of natural vents and where the smoke is too cool for natural venting to be economical. It also suits spaces where ceiling vents are impractical, such as intermediate floors of multi-storey buildings.<ref name="CO-235">Cooper (2002b), p. 3-235.</ref><ref name="CO-236" /> The design concept is the same as for natural vents, but the fans must be selected for the temperature of the smoke they extract. An alternative is to push air in with fans below the layer and let the smoke escape through natural vents, which keeps the fans out of the hot gases.<ref name="CO-236" />

A single extraction point with too much capacity can draw clean air from below the layer as well as smoke. This effect, called [[plugholing]], wastes part of the capacity and is worse when the layer is thin.<ref name="VR-6.8" /><ref name="CO-228">Cooper (2002b), p. 3-228.</ref> It was first studied for natural vents, and later work showed that the same analysis applies to fans.<ref name="MI-305">Milke (2002b), p. 4-305.</ref> Remedies are more extraction points with less capacity each, a deeper smoke layer, or directing the suction horizontally rather than vertically.<ref name="VR-6.8" />

[[File:fse-smoke-and-heat-exhaust-ventilation-v4.svg|thumb|upright=1.5|alt=Left: one large extract point pulling a funnel of clean air through a thin smoke layer. Right: three smaller extract points removing only smoke.|Plugholing: one extraction point with too much capacity draws clean air through the smoke layer (left); several smaller points remove only smoke (right). Based on Van Rossum (2027), §6.8 and Cooper (2002b), p. 3-228.]]

== Design ==
Most designs are set up as a steady-state problem. The designer chooses a design fire, a minimum acceptable height of the smoke layer interface and the ceiling height, and then determines the vent area or fan capacity needed.<ref name="CO-234" /> The design fire must represent a realistic arrangement of combustibles in the building.<ref name="CO-234" /> Some Dutch methods describe fire growth by the time the fire takes to reach 2 m² rather than 1 MW; NEN 6093, the Dutch standard for SHEV systems, is an example.<ref name="VR-4.5">Van Rossum (2027), §4.5.</ref> Time-dependent problems with growing fires and vents opening one by one require a computer model such as LAVENT.<ref name="CO-234" />

The central quantity is the mass flow that the plume delivers at the underside of the smoke layer. For relatively large fires, the book uses the plume model of Thomas, in which the mass flow depends on the perimeter of the fire and the height the smoke rises:<ref name="VR-6.7" />

<math display="block">\dot{m}_\mathrm{p} = 0.188 \, W_\mathrm{f} \, z^{3/2}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>\dot{m}_\mathrm{p}</math> || mass flow in the plume at the underside of the smoke layer || kg/s
|-
| <math>W_\mathrm{f}</math> || perimeter of the fire || m
|-
| <math>z</math> || height from the floor to the underside of the smoke layer || m
|}

The book takes the rule for choosing a plume model from NEN 6093. The Thomas model is the more reliable choice when the mean flame length is shorter than the diameter of the fire; for longer flames, the Heskestad plume is preferred.<ref name="VR-6.7" /> Because the height appears to the power 1.5, a higher clear layer needs much more exhaust. In SHEV design, the plume mass flow is the smoke that must be removed to keep the layer at its design height.<ref name="VR-6.7" /><ref name="CO-236" />

[[File:fse-smoke-and-heat-exhaust-ventilation-v2.svg|thumb|upright=1.5|alt=Curves of plume mass flow against clear height from 1 to 8 m for fire perimeters of 8, 12 and 16 m, rising steeply with height.|Plume mass flow at the underside of the smoke layer according to the Thomas model, for three fire perimeters. Based on Van Rossum (2027), §6.7.]]

=== Worked example ===
Cooper uses wood pallets in 3 m by 3 m arrays as an example of stored goods.<ref name="CO-220" /> Such a fire has a perimeter of 12 m. Using the book's flame length relation, the Thomas model applies as long as the fire stays below about 3.3 MW.<ref name="VR-6.7" /> Suppose the hall is 7 m high and the smoke layer must stay 3 m deep, so 4 m above the floor. The plume then delivers 0.188 × 12 × 41.5 ≈ 18.1 kg/s to the layer. With a clear height of 3 m it would be 11.7 kg/s, and with 5 m it would be 25.2 kg/s.<ref name="VR-6.7" />

If the 3 m deep layer were 100 K warmer than the outside air, Cooper's result of 2.4 kg/s per square metre gives the vent area. Removing 18.1 kg/s then needs about 7.6 m² of open roof vents.<ref name="CO-224" /> According to Cooper's assumption, the inlets near the floor must have at least 1.5 to 2 times this area, about 11 to 15 m².<ref name="CO-223" /> US vent units range from 1.5 m² to 4.7 m² each, so this means two to six vents.<ref name="CO-228" /> If heat losses kept the layer cooler than assumed, the flow per square metre would fall and the vents would have to be larger.<ref name="CO-224" />

The example is a simplification. Real layer temperatures depend on heat losses to the roof and walls, and the assumption of no heat loss can overestimate them in spaces lower than about 10 m.<ref name="MI-302">Milke (2002b), p. 4-302.</ref> In zone models, the extraction capacity should ideally be larger than the supply capacity in volume terms, and more so the more the fire heats the air.<ref name="VR-9.3a">Van Rossum (2027), §9.3 (How a zone model works).</ref>

== Interaction with sprinklers ==
Combining SHEV with sprinklers has long been controversial. Sprinkler spray cools the smoke, so less smoke leaves through the vents than designed. At the time of Cooper's review, no generally accepted guidelines existed for designing combined systems.<ref name="CO-236" />

This [[sprinkler-vent interaction]] can also work the other way. The Australian Building Codes Board notes that a smoke vent or exhaust system may remove hot gases before sprinklers activate, which may affect especially early suppression sprinklers.<ref name="AB-B1-21">Australian Building Codes Board (2020a), Data Sheet B1, p. 21.</ref> Sprinkler failure also matters for fan systems. If the sprinklers do not limit the fire to its design size, a smoke exhaust system that depends on them can be overwhelmed or its components overheated.<ref name="AB-C4-132">Australian Building Codes Board (2020a), Data Sheet C4, p. 132.</ref>

A review of 34 position papers, summarised by Cooper, reached the following conclusions. Venting has no negative effect on sprinkler performance and does not change the activation of early sprinklers or the total number that open. A fire directly below a vent may delay the first sprinklers slightly, without significant effect.<ref name="CO-237">Cooper (2002b), p. 3-237.</ref> Venting does limit the spread of smoke, improves visibility for occupants and firefighters and reduces damage. If the sprinklers fail, venting remains valuable for manual firefighting.<ref name="CO-237" /> The review recommends early, grouped opening of vents and placing curtains over aisles rather than over storage.<ref name="CO-237" />

== Reliability and limitations ==
For performance-based design in Australia, the Board uses a typical reliability of 80 % for both smoke exhaust systems and smoke-and-heat vents, with a range of 70 to 85 %.<ref name="AB-C4-134">Australian Building Codes Board (2020a), Data Sheet C4, p. 134.</ref> This is lower than the 90 to 95 % it uses for wet pipe sprinkler systems in general and residential buildings.<ref name="AB-C2-71">Australian Building Codes Board (2020a), Data Sheet C2, p. 71.</ref> For vents, this value was adopted from exhaust systems because no separate data were available.<ref name="AB-C4-140">Australian Building Codes Board (2020a), Data Sheet C4, p. 140.</ref> Smoke-and-heat vents are sensitive to external conditions, fire size and the buoyancy of the smoke, so their efficacy may be relatively low.<ref name="AB-C4-135">Australian Building Codes Board (2020a), Data Sheet C4, p. 135.</ref>

SHEV is not useful everywhere. Dutch guidance on car parks notes that smoke and heat extraction systems, mostly providing up to ten-fold ventilation, can also spread fire and smoke within the car park, depending on where they are placed. They are basically only effective for clearing smoke once the fire is out. Such systems have not been used since 2012 because their added value was found to be virtually non-existent.<ref name="VG-38">Van der Graaf et al. (2021/2023), p. 38.</ref> The same guidance still lists ventilation and smoke and heat removal among the facilities that can improve an interior attack by the fire service.<ref name="VG-50">Van der Graaf et al. (2021/2023), p. 50.</ref> Outlets must be placed away from windows, air intakes and escape routes, so that the smoke does not harm the surroundings.<ref name="VG-40">Van der Graaf et al. (2021/2023), p. 40.</ref>

A common misconception is that a SHEV system removes all smoke from a space. A successful design removes the hottest and most hazardous smoke and slows or stops the descent of the layer, for example above eye level; it does not clear the space.<ref name="CO-237" /> Another is that more exhaust always helps; plugholing shows that a single large outlet can waste its capacity.<ref name="VR-6.8" /> SHEV also interacts with ordinary building services. Ordinary ventilation systems can strongly influence smoke spread, so it can be difficult to decide what they should do when the fire alarm operates.<ref name="VR-6.6" />

== See also ==
* [[Warehouse fire]]
* [[Thomas plume]]
* [[Atrium smoke control]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Cooper, L. Y. (2002b). Smoke and heat venting. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-219–3-242). National Fire Protection Association; Society of Fire Protection Engineers.
* Milke, J. A. (2002b). Smoke management in covered malls and atria. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-292–4-310). National Fire Protection Association; Society of Fire Protection Engineers.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Smoke Control]]

Evacuation of people with disabilities

2026-11-02T17:14:06Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''evacuation of people with disabilities''' covers the measures and design assumptions that allow people with a physical, sensory or cognitive limitation to leave a building, or reach a place of safety, during a fire. In fire safety engineering it means that these occupants are included in the escape design, with their own walking speeds, response times and need for help.<ref name="Proulx-3-359">Proulx (2002), p. 3-359.</ref><ref name="ABCB-D1-260">Australian Building Codes Board (2020a), Data Sheet D1, p. 260.</ref> Proulx describes the aim as "egressibility": everyone should be able to leave a building or reach an area of safety, though not necessarily by the same route or in the same way.<ref name="Proulx-3-360">Proulx (2002), p. 3-360.</ref>

People with disabilities are a sizeable part of the population, and many limitations are temporary, such as a recent injury or pregnancy.<ref name="Averill-169">Averill et al. (2005), p. 169.</ref> Measured horizontal walking speeds of people with a mobility impairment range from about 0.57 m/s with a walking frame to 0.95 m/s without a walking aid, compared with 1.25 m/s for people without a disability. On stairs the difference is larger still.<ref name="Proulx-3-360" /><ref name="Proulx-3-361">Proulx (2002), p. 3-361.</ref> No single measure solves the problem. Refuge areas, lifts, evacuation devices, visual alarms, trained helpers and staff-assisted evacuation in care buildings are used in combination.<ref name="Proulx-3-360" /> The evacuation of the World Trade Center towers in 2001 showed both the value and the limits of these measures.<ref name="Averill-169" />

== Who is affected ==
Over the last decades, building regulations in developed countries have increasingly required new and refurbished public buildings to be accessible to everyone. Because most people with disabilities and many older people want to stay independent and go out, buildings are more and more used by occupants of mixed ability. Proulx notes that fire safety design has to take this into account.<ref name="Proulx-3-359" /> In the 1991 Canadian census, 15.5 % of the population reported some level of disability. The most common limitations were in mobility (53 % of disabled adults), agility (50 %), intellect (32 %) and hearing (24 %).<ref name="Proulx-3-359" />

A survey in the United States in 1994 estimated that 20.6 % of the population had some level of disability. Among people over 6 years old, 3.3 % used a wheelchair and 9.6 % used a cane, crutches or a walker for six months or longer.<ref name="Proulx-3-360" /> In Northern Ireland, disabled people made up 12 % of the mobile population out and about in the community. Of those who go out, 55 % do so every day, so disabled people regularly visit public buildings.<ref name="Proulx-3-360" /> Practice evacuations of high-rise offices in Canada suggested that about 3 % of occupants cannot use the stairs, because of conditions such as heart problems or recovery from surgery or accidents.<ref name="Bryan-3-335">Bryan (2002), p. 3-335.</ref>

In Australia, the 2020 data sheets for the Fire Safety Verification Method group disabilities into mobility, vision, hearing, cognition and memory, and speech. Based on a national survey from 2016, 4.1 % of the population are wheelchair users and 0.9 % have a severe vision limitation. A further 1.6 % have a severe hearing limitation and another 1.6 % a severe cognitive limitation. People with heart or lung disease can use the escape routes but may need rest breaks over long distances.<ref name="ABCB-D1-259">Australian Building Codes Board (2020a), Data Sheet D1, p. 259.</ref> Ageing reduces the abilities of many people, conditions such as obesity or arthritis reduce stamina or cause pain, and temporary conditions also affect people's usual abilities.<ref name="ABCB-D1-259" />

== Movement on the escape route ==
The most detailed data on the movement of disabled people come from the Fire SERT research group in Northern Ireland, which studied 155 participants. On a horizontal route the mean speed of all disabled participants who walked without help was 1.00 m/s. People with a locomotion disability walked at 0.95 m/s without an aid, 0.94 m/s with crutches and 0.81 m/s with a walking stick. With a walking frame or rollator the mean speed was 0.57 m/s.<ref name="Proulx-3-360" /> Electric wheelchair users moved at 0.89 m/s and manual wheelchair users at 0.69 m/s, against 1.25 m/s for participants without a locomotion disability.<ref name="Proulx-3-360" /> The book ''Introduction to Fire Safety Engineering'' gives the same values in its table of [[walking speed|walking speeds]] for target groups, with 0.58 m/s for a walking frame.<ref name="VR-7.4">Van Rossum (2027), §7.4.</ref>

[[File:fse-evacuation-of-people-with-disabilities-v1.svg|thumb|upright=1.5|alt=Bar chart of mean horizontal walking speeds: from 0.57 m/s with a walking frame to 1.25 m/s for people without a locomotion disability.|Mean speed on a horizontal route for disabled people moving without help, by mobility aid, compared with people without a locomotion disability. Based on Proulx (2002), p. 3-360.]]

The spread within each group is large. The range for all disabled participants ran from 0.10 to 1.77 m/s.<ref name="Proulx-3-360" /> In the same study, 16 of 28 manual wheelchair users needed help to complete a 50 m corridor with one 90-degree turn.<ref name="Bryan-3-336">Bryan (2002), p. 3-336.</ref> The researchers therefore suggested that users of different mobility aids should be treated separately in evacuation models. For some people, rest periods and the time needed to turn corners should also be included.<ref name="Bryan-3-337">Bryan (2002), p. 3-337.</ref>

Stairs are the main obstacle. Only 30 participants with a locomotion disability passed the screening test for the stair study. Their mean speed was 0.38 m/s going up and 0.33 m/s going down, compared with 0.70 m/s in both directions for people without a disability.<ref name="Proulx-3-361" /> Participants had to rest repeatedly, and more than 35 % went up one step at a time; 44 % came down that way. Nearly all used the handrail, 91 % going up and 94 % going down, and they were more confident going up than coming down.<ref name="Proulx-3-361" />

[[File:fse-evacuation-of-people-with-disabilities-v2.svg|thumb|upright=1.5|alt=Grouped bar chart of stair speeds: 0.38 m/s up and 0.33 m/s down with a locomotion disability, 0.70 m/s both ways without.|Mean speed on stairs going up and going down, for people with and without a locomotion disability. Based on Proulx (2002), p. 3-361.]]

Doors are a further barrier. In tests with a standard single door 750 mm wide, manual wheelchair users took three to four times longer than ambulant disabled people to pass through.<ref name="Bryan-3-338">Bryan (2002), p. 3-338.</ref> When the door had to be pulled open against a closing force of 60 or 70 N, five of the seven wheelchair users failed to get through.<ref name="Bryan-3-339">Bryan (2002), p. 3-339.</ref> In the Netherlands, door experiments at TU Delft in 2009 found the lowest flow when 5 % of the group could not walk unaided: 10 % below 135 persons per minute.<ref name="VR-7.4" />

A simple comparison shows what these speeds mean, using the values from the Fire SERT study.<ref name="Proulx-3-360" /> A corridor of 30 m takes 24 s at 1.25 m/s, 44 s in a manual wheelchair at 0.69 m/s and 53 s with a walking frame at 0.57 m/s. Going down the stairs, the mean speed of 0.33 m/s for people with a locomotion disability is less than half the 0.70 m/s of people without one.<ref name="Proulx-3-361" /> In a tall building these differences add up over many floors, and slower people can also hold up those behind them.<ref name="Averill-159">Averill et al. (2005), p. 159.</ref>

== Recognising the alarm and responding ==
A disability can also delay the earlier phases of escape. The Australian data sheets define design occupant profiles that show where adjustments are needed. People with a hearing limitation risk missing an alarm or spoken instructions, which visual or physical alarms can improve. People with a vision limitation do not see visual cues and find their way more slowly.<ref name="ABCB-D1-261">Australian Building Codes Board (2020a), Data Sheet D1, p. 261.</ref> People with a cognitive or memory limitation may not recognise or respond to an alarm at all and may need help to leave.<ref name="ABCB-D1-261" /> For mobility, the response may be delayed in some occupancies such as homes, travel is slower and needs more space, and the evacuation plan must cover vertical travel.<ref name="ABCB-D1-260" />

Proulx reports that people with an intellectual disability can learn to evacuate a building, but only with substantial, repeated training. Their walking speed may be similar to that of other occupants, although the delay before they start can be very long, especially at night.<ref name="Proulx-3-361" /><ref name="Proulx-3-362">Proulx (2002), p. 3-362.</ref> For people with a vision limitation, the Fire SERT studies found that LED exit signs were the easiest to see and read.<ref name="Proulx-3-361" />

== Strategies and measures ==
Proulx lists several options and notes that each has advantages and disadvantages.<ref name="Proulx-3-360" /> A refuge area is a protected space where people wait for help. It had limited success in tests but remains of interest when combined with other measures. Lifts that are safe to use in a fire are another option.<ref name="Proulx-3-360" />

[[File:fse-evacuation-of-people-with-disabilities-v3.svg|thumb|upright=1.5|alt=Section of a multi-storey building with a refuge area by the stair, a protected evacuation lift, helpers carrying an evacuation chair and a flashing alarm.|Measures that can be combined to help people with disabilities escape: a refuge area, a protected lift, an evacuation chair on the stairs and visual alarms. Based on Proulx (2002), p. 3-360.]]

A buddy system assigns one or more people to help or report a person with a limitation, although in practice it has not always worked. Evacuation devices for carrying people need little effort but must be kept nearby, and users need some training. Lists of people who may need help are often incomplete and out of date.<ref name="Proulx-3-360" />

Proulx concludes that no single option solves all problems, and that a combination of measures, tailored to the occupancy, is more likely to work.<ref name="Proulx-3-360" /> He also stresses that many people with a mobility impairment are still mobile to some degree. Some of them will try to evacuate whatever [[evacuation strategy]] is in place, so their movement capacity must be known.<ref name="Proulx-3-360" /> The International Fire Engineering Guidelines (2005) likewise state that people with disabilities can form a significant part of the occupants. They name refuges and the use of lifts as areas of growing interest and research.<ref name="IFEG-2.8-2">National Research Council of Canada et al. (2005), p. 2.8-2.</ref>

Lifts play a growing role. US building codes, including the New York City code, require accessible lifts as part of the [[means of escape]], to be used by the fire service to evacuate people with disabilities. These lifts need emergency power, access from a refuge area or horizontal exit unless the building is fully sprinklered, and a smoke-protected shaft.<ref name="Averill-35">Averill et al. (2005), p. 35.</ref> In Australia, Performance Requirement DP7 of the National Construction Code covers lifts intended to help occupants evacuate. Their type, number, location and fire isolation must suit the travel distance, the number and mobility of occupants, the waiting and travel time, the reliability of the lift and the emergency procedures.<ref name="ABCB-H-106">Australian Building Codes Board (2020b), p. 106.</ref> Such an [[evacuation lift]] is one of the enhancements the Australian handbook suggests for very tall buildings, together with protect-in-place or partial evacuation strategies.<ref name="ABCB-H-49">Australian Building Codes Board (2020b), p. 49.</ref>

== Care buildings ==
In hospitals and care homes, most residents may need help to leave. The Australian data sheets note that in hospitals, residential care and aged care the share of occupants with disabilities may approach 100 %, and staff assistance may be needed for most of them.<ref name="ABCB-D1-261" /> Bryan notes that in such buildings fire problems have been reduced by building design, staff training and preparation to protect occupants until evacuation is possible. Studies of health care staff found that nurses carried out their roles even at high personal risk.<ref name="Bryan-3-335" />

The usual response is [[progressive horizontal evacuation]]. Nursing staff first move patients horizontally to an adjacent smoke zone or fire compartment, and move them further only if the fire continues to develop.<ref name="IFEG-2.8-2" /> In New Zealand, C/VM2 (second edition, 2025) gives fixed times for this. In hospitals and rest homes, staff take 60 s to respond to the alarm and then 120 s per patient for each team of two staff. For early childhood centres it allows 60 s per child for each staff member.<ref name="MBIE-30">Ministry of Business, Innovation and Employment (2025b), p. 30.</ref> Three teams of two staff are considered a realistic maximum for shuttling patients to a place of safety.<ref name="MBIE-30" />

[[File:fse-evacuation-of-people-with-disabilities-v4.svg|thumb|upright=1.5|alt=Ward plan split into two fire compartments; pairs of staff move patients in beds through a fire door; a timeline shows 60 s response and 120 s per patient.|Horizontal evacuation in a hospital: pairs of staff move patients to the adjacent compartment, with 60 s to respond and 120 s per patient per team in C/VM2. Based on Ministry of Business, Innovation and Employment (2025b), p. 30.]]

The commentary to the first edition explains that the two minutes per patient assume an adjacent enclosure for horizontal evacuation. Trial evacuations of intensive care units, in which patients were strapped to stretchers and carried down stairs, needed three to eight minutes just to prepare each patient.<ref name="MBIE-2013-97">Ministry of Business, Innovation and Employment (2013), p. 97.</ref> C/VM2 also excludes some simplified analyses for these groups. Its check of radiant heat on people passing a fire along an escape route is not appropriate where occupants are likely to have impaired mobility.<ref name="MBIE-35">Ministry of Business, Innovation and Employment (2025b), p. 35.</ref>

== Experience from the World Trade Center ==
After the bombing of the World Trade Center in 1993, a study of 27 occupants with disabilities found an average evacuation time of 3.34 hours, ranging from 40 minutes to more than 9 hours. Most went down the stairs with help from other evacuees or emergency personnel.<ref name="Bryan-3-335" /> A review committee in 1995 called the evacuation of disabled people slow and arduous. Evacuation chairs and a buddy system for known mobility-impaired occupants were then introduced.<ref name="Averill-159" /><ref name="Averill-42">Averill et al. (2005), p. 42.</ref>

On 11 September 2001, about 6 % of the surviving occupants of the two towers, roughly 1,000 people, reported a limitation that affected their evacuation. Recent injuries and chronic illnesses were the most common; very few people used a wheelchair.<ref name="Averill-168">Averill et al. (2005), p. 168.</ref><ref name="Averill-169" /> Many evacuated with help from colleagues or strangers. In one case four people carried a wheelchair user down from a floor in the 60s in an evacuation chair.<ref name="Averill-148">Averill et al. (2005), p. 148.</ref> Of the survivors, 51 % in WTC 1 and 33 % in WTC 2 named injured and disabled people in the stairs as an obstacle. Two helpers supporting one person could fill the whole width of a stair.<ref name="Averill-159" />

In WTC 1 some mobility-impaired occupants were taken off the stairs to a holding floor low in the building, so that others could pass.<ref name="Averill-169" /> Firefighters found 40 to 60 people on the 12th floor and were helping about 20 of them down just before the collapse; it is unknown how many survived.<ref name="NIST-159">National Institute of Standards and Technology (2005), pp. 159–160.</ref> Mobility-impaired people did not appear to be over-represented among those who died below the impact floors. However, they were not consistently covered by the evacuation procedures, because some were left by colleagues and others had not told anyone of their limitation.<ref name="Averill-169" /> In WTC 2, 19 % of the mobility-impaired survivors used lifts, close to the 16 % of all survivors.<ref name="Averill-152">Averill et al. (2005), p. 152.</ref>

NIST concluded that, as far as possible, mobility-impaired occupants should have a means to evacuate on their own. It observed that current strategies generally required them to shelter in place and wait for help. New procedures should consider full building evacuation and could include fire-protected and structurally hardened lifts and motorised evacuation devices.<ref name="NIST-216">National Institute of Standards and Technology (2005), p. 216.</ref>

== Use in fire safety engineering ==
In a performance-based design, people with disabilities enter the analysis through occupant profiles and through the evacuation strategy. The Australian handbook divides occupants into three groups: prompt evacuation without assistance, slow evacuation without assistance and assisted evacuation.<ref name="ABCB-H-62">Australian Building Codes Board (2020b), p. 62.</ref> For the last group the pre-movement time is taken as infinite unless help arrives. These occupants rely on the fire brigade's search and rescue if they cannot leave.<ref name="ABCB-H-62" /><ref name="ABCB-H-123">Australian Building Codes Board (2020b), p. 123.</ref> The handbook also requires the fire safety strategy to document the evacuation strategy for all occupants, including people with disabilities.<ref name="ABCB-H-28">Australian Building Codes Board (2020b), p. 28.</ref>

The default Australian profiles assume 4 % of occupants with a mobility limitation, 1 % with a vision limitation and 1.5 % each with a hearing or cognitive limitation. These shares follow the general population and must be adjusted for buildings such as hospitals.<ref name="ABCB-D1-260" /><ref name="ABCB-D1-261" /> Evacuation software increasingly includes lifts, groups in wheelchairs or hospital beds, and staff who guide people out.<ref name="VR-8.3">Van Rossum (2027), §8.3.</ref> NIST recommended that egress models and design methods take account of human factors such as occupant size and mobility status.<ref name="NIST-215">National Institute of Standards and Technology (2005), p. 215.</ref>

== See also ==
* [[Occupant load]]
* [[Pre-movement time]]
* [[Human behaviour in fire]]
* [[Evacuation modelling]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Multi-sensor detector

2026-11-02T17:14:04Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''multi-sensor detector''' is an automatic fire detector that senses more than one fire phenomenon, or uses more than one operating principle, in a single unit.<ref name="IS-11-4">Iqbal and Salley (2004), p. 11-4.</ref> In the terms of the US fire alarm code NFPA 72, as quoted in NUREG-1805, this is called combination detection.<ref name="IS-11-4" /> Typical examples are a combined heat and smoke detector and a heat detector that combines a rate-of-rise and a fixed-temperature element.<ref name="IS-11-4" /> Units that combine thermal, photoelectric and ionisation sensing are also available.<ref name="IS-11-6">Iqbal and Salley (2004), p. 11-6.</ref>

The idea behind such detectors is that each type of sensor responds best to a different part of a fire. Ionisation detectors respond somewhat faster to open flaming fires, while photoelectric detectors respond faster to smouldering fires.<ref name="IS-11-5">Iqbal and Salley (2004), p. 11-5.</ref> Modern detectors also use response algorithms instead of a simple threshold, partly to reduce false and nuisance alarms and to improve fire signature matching.<ref name="SMC-4-22">Schifiliti et al. (2002), p. 4-22.</ref> These algorithms differ between detectors and are generally not published.<ref name="SMC-4-22" /> For fire safety engineering this creates a problem, because the actual response of each model is affected by its signal sampling algorithm.<ref name="SMC-4-22" /> In New Zealand, C/VM2 (second edition, 2025) avoids the problem by simulating every detector with fixed heat or smoke detector criteria, regardless of the make and model installed.<ref name="MBIE25-27">Ministry of Business, Innovation and Employment (2025b), p. 27.</ref>

== Fire signatures and why sensors are combined ==
A fire produces several signatures that a detector can sense. The SFPE Handbook lists electromagnetic radiation, invisible products of combustion smaller than 0.1 micrometre, visible smoke larger than 0.1 micrometre, a rapid change in temperature and a high temperature.<ref name="SMC-4-4">Schifiliti et al. (2002), p. 4-4.</ref> Each common detector type responds to only one of these.<ref name="SMC-4-4" /> Ionisation smoke detectors sense the invisible products, photoelectric detectors the visible smoke, rate-of-rise heat detectors a rapid change in temperature, and fixed-temperature heat detectors a high temperature.<ref name="SMC-4-4" />

[[File:fse-multi-sensor-detector-v1.svg|thumb|upright=1.5|alt=Grid of five fire signatures against detector types; each single detector has one mark, a combined smoke and heat unit has marks in several columns.|Fire signatures and the detector types that sense them; a combination detector covers more than one column. Based on Schifiliti et al. (2002), p. 4-4, and Iqbal and Salley (2004), pp. 11-4 to 11-6.]]

The smoke itself differs between fires. Early in a fire, heating of materials produces very small particles, and the particle size grows as the fire develops and particles coagulate.<ref name="IS-11-3">Iqbal and Salley (2004), p. 11-3.</ref> High-energy flaming fires produce large numbers of small particles, which suit the ionisation principle.<ref name="IS-11-5" /> Low-energy smouldering fires produce more of the larger particles, to which photoelectric detectors respond faster.<ref name="IS-11-5" /> Every listed smoke detector must still pass the same set of test fires.<ref name="IS-11-5" />

Full-scale tests show the difference in practice. Purser describes a flaming furniture fire in the lounge of a test house. There, the ionisation detector in the lounge operated after 0.5 minutes and the optical detector after 1.5 minutes.<ref name="Purser-2-137">Purser (2002), p. 2-137.</ref> The response of a light-scattering detector also depends on the colour of the smoke.<ref name="SMC-4-23">Schifiliti et al. (2002), p. 4-23.</ref> One such detector may alarm at an optical density of 0.029 m−1 for light grey smoke from a smouldering cotton wick but only at 0.15 m−1 for black smoke from a kerosene fire.<ref name="SMC-4-23" /> A unit that combines several sensors aims to cover more of these fire types with one device.<ref name="IS-11-4" /><ref name="IS-11-5" />

== Types and combinations ==
One combination adds a heat sensor to a smoke detector. NUREG-1805 notes that some ionisation and photoelectric detectors are made with a dual mode of operation, with a fixed-temperature thermal element built into the unit.<ref name="IS-11-5" /> A heat detector can itself combine two principles: a rate-of-rise element and a fixed-temperature element.<ref name="IS-11-4" /> According to the SFPE Handbook, most commercial rate-of-rise heat detectors have a fixed-temperature element as a backup. The reason is that many slowly developing fires cause high ceiling temperatures without ever reaching the rate of rise needed for an alarm.<ref name="SMC-4-18">Schifiliti et al. (2002), p. 4-18.</ref>

A worked example in the SFPE Handbook shows how the two elements of such a combined heat detector complement each other.<ref name="SMC-4-18" /> The detector has a fixed-temperature threshold of 57 °C, a rate-of-rise threshold of 11 °C per minute and a response time index of 42 m1/2s1/2.<ref name="SMC-4-18" /> It is installed at a spacing of 10.0 m under a 2.4 m ceiling, and the fire grows as a t-squared fire with different growth times.<ref name="SMC-4-18" /> For a fast fire with a growth time of 100 s, the rate-of-rise element responds after 53 s and the fixed-temperature element after 135 s.<ref name="SMC-4-18" />

For a slow fire with a growth time of 600 s, the order reverses: 521 s for the fixed-temperature element against 835 s for the rate-of-rise element.<ref name="SMC-4-18" /> Both respond at 454 s when the growth time is 509 s.<ref name="SMC-4-18" /> The crossover depends on the spacing: at a spacing of 25 m, the rate-of-rise element is faster only for growth times up to 228 s.<ref name="SMC-4-18" /> The handbook therefore advises considering the two elements separately when a system is designed or analysed.<ref name="SMC-4-18" />

[[File:fse-multi-sensor-detector-v2.svg|thumb|upright=1.5|alt=Two response-time curves against fire growth time crossing at 509 s; rate-of-rise is lower for fast fires, fixed-temperature for slow fires.|Response time of the two elements of a combined heat detector at 10 m spacing as a function of fire growth time; the curves cross at a growth time of 509 s. Based on Schifiliti et al. (2002), p. 4-18.]]

Analogue sensors go a step further. Instead of a simple alarm or no-alarm signal, they measure the conditions in the space and send this information to a computer-based fire alarm control unit.<ref name="IS-11-5" /><ref name="IS-11-6" /> They can report when a detector is too dirty to work properly or has become too sensitive.<ref name="IS-11-6" /> The system designer can adjust the sensitivity to the ambient environment or use an extra-sensitive setting for a high-value area.<ref name="IS-11-6" /> NUREG-1805 lists these sensors as photoelectric, ionisation, or combination thermal, photoelectric and ionisation units.<ref name="IS-11-6" />

[[File:fse-multi-sensor-detector-v3.svg|thumb|upright=1.5|alt=Block diagram: smoke and heat sensors in one detector send values over a loop to a panel that gives pre-alarm, alarm or dirty warnings.|An analogue multi-sensor detector sends measured values, not only an alarm state, to the control panel. Based on Iqbal and Salley (2004), pp. 11-5 to 11-6, and Australian Building Codes Board (2020a), pp. 96-97.]]

The Australian verification method data sheets describe the same capability at system level. An analogue addressable detector reports its exact location and also data on its own condition.<ref name="ABCB-96">Australian Building Codes Board (2020a), p. 96.</ref> This information can identify dirty detectors and give an early alert before the main alarm, as smoke or heat levels rise.<ref name="ABCB-97">Australian Building Codes Board (2020a), p. 97.</ref> According to the data sheets, it may improve the efficacy of the system and reduce false alarms.<ref name="ABCB-97" />

Combining signals is also possible between separate detectors. In coincident detection, two or more detectors must operate before the alarm is recognised.<ref name="IFEG-2.7-2">National Research Council of Canada et al. (2005), p. 2.7-2.</ref> These can be two similar detectors at different locations, or two detectors of different types.<ref name="IFEG-2.7-2" /> A multi-sensor detector combines the signals within one unit, whereas coincident detection combines the signals of several units in the [[fire alarm system]].<ref name="IS-11-4" /><ref name="IFEG-2.7-2" />

== Carbon monoxide sensing ==
Gas sensing is a further possible input. The International Fire Engineering Guidelines list [[carbon monoxide]] concentration among the fire conditions that can be used to determine detector activation.<ref name="IFEG-1.7-5">National Research Council of Canada et al. (2005), p. 1.7-5.</ref> Carbon monoxide matters especially in smouldering fires. Purser notes that a smouldering fire in a dwelling can last for several hours, with a hazard that develops mainly from carbon monoxide.<ref name="Purser-2-140">Purser (2002), p. 2-140.</ref> The slowly thickening and irritant smoke may trigger detectors before the situation becomes dangerous.<ref name="Purser-2-140" />

A Dutch study by the Netherlands Institute for Public Safety (NIPV) deals with indoor car parks for electric vehicles. It recommends a fire detection system with smoke and/or heat detectors together with a gas detection system for carbon monoxide.<ref name="NIPV-37">Van der Graaf et al. (2021/2023), p. 37.</ref> It cites a test of several detectors in which carbon monoxide detectors were the fastest to alert.<ref name="NIPV-37" /> The same report notes that in the Netherlands carbon monoxide detectors are in use, often in connection with LPG, but not as a form of fire detection.<ref name="NIPV-37" /> Dutch standardisation and certification are not geared to this use either.<ref name="NIPV-37" />

For gas sensing detectors there is no accepted method to calculate activation times, according to the International Fire Engineering Guidelines (2005).<ref name="IFEG-2.7-6">National Research Council of Canada et al. (2005), p. 2.7-6.</ref> The same chapter does give calculation methods for heat, smoke and flame detectors.<ref name="IFEG-2.7-2" /> Carbon monoxide is also one of the main asphyxiant gases in flaming furniture fires, together with hydrogen cyanide.<ref name="Purser-2-140" />

== Unwanted alarms and reliability ==
One reason to combine sensors and signal processing is to reduce [[unwanted fire alarms]]. The SFPE Handbook states that detector algorithms are used to reduce false and nuisance alarms and to improve fire signature matching.<ref name="SMC-4-22" /> NUREG-1805 describes analogue sensor systems as essentially free of false alarms under the conditions normally found in a building.<ref name="IS-11-6" /> It adds that matching the detector's sensitivity to the environment is part of the design of a stable smoke detection system.<ref name="IS-11-5" />

Processing a signal takes time. The International Fire Engineering Guidelines list detector signal interrogation, verification and processing, and the time for coincident detector operation, among the delays between activation and notification.<ref name="IFEG-1.7-6">National Research Council of Canada et al. (2005), p. 1.7-6.</ref> These delays must be added to the activation time when the time to warn occupants is estimated.<ref name="IFEG-1.7-6" />

Complexity has its own risks. The Australian data sheets note that addressable systems depend on site-specific software and programming, which introduces the risk of software or programming errors.<ref name="ABCB-96" /> They also note that combining different system types adds complexity, which should be considered when estimating the effectiveness of the system.<ref name="ABCB-97" />

== Use in regulation and design ==
Some prescriptive codes accept combination detectors as an alternative to single-sensor detectors. The New York City building code applied to the World Trade Center required, in a Class E fire alarm system, ceiling-mounted ionisation smoke detectors or combination smoke/heat detectors at each elevator landing.<ref name="GJ-80">Grill and Johnson (2005b), p. 80.</ref> These detectors had to be placed directly above the call button.<ref name="GJ-80" />

Performance-based design treats combination detectors more simply. In New Zealand, C/VM2 (second edition, 2025) requires detectors to be simulated with fixed values, regardless of the make and model to be installed.<ref name="MBIE25-27" /> It gives separate criteria for a [[heat detector]] and for a [[smoke detector]].<ref name="MBIE25-27" /> The heat detector has a response time index of 30 m1/2s1/2 and an activation temperature of 57 °C, and the smoke detector alarms at an optical density of 0.097 m−1.<ref name="MBIE25-27" /> C/VM2 gives no separate criterion for multi-sensor detectors.<ref name="MBIE25-27" /> The International Fire Engineering Guidelines similarly allow a point smoke detector to be modelled as a heat detector that operates at a temperature rise of 13 °C, with a low response time index.<ref name="IFEG-2.7-5">National Research Council of Canada et al. (2005), p. 2.7-5.</ref>

== Limits ==
The main limitation for engineers is that the response of a multi-sensor detector is set by its internal algorithm. The SFPE Handbook notes that even if correlations between optical density and detector response were available, the actual response of each model would be affected by its signal sampling algorithm.<ref name="SMC-4-22" /> Because the algorithms are generally not published, accurate modelling is not possible without data for the particular detector.<ref name="SMC-4-22" /> Product performance tests can give a range of possible response, because the test standard sets minimum and maximum permissible performance.<ref name="SMC-4-23" />

For heat detectors with two elements, the SFPE Handbook advises analysing each element on its own, as in the worked example above.<ref name="SMC-4-18" /> For smoke detectors, the International Fire Engineering Guidelines note that there are few well-developed prediction methods and advise a conservative prediction.<ref name="IFEG-2.7-5" /> The data sheets of the Australian verification method credit the extra information from analogue detectors with better efficacy and fewer false alarms.<ref name="ABCB-97" /> Estimation methods are best used to compare how one detector's response changes with spacing or location, not to compare with egress times.<ref name="SMC-4-23" />

== See also ==
* [[Beam smoke detector]]
* [[Flame detector]]
* [[Aspirating smoke detection]]
* [[Detector activation time]]
* [[Voice alarm system]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Grill, R. A., & Johnson, D. A. (2005b). ''Fire protection and life safety provisions applied to the design and construction of WTC 1, 2, and 7 and post-construction provisions applied after occupancy'' (NIST NCSTAR 1-1D). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Purser, D. A. (2002). Toxicity assessment of combustion products. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-83–2-171). National Fire Protection Association; Society of Fire Protection Engineers.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)

[[Category:Detection & Alarm]]

Performance-based design

2026-11-02T17:14:01Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Performance-based design''' is an approach to fire safety design in which measures are chosen for a specific building and then shown by engineering analysis to meet stated objectives or performance requirements.<ref name="HB-12">Australian Building Codes Board (2020b), p. 12.</ref><ref name="IFEG-1221">National Research Council of Canada et al. (2005), p. 1.2-21.</ref> It is the alternative to following a prescribed set of rules. Building codes that allow this approach state what a building must achieve and leave open how it is achieved.<ref name="HB-12" /> In Australia, a design developed this way is called a Performance Solution, and it is unique to each situation.<ref name="HB-13">Australian Building Codes Board (2020b), p. 13.</ref>

Performance-based design grew out of the limits of prescriptive codes, which cannot cover emerging technologies and every combination of circumstances.<ref name="IFEG-021">National Research Council of Canada et al. (2005), p. 0.2-1 (Part 0, Australia).</ref> It is applied through [[fire safety engineering]]. The engineer and the stakeholders agree the objectives, [[fire scenario|fire scenarios]], design fires and acceptance criteria. One or more trial designs are then analysed until one meets the criteria.<ref name="IFEG-122">National Research Council of Canada et al. (2005), p. 1.2-2.</ref>

The analysis may show that a design is at least equivalent to a design that complies with the prescriptive provisions, or that it meets the performance requirements directly.<ref name="IFEG-1221" /> The freedom of the approach comes with costs. The performance requirements of many codes are not quantified, the results are hard for authorities to check, and the design depends on the building being managed as assumed.<ref name="IFEG-023">National Research Council of Canada et al. (2005), p. 0.2-3 (Part 0, Australia).</ref><ref name="VR-2.1">Van Rossum (2027), §2.1.</ref>

== Principle and terminology ==
The core idea is to regulate the outcome rather than the means. The International Fire Engineering Guidelines (IFEG, 2005) explain that the intent of building regulations is to reduce risks to a level that the community tolerates.<ref name="IFEG-021" /> Traditional codes did this with prescriptive rules. Because such rules constrained designs in ways that did not always suit the building, codes became performance-based, to free designers, increase innovation and facilitate trade.<ref name="IFEG-021" /> The opposite approach is described in the article on [[prescriptive design]].

Performance-based codes are usually arranged as a hierarchy. In the Building Code of Australia described by the IFEG, objectives and functional statements state in general terms what the community expects and how a building could satisfy it.<ref name="IFEG-022">National Research Council of Canada et al. (2005), p. 0.2-2 (Part 0, Australia).</ref> Below them are the performance requirements, which every design must meet but which are not quantified.<ref name="IFEG-023" /> A design that complies with the deemed-to-satisfy (DTS) provisions is deemed to meet the performance requirements. A design that differs and is shown by analysis to comply was called an Alternative Solution.<ref name="IFEG-023" />

[[File:fse-performance-based-design-v1.svg|thumb|upright=1.5|alt=Pyramid: objectives, functional statements, performance requirements; below, two routes: deemed-to-satisfy solution and performance solution.|Hierarchy of a performance-based building code: the performance requirements are met either by deemed-to-satisfy provisions or by a performance solution shown to comply by analysis. Based on National Research Council of Canada et al. (2005), pp. 0.2-2–0.2-3 and Australian Building Codes Board (2020b), p. 13.]]

The terms differ between countries. The current National Construction Code (NCC) of Australia uses Performance Solution for what earlier editions called an Alternative Solution.<ref name="HB-16">Australian Building Codes Board (2020b), p. 16.</ref> The NCC is performance-based, and a design may use a Performance Solution, a DTS Solution, or a combination of both.<ref name="HB-12" />

In New Zealand, the Building Code sets objectives, functional requirements and performance criteria. Compliance can be shown with an acceptable solution, a verification method such as C/VM2, or another means.<ref name="CVM2-2">Ministry of Business, Innovation and Employment (2025b), p. 2.</ref> In the Netherlands, a risk-based alternative to the prescribed measures of the building regulations is usually called equivalence. It must achieve at least the same level of fire safety as the measure it replaces.<ref name="VR-1">Van Rossum (2027), §1.</ref>

== Reasons for using it ==
The main advantage is freedom of design. The Australian handbook for the Fire Safety Verification Method (FSVM) points out that a performance-based code imposes no obligation to use any particular material, component or construction method. The only condition is that the performance requirements are met.<ref name="HB-12" /> Performance Solutions are often flexible in achieving the outcomes and encourage innovative design and use of technology.<ref name="HB-13" /> A DTS Solution, by contrast, follows a set recipe of what, when and how to build.<ref name="HB-13" />

The IFEG list further benefits. They include a disciplined approach to fire safety design and a better understanding of how the parts of a fire safety system interact. Others are a method for comparing alternative solutions, monetary savings, and guidance for construction, commissioning and maintenance.<ref name="IFEG-033">National Research Council of Canada et al. (2005), p. 0.3-3 (Part 0, Australia).</ref> Fire engineering can also assess existing buildings when their use changes and propose solutions for upgrading them.<ref name="IFEG-033" />

Some buildings fall outside what prescriptive rules can handle. ''Introduction to Fire Safety Engineering'' notes that in the Netherlands, fire safety engineering is applied mainly to buildings that deviate from the norm. Examples are public transport terminals, tunnels, towers with a floor above 200 m, buildings with large atria, and large compartments outside the scope of the Dutch standards.<ref name="VR-1.5">Van Rossum (2027), §1.5.</ref> Even in ordinary buildings, an analysis may address a single aspect to represent reality better than the standard rules do.<ref name="VR-1.5" />

New Zealand gives an example of why a structured performance framework was introduced. A review of its Building Code found that the fire safety performance requirements were not clear or precise enough.<ref name="MBIE13-11">Ministry of Business, Innovation and Employment (2013), p. 11.</ref> The lack of a consistent approach led to disputes over designs, stifled innovation, more conservative designs, and lengthy and costly debates about compliance.<ref name="MBIE13-11" /> The response was C/VM2, a verification method with prescribed design scenarios for complex buildings. Simple buildings remained covered by acceptable solutions.<ref name="MBIE13-11" />

== The design process ==
The IFEG describe a typical process in five stages: preparing a fire engineering brief (FEB), carrying out the analysis, collating and evaluating the results, drawing conclusions, and preparing the report.<ref name="IFEG-112">National Research Council of Canada et al. (2005), p. 1.1-2.</ref> The FEB defines the scope of work and the basis for the analysis, as agreed by the relevant stakeholders.<ref name="IFEG-122" /> For a simple departure from a prescriptive provision it may be a short document, but for a large or complex project it can be a major one.<ref name="IFEG-122" />

[[File:fse-performance-based-design-v2.svg|thumb|upright=1.5|alt=Five boxes from fire engineering brief to report, with a loop from the conclusions back to the brief when a trial design fails.|The fire engineering process: if a trial design does not meet the criteria, the brief is revisited and a new trial design analysed. Based on National Research Council of Canada et al. (2005), pp. 1.1-2 and 1.2-2.]]

The FEB covers the scope, stakeholders, building and occupant characteristics, objectives, hazards and available measures, and the trial designs.<ref name="IFEG-1232">National Research Council of Canada et al. (2005), p. 1.2-32.</ref> It also records the non-compliance issues, the approaches and methods of analysis, the [[acceptance criteria]] and factors of safety, and the fire scenarios and design occupant groups.<ref name="IFEG-1232" /> In principle, the objectives, trial designs, methods and criteria are all agreed before the analysis starts.<ref name="IFEG-122" /> At this stage the [[design fire|design fires]] are described as far as possible without calculation, so the brief works as a [[Qualitative design review|qualitative review]] of the design.<ref name="IFEG-1229">National Research Council of Canada et al. (2005), p. 1.2-29.</ref> If analysis later shows that a trial design is unacceptable, the brief is revisited and a further trial design is developed.<ref name="IFEG-122" />

Australian practice uses a performance-based design brief (PBDB), followed by a performance-based design report.<ref name="HB-125">Australian Building Codes Board (2020b), p. 125.</ref> The report lists all variations from the DTS provisions and the performance requirements they affect. For each design scenario it states whether the proposal reached at least the safety level of a reference building that complies with the DTS provisions.<ref name="HB-125" /> In New Zealand, C/VM2 (second edition, 2025) describes an example process. It runs from concept design and FEB to a trial design tested against ten design scenarios, which is revised until all of them are satisfied.<ref name="CVM2-9">Ministry of Business, Innovation and Employment (2025b), p. 9.</ref>

== Comparative and absolute approaches ==
A key decision in the brief is how compliance will be shown. In a comparative approach, the fire safety of the trial design is compared with that of an identical building designed to the DTS or prescriptive provisions.<ref name="IFEG-1221" /> Both are analysed with the same assumptions, models and input data, and the trial design must be equivalent or better. This is often called an equivalence approach.<ref name="IFEG-1221" /> In an absolute approach, the results are compared directly with the objectives or performance requirements using the agreed acceptance criteria, without a benchmark design.<ref name="IFEG-1221" />

[[File:fse-performance-based-design-v3.svg|thumb|upright=1.5|alt=Two panels: left, trial design and DTS design analysed the same way and compared; right, trial design results compared with fixed criteria.|Comparative approach (left): the trial design must perform at least as well as an identical DTS design. Absolute approach (right): the results must meet agreed acceptance criteria. Based on National Research Council of Canada et al. (2005), p. 1.2-21.]]

The Australian FSVM requires a comparison with a reference building that complies with the DTS provisions.<ref name="HB-16" /> The handbook explains that the reference building indirectly defines the acceptable individual and societal risk levels, so its choice must be agreed by all stakeholders.<ref name="HB-46">Australian Building Codes Board (2020b), p. 46.</ref> An example shows why this matters. A Performance Solution for an apartment building extends the travel distance from an apartment door to a fire-isolated stair from the DTS limit of 6 m to 12 m.<ref name="HB-46" /> A fair reference building with the same footprint and number of occupants would need two stairs under the DTS provisions. The proposal therefore both lengthens the travel distance and removes an exit, which changes the scenarios that must be analysed.<ref name="HB-46" />

[[File:fse-performance-based-design-v5.svg|thumb|upright=1.5|alt=Two floor plans of 12 apartments: left, one central stair with 12 m travel distance; right, the same floor with two stairs and 6 m limit.|Proposed design with one stair and a 12 m travel distance (left) and a reference building of the same size that complies with the DTS provisions with two stairs (right). Based on Australian Building Codes Board (2020b), pp. 46–48.]]

Analyses also differ in method. Qualitative analysis is used only in a minority of cases, for example for a single non-compliance issue, and its logic must be documented.<ref name="IFEG-1221" /> Most cases need a quantitative approach.<ref name="IFEG-1221" /> Deterministic methods, which always give the same outcome for the same inputs, are the most common. Probabilistic methods assign reliabilities and frequencies and combine many scenarios, but they need much statistical data.<ref name="IFEG-1222">National Research Council of Canada et al. (2005), p. 1.2-22.</ref> ''Introduction to Fire Safety Engineering'' adds that both approaches have their own strengths and are often used together.<ref name="VR-1.4">Van Rossum (2027), §1.4.</ref>

== Acceptance criteria and safety margins ==
Acceptance criteria turn the general objectives into tests that an analysis can pass or fail. The IFEG require them to be appropriate to the objectives and the methods, numerical unless the analysis is qualitative, and realistic.<ref name="IFEG-1225">National Research Council of Canada et al. (2005), p. 1.2-25.</ref> Zero risk, for example, is not an appropriate criterion.<ref name="IFEG-1225" /> Typical parameters for protecting occupants are the expected risk to life, the margin between available and required escape time, smoke layer height, layer temperature, radiant heat, smoke optical density and carbon monoxide level.<ref name="IFEG-1225" />

For life safety, the most common deterministic test is a timeline comparison. The [[available safe egress time]] (ASET) is compared with the required safe egress time (RSET), and ASET must exceed RSET by a margin agreed in the brief.<ref name="IFEG-1222" /> In a comparative evaluation, the margin of the trial design must be the same as or greater than that of the DTS design.<ref name="IFEG-1222" /> ''Introduction to Fire Safety Engineering'' notes that many inputs to both times are uncertain, so the safety margin is important. It can be expressed either as a factor on RSET or as a minimum difference between ASET and RSET.<ref name="VR-7.1">Van Rossum (2027), §7.1.</ref> The more reliable the inputs, the smaller the margin needs to be.<ref name="VR-7.1" />

[[File:fse-performance-based-design-v4.svg|thumb|upright=1.5|alt=Timeline from ignition: RSET bar ends before ASET bar; the gap between them is the safety margin.|The available safe egress time must exceed the required safe egress time by an agreed safety margin. Based on National Research Council of Canada et al. (2005), p. 1.2-22 and Van Rossum (2027), §7.1.]]

Factors of safety should reflect the redundancy and reliability of the design, the methods and assumptions, the uncertainty and the consequences of a fire.<ref name="IFEG-1226">National Research Council of Canada et al. (2005), p. 1.2-26.</ref> They should be applied only at the end of a calculation, because applying them at every step leads to overly conservative results.<ref name="IFEG-1226" /> In a comparative evaluation, explicit factors are usually unnecessary because both designs are analysed in the same way.<ref name="IFEG-1226" /> Sensitivity studies test how the result changes when key inputs vary. Examples are a fire growth rate several times greater, partial failure of smoke exhaust, lower walking speeds or fewer available exits.<ref name="IFEG-1224">National Research Council of Canada et al. (2005), p. 1.2-24.</ref>

== Verification, competence and the building life cycle ==
Performance-based designs are harder to check than prescriptive ones. ''Introduction to Fire Safety Engineering'' points out that the justification can be so complex that the approving authority cannot easily verify it, unlike a rule-based design.<ref name="VR-2.1" /> Computational fluid dynamics is given as an example: authorities often have no specialist in it, so the integrity of the fire safety engineer must be relied upon.<ref name="VR-2.1" /> A second opinion is advised where results are unexpected or the party is unknown, but it is not economically feasible for every project.<ref name="VR-2.1" /> The book also notes incentives, such as obtaining a permit or saving on construction and installations, that can make certain outcomes attractive to the parties involved.<ref name="VR-2.2">Van Rossum (2027), §2.2.</ref>

Regulators therefore restrict who may use the formal methods. In Australia, the FSVM must only be used by a professional engineer or other suitably qualified person recognised as competent in fire safety engineering.<ref name="HB-6">Australian Building Codes Board (2020b), p. 6.</ref> C/VM2 is intended for professional fire engineers who are proficient in fire engineering modelling methods.<ref name="CVM2-7">Ministry of Business, Innovation and Employment (2025b), p. 7.</ref> The IFEG describe third-party review by an independent peer or specialist, including check calculations, as a constructive way to support approval.<ref name="IFEG-038">National Research Council of Canada et al. (2005), p. 0.3-8 (Part 0, Australia).</ref>

A performance-based design is specific to its building, occupants and site. The IFEG call this both a strength and a weakness. It allows the fire safety system best suited to the building, but changes to the building, its occupants or its site may require a new evaluation.<ref name="IFEG-037">National Research Council of Canada et al. (2005), p. 0.3-7 (Part 0, Australia).</ref> Such a re-evaluation might not have been needed if the broader DTS provisions had been followed.<ref name="IFEG-037" />

== Limits and criticism ==
The main weakness is the lack of quantified targets. The IFEG note that the level of safety intended by the Australian building code is not stated explicitly, so acceptance criteria must be developed for each design.<ref name="IFEG-023" /> Because the link between criteria and performance requirements is a matter of engineering judgement, it varies between practitioners and projects. Involving all stakeholders reduces this variation.<ref name="IFEG-023" /> The Australian handbook calls unquantified performance requirements a recognised limitation of performance-based codes and a barrier to their use.<ref name="HB-14">Australian Building Codes Board (2020b), p. 14.</ref>

The Netherlands faces the same problem. ''Introduction to Fire Safety Engineering'' states that Dutch regulations do not lay down how many casualties are acceptable or how large the risk of fire spread to a neighbouring plot may be.<ref name="VR-1.2">Van Rossum (2027), §1.2.</ref> There are also no risk-based limit values, and the limits in the rule-based approach cannot be translated into them unambiguously.<ref name="VR-1.4" /> Benchmarks themselves can be weak: the commentary to C/VM2 notes that many values in international codes are commonly accepted but have no technical basis.<ref name="MBIE13-12">Ministry of Business, Innovation and Employment (2013), p. 12.</ref>

Tools and data are a further limit. The IFEG were deliberately written as guidelines rather than as a mandatory code, because fire engineering was judged to lack the necessary validated tools and data.<ref name="IFEG-012">National Research Council of Canada et al. (2005), p. 0.1-2 (Part 0, Australia).</ref> Fire engineering designs therefore rely heavily on engineering judgement, and approving authorities need to understand the process to assess them.<ref name="IFEG-012" />

== See also ==
* [[Building regulations]]
* [[SFPE]]
* [[Ethics in fire safety engineering]]
* [[Sensitivity analysis]]
* [[Equivalency]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Fire Safety Design & Regulation]]

Phased evacuation

2026-11-02T17:13:59Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''phased evacuation''' is an [[evacuation strategy]] in which the occupants of a building leave in a planned sequence, starting with the people most at risk from the fire. In a [[high-rise building]], the fire floor and the floors next to it are evacuated first, while occupants on floors further away are placed on alert. They are evacuated progressively only if the fire continues to develop.<ref name="IFEG-2.8-2">National Research Council of Canada et al. (2005), p. 2.8-2.</ref> In hospitals, the first step is usually a horizontal move of patients to an adjacent smoke zone or fire compartment.<ref name="IFEG-2.8-2" /> The literature uses several names for the strategy, including staged evacuation, controlled selective evacuation and defend-in-place.<ref name="IFEG-2.8-2" /><ref name="Proulx-355">Proulx (2002), p. 3-355.</ref><ref name="Averill-160">Averill et al. (2005), p. 160.</ref>

A phased evacuation depends on people receiving different messages on different floors at different times. Such evacuations are therefore managed by trained personnel, using a [[voice alarm system]] or a zoned [[evacuation alarm system]] with intercommunication.<ref name="IFEG-2.8-2" /><ref name="ABCB-C3-97">Australian Building Codes Board (2020a), Data Sheet C3, p. 97.</ref> In simulations of a World Trade Center tower, a phased evacuation of three floors took about 4 minutes when occupants started immediately. A full evacuation of the fully occupied tower took 92 to 112 minutes.<ref name="Averill-xxxv">Averill et al. (2005), p. xxxv.</ref> The strategy differs from a [[stay-put strategy]], in which occupants remain where they are rather than move through corridors or stairs that may be filled with smoke.<ref name="IFEG-2.8-2" />

== Terminology and related strategies ==
The International Fire Engineering Guidelines (2005) describe the evacuation plan of a building first as controlled or uncontrolled. A controlled evacuation can then be full, zone or staged.<ref name="IFEG-1.8-6">National Research Council of Canada et al. (2005), p. 1.8-6.</ref> The guidelines recommend that this choice is made during the preparation of the fire engineering brief.<ref name="IFEG-2.8-4">National Research Council of Canada et al. (2005), p. 2.8-4.</ref> Proulx distinguishes between "uncontrolled total evacuation" and "controlled selective evacuation" of multistorey office buildings.<ref name="Proulx-355" /> The first type depends mainly on how evacuees defer to each other in the stairs. The second depends on the type of control imposed and on the availability of a voice communication system.<ref name="Proulx-355" />

The NIST investigation of the World Trade Center uses "phased evacuation" and notes that the strategy is also known as defend-in-place.<ref name="Averill-160" /> In the procedures of the towers, it was called a local evacuation. The people on the floors near the threat would move to three floors below the incident.<ref name="NIST-155">National Institute of Standards and Technology (2005), p. 155.</ref> Only under more severe circumstances would a full building evacuation be ordered.<ref name="NIST-155" />

Phased evacuation sits between two other approaches. In a total evacuation, all occupants are alerted at once and leave at the same time. In the "protect in place" concept for fully sprinklered residential buildings, occupants are encouraged to stay where they are.<ref name="IFEG-2.8-2" /> The concept of a fire safe refuge, a special fire compartment where people await rescue by the fire service, is a further variant.<ref name="IFEG-2.8-2" /> In New Zealand, C/VM2 (second edition, 2025) groups strategies in which occupants stay in place or move to a place of safety inside the building under "delayed evacuation strategies".<ref name="MBIE25-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref> Such a place of safety must meet the definition in C/VM2, which requires fire separations that withstand burnout and an automatic sprinkler system.<ref name="MBIE25-34" /><ref name="MBIE25-68">Ministry of Business, Innovation and Employment (2025b), p. 68.</ref>

== Why buildings use phased evacuation ==
In a tall building, all occupants cannot use the stairs at the same moment without long queues. Proulx describes a hypothetical fire on the tenth floor of a 15-storey office building in which everybody attempts to leave at once.<ref name="Proulx-356">Proulx (2002), p. 3-356.</ref> The occupants of the most threatened floors would then have to queue near those floors for several minutes.<ref name="Proulx-356" /> With the stair doors open on the fire floor, smoke would enter the stairs. Occupants descending from the floors above the fire would have to move through it.<ref name="Proulx-356" />

Field studies confirm that this happens in real fires. In two high-rise residential fires, about 96 percent of the occupants located above the fire floor moved through smoke.<ref name="Proulx-362">Proulx (2002), p. 3-362.</ref> Proulx explains this by the simultaneous movement of many occupants. People on the fire floor open the stair door, and occupants on upper floors open the same stair on their way down.<ref name="Proulx-362" /> Even a pressurised stair can then become contaminated, so that the stairs may be the most dangerous place in the building after the room of fire origin and the fire floor.<ref name="Proulx-362" />

A phased evacuation reduces this problem because fewer doors to the stairs are open at the same time. Klote notes that only a few stair doors may be open simultaneously when a staged evacuation plan is used.<ref name="Klote-4-286">Klote (2002), p. 4-286.</ref> Some pressurisation designs, such as stairs divided into compartments, only work when the number of open doors is limited. They suit a staged evacuation plan but not a densely populated building in which total evacuation is planned.<ref name="Klote-4-287">Klote (2002), p. 4-287.</ref>

There is also a practical and social reason. An uncontrolled total evacuation is increasingly seen as too upsetting for the many occupants who are not threatened.<ref name="Proulx-356" /> NIST concluded that, under certain circumstances, a phased evacuation moves the occupants most at risk to relative safety much faster and with less total impact on building tenants.<ref name="Averill-164">Averill et al. (2005), p. 164.</ref>

== How a phased evacuation is carried out ==
The sequence starts on the fire floor. In the World Trade Center towers, occupants were trained to move three floors below the fire floor, as instructed by the floor warden or the deputy fire safety director.<ref name="Averill-41">Averill et al. (2005), p. 41.</ref> A prepared public address announcement informed other floors that "an orderly evacuation of floors" was taking place. Occupants were asked to wait until their floor number was announced.<ref name="Averill-35">Averill et al. (2005), pp. 35–36.</ref> A separate announcement then told each floor that it was time to evacuate and to use the nearest stairs.<ref name="Averill-35" />

[[File:fse-phased-evacuation-v1.svg|thumb|upright=1.5|alt=Section through a tall building: the fire floor and its neighbours walk down a stair to three floors below; other floors are on alert.|Phased evacuation in a tall building: the fire floor and its neighbours move first to floors below the fire, while the other floors are placed on alert. Based on National Institute of Standards and Technology (2005), p. 155, and Averill et al. (2005), pp. 41 and 247.]]

The order of the floors depends on the expected spread of smoke. In a drill in a 21-storey office building in Ottawa in 1971, the presumed fire was on the third floor.<ref name="Proulx-358">Proulx (2002), p. 3-358.</ref> Because smoke was expected to move up the building, the procedure first cleared the fire floor. It then cleared all floors above it, starting with the 21st floor and progressing downwards.<ref name="Proulx-358" />

The Australian data sheets for the Fire Safety Verification Method describe how alarm systems support this sequence. Systems for phased evacuation commonly have two tones: an alert tone and an evacuation tone.<ref name="ABCB-243">Australian Building Codes Board (2020a), Data Sheet D1, pp. 243–244.</ref> The alert tone means "prepare for an evacuation". Occupants may then turn off equipment, pack up personal valuables and wait for instructions from wardens or announcements.<ref name="ABCB-243" /> The evacuation tone means "evacuate the building immediately".<ref name="ABCB-243" />

In a two-stage system, a building-wide alert sounds while the evacuation tone sounds on the fire floor. The system then cascades through the other floors, giving priority to those closest to the fire floor.<ref name="ABCB-258">Australian Building Codes Board (2020a), Data Sheet D1, p. 258.</ref> The data sheets include this alarm type for consistency with modern systems set up for phased evacuations.<ref name="ABCB-258" />

[[File:fse-phased-evacuation-v2.svg|thumb|upright=1.5|alt=Chart of floors against time: the fire floor switches to the evacuation tone first, neighbouring floors follow later, other floors keep the alert tone.|Two-stage alarm in a phased evacuation: the evacuation tone starts on the fire floor and cascades outwards, while the other floors hear the alert tone. Based on Australian Building Codes Board (2020a), Data Sheet D1, pp. 243–244 and 258.]]

In hospitals and care buildings, the phases are horizontal before they are vertical. Nursing staff first move patients to an adjacent smoke zone or fire compartment, and other floors are evacuated only if the threat continues.<ref name="IFEG-2.8-2" /> The deemed-to-satisfy provisions of the Australian National Construction Code (NCC), as quoted in the 2020 data sheets, reflect this. Patient care areas are divided into smoke compartments of no more than 500 m².<ref name="ABCB-247">Australian Building Codes Board (2020a), Data Sheet D1, p. 247.</ref> Horizontal exits must offer at least 2.5 m² of clear floor area per patient on the safe side.<ref name="ABCB-247" /> The data sheets describe these rules as consistent with a phased evacuation with staff assistance.<ref name="ABCB-247" />

[[File:fse-phased-evacuation-v3.svg|thumb|upright=1.5|alt=Plan of a hospital floor split into two smoke compartments; staff move patients in beds through a door in the dividing wall to the safe side.|Horizontal phase of a hospital evacuation: staff move patients into the adjacent smoke compartment before any vertical evacuation. Based on National Research Council of Canada et al. (2005), p. 2.8-2, and Australian Building Codes Board (2020a), Data Sheet D1, p. 247.]]

== Alarm, communication and management ==
A phased evacuation cannot work with a simple bell that sounds everywhere at once. NIST notes that voice communication systems in the United States were a response to phased evacuation.<ref name="Lew-150">Lew et al. (2005), p. 150.</ref> Occupants who were relocated or held inside the building needed instructions, at least until they were told to leave.<ref name="Lew-150" /> In the 1970s, shortly after New York City adopted Local Law 5 of 1973, discussions of phased evacuation led to high-rise voice communication systems and fire command centres from which incident commanders managed fires.<ref name="Lew-162">Lew et al. (2005), p. 162.</ref> Requirements for voice systems appeared in national standards in the mid-1980s, for example in the 1985 edition of NFPA 72F.<ref name="Lew-150" />

In Australia, multi-zoned evacuation systems are used where buildings are split into zones to facilitate phased evacuations.<ref name="ABCB-C3-97" /> Each zone is controlled individually and has its own speakers for evacuation tones and public address.<ref name="ABCB-C3-97" /> In Australia, clause E4.9 of the NCC, as quoted in the 2020 data sheets, requires an emergency warning and intercom system in buildings with an effective height of more than 25 m, among others.<ref name="ABCB-248">Australian Building Codes Board (2020a), Data Sheet D1, p. 248.</ref> The data sheets regard such systems as consistent with a managed, phased evacuation.<ref name="ABCB-247" />

The content of the message matters. According to the ''Introduction to Fire Safety Engineering'', a spoken message is more effective than a slow-whoop tone. Live announcements, by public address, in person or by telephone, are followed faster than automatic alarms.<ref name="VR-7.3">Van Rossum (2027), §7.3.</ref> The same source notes that a well-trained evacuation organisation greatly increases the chance that an alarm is taken seriously.<ref name="VR-7.3" />

Proulx regards a voice communication system with informative live messages as probably the best way to initiate a fast response.<ref name="Proulx-350">Proulx (2002), p. 3-350.</ref> In New Zealand, C/VM2 (second edition, 2025) reflects this in its design values. For hotel guests remote from the fire, the pre-travel activity time drops from 600 s with a standard alarm to 300 s with a voice alarm.<ref name="MBIE25-30">Ministry of Business, Innovation and Employment (2025b), p. 30.</ref>

Phased evacuation also demands more from building management. Proulx notes that an uncontrolled total evacuation is simple and needs no sophisticated systems or training.<ref name="Proulx-356" /> The systems and training for a controlled evacuation are expensive. They may also cause a long delay before the occupants who should leave are told to do so.<ref name="Proulx-356" /> The Australian data sheets therefore require the fire safety strategy to describe the evacuation management strategy and the performance it must achieve.<ref name="ABCB-248" /><ref name="ABCB-249">Australian Building Codes Board (2020a), Data Sheet D1, p. 249.</ref>

== Evacuation times ==
Proulx compares the two procedures for a 15-storey office building with 70 people on each floor and two exit stairs of 1,120 mm.<ref name="Proulx-355" /> In an uncontrolled total evacuation, about 9 minutes are needed to move all 980 occupants to the ground.<ref name="Proulx-355" /> A controlled selective evacuation clears the tenth-floor fire area, the floors above it and the ninth floor. Because of the time needed to start and control the sequence, it also needs about 9 minutes to move 490 people below the ninth floor.<ref name="Proulx-355" /> The advantage is therefore not speed for the building as a whole. It lies in sparing the people on the threatened floors from queuing near the fire and in smoke.<ref name="Proulx-356" />

The Ottawa drill of 1971 cleared the whole 21-storey building, about 2,100 people, over 30 minutes.<ref name="Proulx-358" /> The sequential procedure could, however, have been stopped after 6 minutes, once the fire floor and two adjacent floors were clear.<ref name="Proulx-358" /> An uncontrolled total evacuation of the same building through its two 1,040 mm stairs would have taken nearly 20 minutes. People from the highest floors might have queued for 14 of those 20 minutes.<ref name="Proulx-358" /><ref name="Proulx-359">Proulx (2002), p. 3-359.</ref>

NIST simulated a phased evacuation in a World Trade Center tower with three evacuation models. Two hundred people on each of floors 47 to 49 had to reach floor 45, three levels below the fire floor, floor 48.<ref name="Averill-247">Averill et al. (2005), p. 247.</ref> All 600 people were clear in 3.5 to 4 minutes if they started immediately. With random start delays between 0 and 10 minutes, the time rose to 11 to 11.5 minutes.<ref name="Averill-160" /> A full evacuation of a tower with 19,800 occupants took 92 to 112 minutes in the same models.<ref name="Averill-xxxv" /> NIST adds that some mid-rise buildings planned for phased evacuation can be fully evacuated at full capacity in less than 20 minutes.<ref name="Averill-257">Averill et al. (2005), p. 257.</ref>

[[File:fse-phased-evacuation-v4.svg|thumb|upright=1.5|alt=Range bars: phased evacuation of 600 people 3.5 to 11.5 min; full evacuation of 19,800 people 92 to 112 min, of 25,500 people 114 to 142 min.|Simulated evacuation times for a World Trade Center tower: a phased evacuation of three floors compared with full evacuations. Based on Averill et al. (2005), pp. xxxv and 160.]]

== Phased evacuation in fire engineering analysis ==
In a performance-based design, a phased evacuation changes the [[required safe egress time]] for each part of the building. The Australian data sheets state that, where the alarm is delayed to certain areas, these delays must be included in the analysis.<ref name="ABCB-243" /> On floors distant from the fire, access to the fire-isolated stair may be delayed to give priority to the floor of fire origin.<ref name="ABCB-246">Australian Building Codes Board (2020a), Data Sheet D1, p. 246.</ref> In evacuation calculations, waiting until instructed to leave counts as a delay at a node of the route.<ref name="ABCB-276">Australian Building Codes Board (2020a), Data Sheet D1, p. 276.</ref> With a two-stage alarm, the response on the fire floor resembles that of a single-stage system. The start of evacuation from other floors is delayed accordingly.<ref name="ABCB-258" />

C/VM2 (second edition, 2025) uses a notification time of 30 s for standard evacuation strategies. For non-standard strategies, the extended notification time must be taken into account.<ref name="MBIE25-29">Ministry of Business, Innovation and Employment (2025b), p. 29.</ref> For hospital wards, it allows 60 s for staff to respond and then 120 s per patient for each team of two staff to move patients to the next firecell.<ref name="MBIE25-30" /> This assumes that an adjacent enclosure is available for horizontal evacuation. Without one, the pre-travel activity time may need to be increased.<ref name="MBIE25-31">Ministry of Business, Innovation and Employment (2025b), p. 31.</ref>

== Limits and risks ==
Phased evacuation is designed for fires that stay within the design envelope of the life safety systems. NIST describes it as efficient for "routine" emergencies or emergencies that fit within that envelope.<ref name="Averill-160" /> It was not an appropriate strategy for an emergency on the scale of 11 September 2001.<ref name="Averill-164" /> NIST therefore recommended that tall buildings be designed to allow a timely full building evacuation when required, for example in major earthquakes, fires, explosions or terrorist attacks.<ref name="NIST-215">National Institute of Standards and Technology (2005), p. 215.</ref>

The strategy is also sensitive to fire spread. The Australian data sheets warn that rapid fire spread to several floors via the façade can defeat a phased approach. They ask for procedures for that case and for scenarios in which the alarm or intercom system fails.<ref name="ABCB-250">Australian Building Codes Board (2020a), Data Sheet D1, p. 250.</ref>

Communication can fail in practice. In the Ottawa drill, an incorrectly set switch delayed the first announcement until nearly 3 minutes into the drill.<ref name="Proulx-358" /> The announcement was ambiguous, and as many as 350 people left their floors before their turn. This delayed the intended earlier evacuation of floors 3, 4 and 2.<ref name="Proulx-358" />

Before the announcement, only 17 percent of the questioned occupants thought the alarm meant a fire, while nearly 60 percent thought of a circuit malfunction.<ref name="Proulx-358" /> On 11 September 2001, occupants of the second tower first heard that their building was secure and that they should return to their offices.<ref name="Averill-98">Averill et al. (2005), p. 98.</ref> A contradicting announcement followed at 9:02 a.m., one minute before the tower was struck.<ref name="Averill-98" />

Finally, people do not always wait for their turn. The ''Introduction to Fire Safety Engineering'' notes that group behaviour and authority shape the response to an alarm. People may conform to those around them or wait until a manager gives permission to leave.<ref name="VR-7.3" /> The same tendencies can also help: when enough people make the right decision, as in a well-practised evacuation organisation, the others tend to follow.<ref name="VR-7.3" />

== See also ==
* [[Evacuation lift]]
* [[Progressive horizontal evacuation]]
* [[Occupant load]]
* [[Pre-movement time]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Lew, H. S., Bukowski, R. W., & Carino, N. J. (2005). ''Design, construction, and maintenance of structural and life safety systems'' (NIST NCSTAR 1-1). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Deflagration

2026-11-02T17:13:57Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''deflagration''' is the propagation of a flame through a flammable mixture of fuel and air at a speed below the speed of sound in that mixture.<ref name="IS-F-20">Iqbal and Salley (2004), p. F-20.</ref><ref name="Z-3-406">Zalosh (2002), p. 3-406.</ref> The mechanism is thermal in nature: heat conducted from the reaction zone raises the temperature of the adjacent unburned gas until it reacts.<ref name="IS-F-20" /><ref name="S-1-148">Simmons (2002), p. 1-148.</ref> When a gas-air mixture or a dust cloud is ignited inside an unvented enclosure, the result is usually a deflagration.<ref name="Z-3-406" /> Most accidental [[explosion]]s are deflagrations.<ref name="Z-3-413">Zalosh (2002), p. 3-413.</ref>

Because the flame travels well below the speed of sound, the pressure rises almost uniformly throughout the enclosure while the flame spreads.<ref name="Z-3-410">Zalosh (2002), p. 3-410.</ref> This sets it apart from a [[detonation]], in which a shock wave runs through the mixture at supersonic speed and the pressure rises suddenly and unevenly.<ref name="Z-3-410" /><ref name="IS-15-3">Iqbal and Salley (2004), p. 15-3.</ref> In a closed vessel, a near-stoichiometric hydrocarbon-air deflagration can reach about 7 to 9 times the initial absolute pressure, enough to rupture a vessel or a room that is not designed for it.<ref name="S-1-147">Simmons (2002), p. 1-147.</ref><ref name="Z-3-406" /> Protection therefore aims to prevent the flammable mixture, to remove ignition sources, or to limit the pressure by venting, suppression or containment.<ref name="Z-3-402">Zalosh (2002), p. 3-402.</ref><ref name="IS-F-20" />

== Definition ==
NUREG-1805, citing Nolan (2000), defines a deflagration as a thermal mechanism by which an explosion reaction propagates through a flammable gas mixture. The reaction velocity is always less than the speed of sound in the mixture.<ref name="IS-F-20" /> Even so, a deflagration is capable of causing damage.<ref name="IS-F-20" /> Zalosh describes it in the same way as flame propagation at subsonic speed away from the ignition site.<ref name="Z-3-406" />

The difference from a detonation is usually expressed in speed. In a detonation the combustion wave runs through the reactants at supersonic speed, of the order of 2,000 m/s, led by a shock wave that compresses the reactants.<ref name="IS-15-3" /> In a deflagration the propagation rate stays below the speed of sound in air at 20 °C, approximately 330 m/s.<ref name="IS-15-3" /> NUREG-1805 also describes deflagrations as a slow subsonic flame front with a slow but uniform rise in pressure and temperature, and estimates detonation pressures at 2 to 4 times those of a deflagration.<ref name="IS-16-8">Iqbal and Salley (2004), p. 16-8.</ref>

[[File:fse-deflagration-v1.svg|thumb|upright=1.5|alt=Two tubes of gas mixture: top, a slow flame front with even pressure throughout; bottom, a fast front led by a shock with a sharp pressure jump.|A deflagration (top) moves below the speed of sound and the pressure rises almost evenly; a detonation (bottom) is led by a shock wave and the pressure jumps at the front. Based on Iqbal and Salley (2004), p. 15-3, and Zalosh (2002), p. 3-410.]]

The flame in a deflagration is a [[premixed flame]]: fuel and air are mixed before the flame reaches them.<ref name="Z-3-402" /> ''Introduction to Fire Safety Engineering'' gives a gas burner and a Bunsen burner as everyday examples of a premixed flame.<ref name="VR-4.3">Van Rossum (2027), §4.3.</ref> In a deflagration the same kind of flame does not stand still on a burner but moves through a cloud or a filled enclosure.<ref name="S-1-146">Simmons (2002), p. 1-146.</ref> Most fires, by contrast, burn as diffusion flames, in which fuel and air mix in the flame itself.<ref name="S-1-147" /><ref name="VR-4.3" />

== Conditions for a deflagration ==
NUREG-1805 names three elements that must be present at the same time: a flammable mixture of fuel and oxygen, usually from air; a means of ignition; and an enclosure.<ref name="IS-15-3" /> The fuel can be a gas, a vapour, a mist or a dust.<ref name="Z-3-414">Zalosh (2002), p. 3-414.</ref> Zalosh describes the hazard as the premixing of fuel and oxidant in a confined space before an ignition source appears; after ignition, the confinement prevents the free expansion of the combustion products.<ref name="Z-3-402" />

A mixture is flammable only between its lower and upper [[flammability limits]].<ref name="Z-3-402" /> A deflagration becomes possible once the gas concentration rises above the lower flammability limit.<ref name="IS-16-8" /> At the limits themselves a mixture burns with insignificant flame propagation, and between them there is a concentration at which flame propagation reaches its maximum.<ref name="IS-15-4">Iqbal and Salley (2004), p. 15-4.</ref> The terms "explosive limits" and "flammable limits" are used interchangeably; the word "explosive" refers to the pressure rise that a deflagration produces as it passes through a vessel.<ref name="IS-B-25">Iqbal and Salley (2004), p. B-25.</ref>

Combustible dusts behave in a similar way. They are an explosion hazard when the particles are smaller than about 100 to 400 μm, depending on the material, and when they are suspended in air between the lower and upper explosive limits.<ref name="Z-3-403">Zalosh (2002), p. 3-403.</ref> Typical lower explosive limits for fine dusts are 30 to 60 g/m³, roughly the same range as for many flammable gases and vapours expressed in the same units.<ref name="Z-3-403" />

== How the flame propagates ==
The basic property of a premixed flame is its burning velocity: the speed with which the flame moves, normal to its surface, into the unburned mixture next to it.<ref name="S-1-144">Simmons (2002), p. 1-144.</ref> Maximum laminar burning velocities at room temperature and atmospheric pressure are 41 cm/s for propane-air, 68 cm/s for ethene-air, 175 cm/s for acetylene-air and 320 cm/s for hydrogen-air.<ref name="S-1-146" /> For the alkanes and many other hydrocarbons the values lie between 40 and 47 cm/s.<ref name="Z-3-408">Zalosh (2002), p. 3-408.</ref>

The flame front moves much faster than its burning velocity suggests. The burned gas expands and pushes the unburned gas away from the ignition point, so an observer sees a flame speed that is significantly larger than the burning velocity.<ref name="Z-3-408" /> Before compression starts, the flame speed equals the burning velocity multiplied by the ratio of burned to unburned gas temperature.<ref name="Z-3-408" /> For many hydrocarbons near stoichiometric concentration this gives 350 to 440 cm/s.<ref name="Z-3-408" />

[[File:fse-deflagration-v2.svg|thumb|upright=1.5|alt=Circular vessel with a central ignition point, a spherical flame front, burned gas inside and arrows showing unburned gas pushed outwards.|A spherical flame in a closed vessel: the flame burns into the unburned mixture at the burning velocity, while the expanding burned gas pushes the mixture outwards, so the flame front moves several times faster. Based on Zalosh (2002), pp. 3-407 to 3-408.]]

Turbulence increases the speed further. It wrinkles the flame front and so enlarges its effective area; with propane, ethene and acetylene, the burning velocity roughly doubles as the Reynolds number of the flow increases to 40,000.<ref name="S-1-146" /> Theory suggests an increase by a factor of five at very high turbulence.<ref name="S-1-146" /> In a duct, the linear speed of the flame equals the burning velocity times the ratio of flame area to duct cross-section, and the flame area is typically two to three times the cross-section.<ref name="S-1-147" /> Higher initial temperature also helps: raising a propane-air mixture from 300 to 480 K doubles its maximum burning velocity.<ref name="S-1-146" />

If the hot products cannot vent, they push the flame and the unburned gas forward with increasing velocity, which creates more turbulence ahead of the flame.<ref name="S-1-147" /> In a long enough duct the flame front then acts as an accelerating piston, a shock wave forms, and the flame can become a detonation travelling at 1,500 to 3,000 m/s.<ref name="S-1-147" /> This process is the [[deflagration-to-detonation transition]]. NUREG-1805 notes that whether it can happen depends on both the mixture and its environment.<ref name="IS-B-26">Iqbal and Salley (2004), p. B-26.</ref>

== Pressure in a closed vessel ==
If the flame burns through the whole enclosure, the final pressure follows from the ideal gas law applied to the mixture before and after burning, since both fill the same volume.<ref name="Z-3-406" /> In words: the pressure ratio equals the ratio of the number of moles of gas times the absolute temperature, after and before the deflagration.<ref name="Z-3-406" /> More moles of gas or a hotter gas in the same volume both mean a higher pressure.

<math display="block">\frac{P_\mathrm{m}}{P_0} = \frac{n_\mathrm{b}\, T_\mathrm{b}}{n_0\, T_0}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>P_\mathrm{m}</math> || pressure at the end of a closed vessel deflagration (absolute) || Pa or bar
|-
| <math>P_0</math> || initial pressure in the enclosure (absolute) || Pa or bar
|-
| <math>n_\mathrm{b}</math> || number of moles of burned gas at the end of the deflagration || mol
|-
| <math>n_0</math> || number of moles of gas-air mixture initially in the enclosure || mol
|-
| <math>T_\mathrm{b}</math> || temperature of the burned gas at the end of the deflagration || K
|-
| <math>T_0</math> || initial temperature of the gas-air mixture || K
|}

The relation applies to gas and to dust deflagrations, because only the temperature and composition of the gas set the pressure.<ref name="Z-3-406" /> It assumes that the flame has burned through the whole enclosure and that no gas escapes.<ref name="Z-3-406" /> Using the adiabatic flame temperature at constant volume gives a conservative estimate, because real vessels lose heat to their walls.<ref name="Z-3-406" /><ref name="S-1-147" />

'''Worked example.''' Complete combustion of one mole of methane uses two moles of oxygen and forms one mole of carbon dioxide and two moles of water.<ref name="VR-3.3">Van Rossum (2027), §3.3.</ref> The nitrogen from the air does not react, so for a stoichiometric methane-air mixture the number of moles stays the same and ''n''b/''n''0 = 1.<ref name="VR-3.3" /><ref name="VR-4.2">Van Rossum (2027), §4.2.</ref> Simmons gives an initial temperature of 300 K and an adiabatic flame temperature of about 2,200 K for a stoichiometric hydrocarbon-air mixture.<ref name="S-1-147" /> With an initial pressure of 1.013 bar:

<math display="block">P_\mathrm{m} = 1.013\ \mathrm{bar} \times 1 \times \frac{2200\ \mathrm{K}}{300\ \mathrm{K}} \approx 7.5\ \mathrm{bar\ (absolute)}</math>

This is about 6.5 bar above atmospheric pressure (values rounded up, since they are loads).<ref name="S-1-147" /> Simmons states that the pressure can reach a little over 7 bar under adiabatic conditions, and that it will rupture a vessel explosively unless the vessel was designed for it.<ref name="S-1-147" /> At constant volume the flame temperature is 200 to 400 °C higher than at constant pressure, about 2,600 K for stoichiometric methane in the calculations shown by Zalosh.<ref name="Z-3-406" /> With that value the same relation gives about 8.8 bar absolute, close to the calculated value of about 9 atm in the handbook.<ref name="Z-3-406" /> The measured maximum for methane in Zalosh's table is 7.1 bar gauge, about 8.1 bar absolute.<ref name="Z-3-407">Zalosh (2002), p. 3-407.</ref>

The measured maximum pressures of twelve gases and vapours lie between 6.8 and 8.1 bar gauge, with two exceptions: acetylene at 10.6 bar gauge and ammonia at 5.4 bar gauge.<ref name="Z-3-407" /> For twenty dusts the range is 6.7 to 11.2 bar gauge, about the same as for gases.<ref name="Z-3-407" /> The highest pressures occur in slightly rich mixtures, at equivalence ratios of 1.1 to 1.2. Near the flammability limits, measured pressures are much lower than theory because the combustion is incomplete and heat is lost to the walls.<ref name="Z-3-406" /><ref name="Z-3-407" />

[[File:fse-deflagration-v3.svg|thumb|upright=1.5|alt=Horizontal bar chart of maximum deflagration pressure for twelve gases; most bars lie between 6.8 and 8.1 bar, acetylene 10.6, ammonia 5.4.|Measured maximum closed-vessel deflagration pressures of twelve flammable gases and vapours in air. Based on Zalosh (2002), p. 3-407, Table 3-16.3.]]

The pressure does not rise evenly in time. Zalosh gives a calculated example for a vessel of 7.15 m radius. The pressure was only about 2 psig (14 kPa) when the flame had reached half the vessel radius and 16 psig (110 kPa) at 80 % of the radius. The final pressure was 93 psig (641 kPa).<ref name="Z-3-408" /> Most of the pressure rise therefore occurs when the flame is close to the walls.<ref name="Z-3-408" /> The time needed to reach the maximum depends on the size of the enclosure and the fuel, and can extend to a few hundred milliseconds.<ref name="IS-15-3" />

A mixture that fills only part of an enclosure still produces a pressure rise, roughly in proportion to the fraction of the total gas mass that burns.<ref name="Z-3-408" /> In an example by Zalosh, 200 g of n-butane leaks into a filling room of 4 × 4 × 3 m and mixes stoichiometrically with part of the air.<ref name="Z-3-410" /> Filling the whole room would take 3.52 kg of butane, so about 5.7 % of the mass burns and the pressure rises by about 0.46 bar.<ref name="Z-3-410" /> This exceeds the strength of rooms in industrial facilities, even though 200 g spread evenly through the room would be too little to form a flammable mixture at all.<ref name="Z-3-410" />

== Deflagrations in buildings and fires ==
Explosions in open air are fundamentally different from confined explosions in vessels, pipework or buildings, and they require different countermeasures.<ref name="IS-15-5">Iqbal and Salley (2004), p. 15-5.</ref> In a confined explosion the overpressure comes from the expansion of the hot gases.<ref name="IS-15-5" /> When gases escape through an explosion vent, which may simply be a door or a window, the resulting turbulence can produce a second pressure peak.<ref name="IS-15-5" />

Fires can create the premixed mixture themselves. A backdraft occurs when fresh air reaches the fuel-rich smoke of an oxygen-starved fire, and it has the characteristics of a premixed fuel-air deflagration.<ref name="IS-15-4" /> A smouldering fire can produce enough unburned fuel and carbon monoxide to form a premixed atmosphere; if the mixture reaches its autoignition temperature, the smoke cloud deflagrates in a smoke explosion.<ref name="IS-15-4" /> Such explosions have been observed in smouldering fires involving polyurethane foam.<ref name="IS-15-4" />

''Introduction to Fire Safety Engineering'' describes the smoke gas explosion in the same terms. Flammable gases from decomposition or incomplete combustion are premixed with air and lie between the lower and upper explosion limits.<ref name="VR-4.10">Van Rossum (2027), §4.10.</ref> They can be ignited by autoignition or by an ignition source such as a spark from electrical equipment or glowing particles from the fire.<ref name="VR-4.10" />

In industry, dust deflagrations can cause severe secondary explosions. A deflagration inside process equipment can make the equipment fail and send a blast wave through the building, which raises settled dust from walls and floors.<ref name="Z-3-409">Zalosh (2002), p. 3-409.</ref> The flame emerging from the equipment often ignites this dust cloud, even if the cloud fills only a small part of the building.<ref name="Z-3-409" /> NUREG-1805 lists possible secondary explosions among the hazards of a deflagration, together with failure of equipment, ejection of flame and unburned product, and personal injury.<ref name="IS-15-3" />

== Protection ==
Explosion protection interrupts the sequence of events at one of three points. It prevents the formation of a fuel-air mixture, for example by ventilation or inerting, or removes ignition sources, for example by grounding equipment. The third option is to limit the pressure build-up.<ref name="Z-3-402" /> Inerting lowers the oxygen concentration below the limiting oxidant concentration. The NFPA 69 guideline quoted by Zalosh keeps the oxygen at least 2 volume per cent below that value when it is monitored continuously.<ref name="Z-3-403" />

[[Explosion venting]] discharges combustion gases during a deflagration so that the pressure stays below the damage threshold of the enclosure, which may be a room, a building or a piece of equipment.<ref name="Z-3-414" /> The vent opening is usually covered by blow-off panels, rupture discs or other engineered vent devices.<ref name="Z-3-414" /> The most effective vents open early, are as large as possible and let the combustion gases escape without restriction.<ref name="Z-3-414" /> Normal process openings are usually too small to prevent destructive pressures.<ref name="IS-15-3" />

Explosion suppression detects an incipient explosion and discharges an extinguishing agent from pressurised containers before the pressure reaches the damage threshold.<ref name="Z-3-417">Zalosh (2002), p. 3-417.</ref> Unlike venting, it releases no flame or fuel, so it suits indoor equipment and equipment holding toxic materials.<ref name="Z-3-417" /> A third option is deflagration pressure containment: the vessel and its connections are designed to withstand the maximum pressure of an internal deflagration.<ref name="IS-F-20" />

Venting and suppression both rely on the relatively slow, uniform pressure rise of a deflagration. Because vents open only after the explosion has started, they cannot cope with a detonation, whose peak pressure arrives instantly with the shock front.<ref name="Z-3-414" /> The sudden, non-uniform pressure rise of a detonation also rules out explosion suppression.<ref name="Z-3-411">Zalosh (2002), p. 3-411.</ref>

== Limits of simple estimates ==
The closed vessel relation gives an upper bound for an ideal case. Real vessels lose heat to their walls, so measured pressures are somewhat lower, especially for mixtures near the flammability limits.<ref name="S-1-147" /><ref name="Z-3-407" /> The rate of pressure rise depends strongly on turbulence, which varies among gases in a way that cannot be predicted in advance.<ref name="Z-3-409" /> Measured rates of pressure rise, expressed as the KG value for gases and the KST value for dusts, are therefore used in practice.<ref name="Z-3-409" /> The vent area needed depends on the size of the enclosure and on this rate of pressure rise.<ref name="Z-3-414" />

Connected vessels raise the pressure further. Burning in one vessel raises the pressure in a connected vessel; if the flame then enters the second vessel, it starts from a higher pressure and produces a higher final pressure.<ref name="S-1-147" /> Long ducts can accelerate the flame towards a detonation, which the simple relation does not cover.<ref name="S-1-147" />

== See also ==
* [[Vapour cloud explosion]]
* [[Battery energy storage system]]
* [[Deflagration index]]

== Notes ==
<references />

== References ==
* 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, Office of Nuclear Reactor Regulation.
* Simmons, R. F. (2002). Premixed burning. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-144–1-154). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Zalosh, R. (2002). Explosion protection. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-402–3-421). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Explosions & Industrial Fires]]

Walking speed

2026-11-02T17:13:54Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Walking speed''' is the speed at which people move along an escape route during an evacuation, measured along their line of travel in metres per second. Together with the capacity of doors and stairs, it determines the movement part of the evacuation time.<ref name="VR7.4">Van Rossum (2027), §7.4.</ref> ''Introduction to Fire Safety Engineering'' gives about 1.25 m/s for unhindered walking on a level floor, at fewer than 1 person per m².<ref name="VR7.4" /> The SFPE Handbook gives the same average unrestricted speed for densities below about 0.5 persons/m².<ref name="P346">Proulx (2002), p. 3-346.</ref>

Walking speed is not a fixed property of a person. It falls as a crowd becomes denser, it is lower on stairs than on a level floor, and it differs widely between people with and without mobility impairments.<ref name="NM370">Nelson and Mowrer (2002), p. 3-370.</ref><ref name="Bryan337">Boyce et al. (1999), as cited in Bryan (2002), p. 3-337.</ref> Smoke slows people down further, especially when it irritates the eyes.<ref name="Purser118">Purser (2002), p. 2-118.</ref> In the World Trade Center towers in 2001, the median evacuee moved at only about 0.2 m/s along the stairs.<ref name="Av129">Averill et al. (2005), p. 129.</ref> Engineers use walking speeds to calculate the travel part of the [[required safe egress time]].<ref name="VR7.4" /> Design methods in Australia and New Zealand cap the horizontal speed at 1.2 m/s.<ref name="VM2-32">Ministry of Business, Innovation and Employment (2025b), p. 32.</ref><ref name="DS278">Australian Building Codes Board (2020a), p. 278.</ref>

== Typical values ==
For a single person on a level floor without crowding, the sources give values between 1.2 and 1.25 m/s. ''Introduction to Fire Safety Engineering'' uses about 1.25 m/s for level, unhindered evacuation.<ref name="VR7.4" /> Proulx gives 1.25 m/s as the average unrestricted walking speed at densities below about 0.5 persons/m².<ref name="P346" /> The Australian Fire Safety Verification Method (FSVM) data sheets note that walking speed varies with age and gender. They consider a typical unimpeded speed of 1.2 m/s a reasonable upper limit for adults.<ref name="DS278" />

Simple hand calculations often use a round value. ''Introduction to Fire Safety Engineering'' allows 1 m/s for the walk from the last bottleneck to the place of safety, and its worked example uses 1 m/s throughout.<ref name="VR7.5">Van Rossum (2027), §7.5.</ref> Proulx gives an example in which the first person walks 18 m to a door at 61 m per minute, which is about 1 m/s.<ref name="P363">Proulx (2002), p. 3-363.</ref> Evacuation software lets the user choose. The EXIT89 model used by NIST offered an "emergency" horizontal unimpeded speed of 1.36 m/s and a "normal" speed of 0.91 m/s.<ref name="Av246">Averill et al. (2005), p. 246.</ref>

At low density, the person matters more than the crowd. Proulx notes that occupant characteristics such as age, limitations and grouping then decide the speed. A family group, for example, is likely to move at the speed of its slowest member, such as a child or an older person.<ref name="P346" /><ref name="P347">Proulx (2002), p. 3-347.</ref> The FSVM data sheets add other limits: congestion, the dimensions of the route, poor visibility and fatigue.<ref name="DS276">Australian Building Codes Board (2020a), p. 276.</ref>

== Effect of crowd density ==
When people walk close together, they slow down. ''Introduction to Fire Safety Engineering'' bases this on research by the London Transport Board in 1958, whose values are still used.<ref name="VR7.4LTB">London Transport Board (1958), as cited in Van Rossum (2027), §7.4.</ref> The capacity of an escape route keeps rising up to about 4.5 persons/m², because more of the available space is used even though each person walks more slowly. Above that density, capacity falls sharply and congestion develops.<ref name="VR7.4LTB" />

Nelson and Mowrer describe three regimes, derived from the work of Fruin, Pauls, and Predtechenskii and Milinskii.<ref name="NM370" /> Below about 0.54 persons/m², people move at their own pace, independent of others. Above about 3.8 persons/m², no movement takes place until the crowd has thinned. Between these limits, speed falls in a straight line as density rises.<ref name="NM370" /> Proulx gives similar limits, with a standstill at 4 to 5 persons/m², like a fairly crowded lift.<ref name="P346" />

=== The speed-density relation ===
The straight-line relation between speed and density is the basis of the hydraulic method for egress calculations.<ref name="NM370" /> The ''International Fire Engineering Guidelines'' describe this methodology as based on hydraulic flow and effective width.<ref name="IFEG283">National Research Council of Canada et al. (2005), pp. 2.8-3–2.8-4.</ref> It starts from a constant <math>k</math> for the type of route and subtracts a part that grows with the density:<ref name="NM370" />

<math display="block">S = k - a\,k\,D</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>S</math> || walking speed along the line of travel (on stairs, along the line of the treads) || m/s
|-
| <math>k</math> || speed constant for the route element: 1.40 for corridors, aisles, ramps and doorways; 1.00 to 1.23 for stairs || m/s
|-
| <math>a</math> || constant, 0.266 when speed is in m/s and density in persons/m² || m²/person
|-
| <math>D</math> || density of people on the route || persons/m²
|}

The relation is valid for densities between 0.54 and 3.8 persons/m².<ref name="NM370" /> At 0.54 persons/m², it gives about 1.2 m/s in a corridor, which is the maximum unimpeded speed of 1.19 m/s in the SFPE table.<ref name="NM370" /><ref name="Calc">Nelson and Mowrer (2002), p. 3-370; values calculated with their Equation 3.</ref> At 3.8 persons/m², it gives zero. The same equation is prescribed in New Zealand by C/VM2 (second edition, 2025) and recommended in the Australian FSVM data sheets, both with <math>k</math> = 1.4 and <math>a</math> = 0.266.<ref name="VM2-32" /><ref name="DS278" /> C/VM2 adds that a calculated horizontal speed above 1.2 m/s must be taken as 1.2 m/s.<ref name="VM2-32" />

[[File:fse-walking-speed-v1.svg|thumb|upright=1.5|alt=Line chart of walking speed against crowd density: flat up to 0.54 persons per square metre, then falling to zero at 3.8, for a corridor and two stairs.|Walking speed against crowd density for a corridor and two stair geometries, from S = k − akD with a = 0.266. Based on Nelson and Mowrer (2002), p. 3-370.]]

'''Worked example.''' A corridor holds a crowd at a density of 2.0 persons/m². With <math>k</math> = 1.40 m/s and <math>a</math> = 0.266, the speed is 1.40 − 0.266 × 1.40 × 2.0 = 0.655 m/s, or about 0.66 m/s.<ref name="Calc" /> The travel time for a [[travel distance]] of 30 m is the distance divided by the speed. It is 30 / 0.655 = 45.8 s, rounded up to 46 s. At a density of 0.54 persons/m², the same corridor takes 25 s, so the crowd adds about 21 s over 30 m.<ref name="Calc" />

Speed and density together set the flow along a route. The product of speed and density is the [[specific flow]], the number of people passing per second per metre of effective width.<ref name="NM371">Nelson and Mowrer (2002), p. 3-371.</ref> It peaks at 1.9 persons/m², where the speed in a corridor has fallen to about 0.69 m/s.<ref name="NM371" /><ref name="Calc" />

== Stairs ==
People move more slowly on stairs than on a level floor.<ref name="VR7.4" /> The distance travelled is measured along the "climb line", which runs along the front edges of the treads. Its length follows from the horizontal length of the stair and the height between floors.<ref name="VR7.4" /> Speed on a stair also falls as density rises, but it varies differently with density than on a level floor. The London Transport Board measured these descent speeds along the climb line.<ref name="VR7.4LTB" /> Nelson and Mowrer give maximum speeds along the line of travel for four stair geometries, at densities below 0.54 persons/m²:<ref name="NM370" />

[[File:fse-walking-speed-v2.svg|thumb|upright=1.5|alt=Side view of a stair between two floors with a dashed line along the tread edges forming the hypotenuse of a right triangle.|The climb line along the front edges of the treads is the travel distance on a stair; its length follows from the horizontal length and the storey height. Based on Van Rossum (2027), §7.4.]]

{| class="wikitable"
! Riser (mm) !! Tread (mm) !! <math>k</math> (m/s) !! Maximum speed (m/s)
|-
| 191 || 254 || 1.00 || 0.85
|-
| 178 || 279 || 1.08 || 0.95
|-
| 165 || 305 || 1.16 || 1.00
|-
| 165 || 330 || 1.23 || 1.05
|}

Shallower stairs with deeper treads are faster. Within the range of the table, the speed varies roughly with the square root of the ratio of tread to riser. There are not enough data to say whether this holds outside the range.<ref name="NM370" /> The same values appear in ''Introduction to Fire Safety Engineering'', in C/VM2 and in the FSVM data sheets.<ref name="VR7.4" /><ref name="VM2-33">Ministry of Business, Innovation and Employment (2025b), p. 33.</ref><ref name="DS279">Australian Building Codes Board (2020a), p. 279.</ref> Proulx states that relatively fit people at low density can average about 1.1 m/s down the stair slope, which is about 0.8 m/s horizontally.<ref name="P347" />

Real descents in tall buildings are much slower. In a hypothetical evacuation of a 15-storey office building, Proulx assumes a descent of 3.6 floors per minute, about 0.5 m/s along the stair slope.<ref name="P355">Proulx (2002), p. 3-355.</ref> In the north tower of the World Trade Center, the median occupant descended 1.2 floors per minute. With about 10 m of stair and landings per floor, this is about 0.2 m/s, including all stops for crowding, smoke, water and rest.<ref name="Av129" /> NIST concluded that the average surviving occupant descended at about half the slowest speed previously measured in non-emergency evacuations.<ref name="NIST215">National Institute of Standards and Technology (2005), p. 215.</ref>

== People with reduced mobility ==
Average speeds hide large differences between people. ''Introduction to Fire Safety Engineering'' lists mean speeds on a level floor for several groups. A person with a walking frame or rollator moves at 0.58 m/s, in a manual wheelchair at 0.69 m/s and with a walking stick at 0.81 m/s.<ref name="VR7.4" /> In an electric wheelchair the speed is 0.89 m/s, on crutches 0.94 m/s, and unaided with limited mobility 0.95 m/s, against 1.25 m/s for healthy people.<ref name="VR7.4" />

The SFPE Handbook reports measurements by Boyce, Shields and Silcock on people with disabilities who walked without assistance. Of 107 participants, the mean horizontal speed was 1.00 m/s, with a range from 0.10 to 1.77 m/s.<ref name="Bryan337" /> People using a walking frame or rollator averaged 0.57 m/s, and those without a locomotion disability 1.25 m/s.<ref name="Bryan337" />

[[File:fse-walking-speed-v3.svg|thumb|upright=1.5|alt=Horizontal bars of mean walking speed: 1.25 m/s without locomotion disability down to 0.57 m/s with a walking frame or rollator.|Mean horizontal walking speed of unassisted ambulant people by mobility aid, with the measured range. Based on Boyce et al. (1999), as cited in Bryan (2002), p. 3-337.]]

On stairs the differences were larger. People with a locomotion disability averaged 0.38 m/s going up and 0.33 m/s going down. Those without a disability averaged 0.70 m/s in both directions.<ref name="Bryan337" /><ref name="Bryan338">Boyce et al. (1999), as cited in Bryan (2002), p. 3-338.</ref> Some participants also needed rest periods and extra time to change direction.<ref name="Bryan337b">Bryan (2002), p. 3-337.</ref> These data are central to the [[evacuation of people with disabilities]].

Other groups also move more slowly. For people under the influence of alcohol or drugs, the British Standard BS 9999 suggests a 25 % shorter maximum travel distance or a 25 % lower walking speed.<ref name="VR7.4BS">British Standards Institution (2017), as cited in Van Rossum (2027), §7.4.</ref> The FSVM data sheets state that the speed-density equation applies to the general population and needs further reduction for older people and people with restricted mobility.<ref name="DS278" />

== Walking in smoke ==
Smoke reduces walking speed in two ways: it blocks the view, and irritant smoke makes it hard to keep the eyes open. Jin tested people in a 20 m corridor filled with either a highly irritant white smoke from wood cribs or a less irritant black smoke from kerosene.<ref name="Jin244">Jin (2002), p. 2-44.</ref> In non-irritant smoke, walking speed fell gradually as the smoke became denser. In irritant smoke, it fell very quickly over the same range, and subjects walked in a zigzag or step by step along the wall.<ref name="Jin244" />

Purser summarises these results in terms of optical density. In non-irritant smoke at an optical density of 0.5 per metre, walking speed fell from about 1.2 m/s to 0.3 m/s. People then behaved as if in total darkness, feeling their way along the walls.<ref name="Purser118" /> In irritant smoke, speed dropped to that of darkness at an optical density of only 0.2 per metre.<ref name="Purser118" />

Smoke also affects whether people walk on at all. The FSVM data sheets put the visibility below which people generally start to slow down at about 3 m, for non-irritant smoke and light-reflecting objects.<ref name="DS281">Australian Building Codes Board (2020a), p. 281.</ref> Studies of building fires found that about 30 % of people turned back rather than continue through smoke. On average they turned back at a visibility of about 3 m.<ref name="Purser118" /> This link between smoke and movement is the subject of [[visibility in smoke]].

Jin also tested adults walking into thick smoke with radiant heat near the far end of the corridor. At extinction coefficients of about 0.9 per metre, 14 of the 31 subjects turned back before the furthest point.<ref name="Jin248">Jin (2002), p. 2-48.</ref> For those who went on, walking speed fell in proportion to the distance from the entrance, and they reported feeling uneasy as they walked forward.<ref name="Jin248" />

Proulx reports that in Jensen's tests in dense smoke, people moved at about 0.2 to 0.4 m/s, whatever wayfinding system was used. Survivors of serious fires had moved on average only 10 m in heavy smoke, which at 0.2 m/s means 50 s of exposure.<ref name="P362">Jensen (1993), as cited in Proulx (2002), p. 3-362.</ref> The FSVM data sheets give reduction factors for irritant smoke, derived from Jin. At an extinction coefficient of 0.3 per metre the factor is 0.9, at 0.4 per metre it is 0.7, and at 0.5 per metre it is 0.3.<ref name="DS283">Australian Building Codes Board (2020a), p. 283.</ref> Below a visibility of 10 m with light-emitting signs, occupants are assumed to look for another route or a safer place and wait for help.<ref name="DS283" />

[[File:fse-walking-speed-v4.svg|thumb|upright=1.5|alt=Line chart: walking speed in irritant smoke falls from 1.1 m/s at extinction coefficient 0.2 to 0.3 m/s at 0.5 per metre.|Walking speed in irritant smoke against the extinction coefficient, as tabulated from Jin's experiments. Based on Australian Building Codes Board (2020a), p. 282.]]

== Use in design and limits ==
In design calculations, the walking speed turns a travel distance into a travel time. C/VM2 in New Zealand and the FSVM data sheets in Australia both divide the travel distance by the speed from the speed-density equation.<ref name="VM2-32" /><ref name="DS279" /> ''Introduction to Fire Safety Engineering'' uses the same step in its eight-step hand calculation, together with the flow capacity of the governing door or stair.<ref name="VR7.5" /> Evacuation simulations go further. Their occupant profiles can include walking speed and mobility, and models such as Simulex and Pathfinder adjust each person's speed to the distance from other people.<ref name="VR10.1">Van Rossum (2027), §10.1.</ref><ref name="VR10.3">Van Rossum (2027), §10.3.</ref>

The hydraulic approach rests on simple assumptions. The ''International Fire Engineering Guidelines'' list them: everyone starts at the same moment, nobody interrupts the flow with individual decisions, and nearly everyone can keep up with the group.<ref name="IFEG283" /> Movement times calculated this way may therefore be shorter than real ones.<ref name="IFEG283" />

The guidelines also advise basing the movement calculation on the design occupant group with the largest effect on the overall movement. The travel speed of that group is then used as the input to the model.<ref name="IFEG284">National Research Council of Canada et al. (2005), p. 2.8-4.</ref> The IFEG suggest testing the result with significantly lower travel speeds as a sensitivity study.<ref name="IFEG124">National Research Council of Canada et al. (2005), p. 1.2-24.</ref> Proulx warns against assuming that people behind a fast first walker will keep up a high flow, because they are rarely close behind.<ref name="P363" />

Data on walking in smoke are limited. A review by Fridolf and colleagues, reported in the FSVM data sheets, found no reliable and valid correlations for predicting walking speed in smoke.<ref name="DS280">Fridolf et al. (2018), as cited in Australian Building Codes Board (2020a), p. 280.</ref> Some measurements rest on small groups: the stair speeds of people without a disability in the study by Boyce and colleagues come from eight participants.<ref name="Bryan337" /><ref name="Bryan338" /> Real evacuations can also be far slower than design values, as the World Trade Center towers showed.<ref name="Av129" />

== See also ==
* [[Human behaviour in fire]]
* [[Hydraulic evacuation model]]
* [[Occupant load]]
* [[Exit capacity]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Jin, T. (2002). Visibility and human behavior in fire smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-42–2-53). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nelson, H. E., & Mowrer, F. W. (2002). Emergency movement. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-367–3-380). National Fire Protection Association; Society of Fire Protection Engineers.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Purser, D. A. (2002). Toxicity assessment of combustion products. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-83–2-171). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Dry pipe sprinkler system

2026-11-02T17:13:52Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''dry pipe sprinkler system''' is an automatic [[sprinkler system]] whose pipes contain compressed air or nitrogen instead of water, so that it can be installed where water in the pipes would freeze.<ref name="FL-72">Fleming (2002), p. 4-72.</ref> A special dry pipe valve holds the water back. The gas pressure in the pipes keeps this valve closed until a sprinkler opens in a fire.<ref name="FL-72" /> The escaping gas lowers the pressure, the valve opens, and water flows through the pipes to the sprinklers that have operated.<ref name="FL-72" /><ref name="IS-10-2">Iqbal and Salley (2004), p. 10-2.</ref>

The dry pipe system is one of the four basic types of sprinkler system, next to the [[Wet pipe sprinkler system|wet pipe system]], the preaction system and the [[deluge system]].<ref name="FL-72" /> Like a wet pipe system, it uses closed automatic sprinklers that open one by one in response to heat.<ref name="FL-72" /> The price for frost protection is a delay. Before water reaches a fire, the gas must first escape and the water must travel through empty pipes, a process that may take up to 60 seconds at the most remote point of the system.<ref name="HO-15">Hopkins et al. (2005), p. 15.</ref> In the United States, NFPA 13, the sprinkler standard of the National Fire Protection Association (NFPA), compensates for this delay by requiring a design area 30 % larger than for a wet pipe system.<ref name="HO-15" /> Dry pipe systems are also more expensive to install and maintain, and they are therefore used only where they are really needed, such as in unheated loading docks, cold stores and attics.<ref name="NO-F59">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-59.</ref><ref name="HO-14">Hopkins et al. (2005), p. 14.</ref>

== How it works ==
The overhead pipes of a dry pipe system are normally filled with pressurised air, or sometimes nitrogen.<ref name="HO-15" /><ref name="IS-10-2" /> The system uses the same closed automatic sprinklers as a wet pipe system. In the description of the US Nuclear Regulatory Commission (NRC), compressed air replaces the water of a wet pipe system. A dry pipe valve replaces its alarm check valve.<ref name="IS-10-2" /> The compressed air holds the dry pipe valve shut and so keeps water out of the system.<ref name="IS-10-2" />

When the heat of a fire opens a [[sprinkler head]], the air escapes through the open sprinkler and the pressure in the pipes falls.<ref name="HO-15" /> At a certain point the air pressure no longer outweighs the water pressure below the valve. This point is called the trip pressure, and it is typically 20 psi (138 kPa).<ref name="HO-15" /> The valve then opens, and the pressure of the water supply drives water into the pipes and out of the sprinklers that have opened.<ref name="HO-15" /><ref name="IS-10-2" /> Sprinklers that the fire has not heated stay closed, just as in a wet pipe system.<ref name="NO-F59" />

[[File:fse-dry-pipe-sprinkler-system-v1.svg|thumb|upright=1.5|alt=Two panels of a sprinkler system: left, air-filled pipes and a closed dry pipe valve; right, one sprinkler open over a fire and water flowing to it.|A dry pipe sprinkler system in stand-by (left) and after one sprinkler has opened (right). Based on Fleming (2002), p. 4-72 and Hopkins et al. (2005), p. 15.]]

A dry pipe valve can hold back water at a higher pressure with air at a lower pressure. The National Institute of Standards and Technology (NIST) describes the principle of the dry pipe valve used in automatic dry standpipes. The valve contains a differential clapper. The lower air pressure on the system side acts on a larger surface than the higher water pressure on the supply side.<ref name="HO-21">Hopkins et al. (2005), p. 21.</ref> When air is released, the force of the air falls until the water pressure can overcome it, and water then fills the system.<ref name="HO-21" />

[[File:fse-dry-pipe-sprinkler-system-v4.svg|thumb|upright=1.5|alt=Cross-section of a dry pipe valve: a clapper with a large top area under air pressure and a small seat area under water pressure.|Principle of a differential dry pipe valve: low air pressure acting on a large area holds back higher water pressure acting on a small area. Based on Hopkins et al. (2005), p. 21.]]

Each automatic sprinkler acts both as a heat detector and as a water distribution nozzle.<ref name="FL-85">Fleming (2002), p. 4-85.</ref> The most common sprinklers open when a soldered metal element melts or a liquid-filled glass bulb breaks at a set temperature.<ref name="IS-10-3">Iqbal and Salley (2004), p. 10-3.</ref> In a dry pipe system the sprinkler therefore still decides where water is discharged; the dry pipe valve only decides when water enters the pipes.<ref name="FL-72" />

== Where dry pipe systems are used ==
A wet pipe system needs enough heat to keep the water in its pipes from freezing.<ref name="HO-14" /> In the United States, NFPA 13 requires a minimum temperature of 40 °F (4.4 °C) at all times and in all places where wet pipe systems are installed.<ref name="HO-14" /> The NRC guidance therefore calls wet pipe systems unsuitable for automatic fire protection in unheated buildings.<ref name="IS-10-1">Iqbal and Salley (2004), p. 10-1.</ref> Dry pipe systems were developed for parts of buildings where water-filled pipes could freeze. They are usually installed where heating the building, or part of it, is impractical or undesirable.<ref name="HO-14" />

Typical locations named by NIST are loading docks, cold storage warehouses and attics.<ref name="HO-14" /> Building codes can offer the dry pipe system as one of several options. The New York City building code applied to the World Trade Center required wet pipe systems in all sprinklered areas. In areas subject to freezing, the system had to be protected by insulation, heat tracing or an antifreeze solution, or a dry pipe system had to be provided.<ref name="GJ-29">Grill and Johnson (2005b), p. 29.</ref>

Car parks that are open to the outside air face the same climatic problem. A report by the Netherlands Institute for Public Safety (NIPV) defines an open indoor car park as a naturally ventilated car park whose sides are all open to the outside air.<ref name="NIPV-48">Van der Graaf et al. (2021/2023), p. 48.</ref> It notes that weather influences make sprinkler systems less suitable for such car parks. Sprinklers nevertheless remain the most obvious way to contain a [[car park fire]] and to stop it spreading from one car to the next.<ref name="NIPV-49">Van der Graaf et al. (2021/2023), p. 49.</ref> In the Netherlands, a consultation on making sprinkler systems mandatory for certain indoor car parks started on 10 July 2021.<ref name="NIPV-6">Van der Graaf et al. (2021/2023), p. 6.</ref>

== Delay in water delivery ==
The main drawback of a dry pipe system is the delay between the operation of a sprinkler and the discharge of water.<ref name="HO-15" /> Fleming splits the total water delivery time into two parts. The trip time is the time needed for the air pressure to bleed down until the dry valve opens. The transit time is the time the water then needs to flow from the valve to the open sprinkler.<ref name="FL-85" /> The water delivery time starts when the first sprinkler opens.<ref name="FL-85" />

[[File:fse-dry-pipe-sprinkler-system-v2.svg|thumb|upright=1.5|alt=Two time bars: wet pipe discharges at time zero; dry pipe has a trip time segment and a transit time segment before discharge, together at most 60 s.|Water delivery time of a dry pipe system compared with a wet pipe system. Based on Fleming (2002), p. 4-85 and Hopkins et al. (2005), p. 15.]]

NFPA 13 limits this delay by limiting the size of the system. As described in 2002, the standard contained no maximum water delivery time if the system volume was no more than 750 US gallons (2,839 L).<ref name="FL-85" /> Larger systems were permitted only if water reached the test point within 60 seconds.<ref name="FL-85" /> The rule of thumb is that no more than 60 seconds should be tolerated, and that systems should be divided into smaller systems where necessary.<ref name="FL-85" /> NIST describes the 60-second limit as a performance-based alternative to the volume limit.<ref name="HO-15" />

The delivery time is checked in the field with an inspector's test connection. This connection sits at the most remote point of the system from the dry valve and has an orifice that simulates a single sprinkler.<ref name="FL-85" /> The time measured in this test is not a realistic indication of the delay in a real fire, for two reasons. The first sprinkler to open is likely to be closer to the dry valve, which shortens the transit time. If more sprinklers open, the air escapes faster, so that both the trip time and the transit time become shorter.<ref name="FL-85" />

Researchers at Factory Mutual Research Corporation derived a relation for the trip time. It grows with the dry volume of the system and with the ratio between the initial air pressure and the trip pressure. It falls with a larger flow area of open sprinklers.<ref name="FL-85" /> The relation also contains the air temperature. Transit time is harder to calculate and requires a numerical model in which the system is divided into sections.<ref name="FL-85" />

The delay adds to the time that the sprinkler itself needs to respond. The International Fire Engineering Guidelines allow suppression to be assumed to start at sprinkler activation only in limited cases. Examples are a low, small enclosure with a relatively small fire that one or a few sprinklers can extinguish.<ref name="IFEG-7">National Research Council of Canada et al. (2005), p. 2.7-7.</ref>

== Design area and water demand ==
A dry pipe system is expected to open more sprinklers than a wet pipe system, which discharges water immediately after the first sprinkler responds.<ref name="HO-15" /> In a sprinkler design, water is supplied for the sprinklers in a design area, which is chosen together with a design density (the depth of water applied per minute) from the hazard of the occupancy.<ref name="FL-74">Fleming (2002), p. 4-74.</ref> NFPA 13 contains special changes to the design area for dry systems.<ref name="FL-74" /> According to NIST, it requires the design area of a dry pipe system to be 30 % larger than that of a wet pipe system, with the same density:<ref name="HO-15" />

<math display="block">A_\mathrm{dry} = 1.3 \times A_\mathrm{wet}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>A_\mathrm{dry}</math> || design area of the dry pipe system || m²
|-
| <math>A_\mathrm{wet}</math> || design area required for a wet pipe system with the same hazard and density || m²
|}

The factor applies to the design area only. The density stays the same, so the required water flow rises in proportion to the area. The minimum flow of each sprinkler is still its area of coverage multiplied by the density.<ref name="FL-75">Fleming (2002), p. 4-75.</ref> The factor is a design rule of NFPA 13 in the United States; it is not derived from a calculation of the actual delay.<ref name="HO-15" />

=== Worked example ===
The design criteria for World Trade Center Building 7 (WTC 7), designed in the mid-1980s, show the rule in practice.<ref name="HO-64">Hopkins et al. (2005), p. 64.</ref> For Ordinary Hazard Group 1 areas with wet pipe systems, the criteria gave a density of 0.16 gpm/ft² (6.5 mm/min) over 1,500 ft² (139 m²).<ref name="HO-64" /> The loading berth had a dry pipe system with the same hazard and density, designed over 1,950 ft² (181 m²).<ref name="HO-65">Hopkins et al. (2005), p. 65.</ref> The rule reproduces this value:

<math display="block">A_\mathrm{dry} = 1.3 \times 139\ \mathrm{m^2} \approx 181\ \mathrm{m^2}</math>

At 6.5 mm/min, the minimum flow over the design area rises from about 910 L/min for the wet system to about 1,190 L/min for the dry system. The loading dock had a layout with about 12 m² (126 ft²) per sprinkler.<ref name="MA-76">McAllister et al. (2008), p. 76.</ref> At that spacing, the design area grows from 12 to 16 sprinklers.

[[File:fse-dry-pipe-sprinkler-system-v3.svg|thumb|upright=1.5|alt=Bar chart comparing wet and dry pipe design: area 139 and 181 m2, sprinklers 12 and 16, flow 910 and 1,190 L/min.|Design area, sprinklers in the design area and minimum flow for a wet and a dry pipe system with the same density of 6.5 mm/min. Based on Hopkins et al. (2005), pp. 64 to 65 and McAllister et al. (2008), p. 76.]]

The dry system also loses more pressure in its pipes. For friction loss calculations with the Hazen–Williams formula, steel pipe in dry and preaction systems is given a coefficient C of 100, against 120 for wet and deluge systems.<ref name="FL-76">Fleming (2002), p. 4-76.</ref> A lower C means a larger friction loss for the same flow. The values are chosen to simulate the expected interior roughness of aged pipe.<ref name="FL-76" />

== Related system types ==
{| class="wikitable"
|+ Comparison of sprinkler system types<ref name="FL-72" /><ref name="NO-F59" />
! Type !! Contents of the pipes !! What starts the water flow !! Main reason for use
|-
| Wet pipe || water under pressure || heat opens a sprinkler || most common type; heated buildings
|-
| Dry pipe || air or nitrogen under pressure || an opened sprinkler releases the air, the dry pipe valve trips || risk of freezing
|-
| Preaction || air, under pressure or not || detection opens the valve; heat opens sprinklers || concern for accidental discharge
|-
| Deluge || no water; open nozzles || detection opens the deluge valve || rapidly spreading, high hazard fires
|}

The preaction system was designed in part to overcome the weaknesses of the dry pipe system. It uses automatic sprinklers in pipes containing air, which may or may not be under pressure, together with a separate detection system in the same area.<ref name="NO-F59" /> The detectors open the valve, so that the pipes fill with water before a sprinkler opens. Preaction systems are meant to counteract the delay of dry pipe systems and to prevent damage from a broken pipe or sprinkler head.<ref name="NO-F59" />

A combined dry pipe and preaction system joins both principles. Its pipes contain air under pressure, and a detection system opens the dry pipe valves without a loss of air pressure. At the same time it opens air exhaust valves at the end of the feed main, so that the system normally fills with water before any sprinkler opens.<ref name="NO-F59" /> In a fire risk analysis for United States nuclear power plants, the detection-suppression event tree of NUREG/CR-6850 treats a preaction or dry pipe system in one example case. It notes that a preaction system needs two signals to succeed: an automatic detection signal and the opening of the sprinkler itself.<ref name="NW-P5">Nowlen et al. (2005b), p. P-5.</ref>

== Limitations ==
Dry pipe systems operate more slowly than wet pipe systems and are more expensive to install and maintain.<ref name="NO-F59" /> They are therefore used only where there is an absolute necessity, such as a risk of freezing.<ref name="NO-F59" /> The NRC report names two reasons to restrict them to low-temperature areas. The first is the delay between the release of the compressed air and the arrival of water. The second is internal corrosion and tuberculation of the pipes caused by alternating wet and dry periods.<ref name="IS-10-2" /> Corrosion or tuberculation in the sprinkler piping is also listed among the general causes of sprinkler failure.<ref name="IS-10-2" />

Sprinkler systems in general fail for reasons that also apply to dry pipe systems. The NRC report lists a closed water supply valve, an empty water tank, a fire pump that does not start, corroded or painted sprinkler heads and an inadequate water supply.<ref name="IS-10-2" /> Most sprinkler systems in office buildings are designed to control a fire, and full extinguishment may require the fire service.<ref name="HO-15" /><ref name="HO-16">Hopkins et al. (2005), p. 16.</ref> In a dry pipe system, the delay means that more sprinklers are expected to open before control is reached.<ref name="HO-15" />

== Examples ==
In the World Trade Center towers and WTC 7, wet pipe systems protected the main office and support spaces. Dry pipe and preaction systems were used on a limited basis for computer rooms and other special hazards, in less than 10 % of the floor area.<ref name="HO-14" /> In WTC 7, a dry pipe system protected two 12,000 gal (about 45,000 L) fuel oil storage tanks beneath the loading berths.<ref name="HO-128">Hopkins et al. (2005), p. 128.</ref> The fuel oil pump rooms also had dry pipe systems, designed as Ordinary Hazard Group 3 with 0.21 gpm/ft² (8.6 mm/min) over the entire pump room.<ref name="HO-65" />

A worked example in the SFPE Handbook shows a dry pipe system combined with another agent. A warehouse with crude rubber in piles has a dry pipe system with a maximum discharge of 0.2 gpm/ft² (8.1 mm/min) over 2,000 ft² (186 m²).<ref name="SH-141">Scheffey and Hickey (2002), p. 4-141.</ref> This density is not considered adequate for the hazard, so a fixed high-expansion foam system supplements the sprinklers.<ref name="SH-141" /> The foam system is arranged to start when the dry pipe valve trips because a sprinkler head has operated.<ref name="SH-143">Scheffey and Hickey (2002), p. 4-143.</ref>

== See also ==
* [[Fire water supply]]
* [[Hydraulic calculation]]
* [[Sprinkler design density]]
* [[Response time index]]

== Notes ==
<references />

== References ==
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* Grill, R. A., & Johnson, D. A. (2005b). ''Fire protection and life safety provisions applied to the design and construction of WTC 1, 2, and 7 and post-construction provisions applied after occupancy'' (NIST NCSTAR 1-1D). National Institute of Standards and Technology.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* McAllister, T. P., Gann, R. G., Averill, J. D., Gross, J. L., Grosshandler, W. L., Lawson, J. R., McGrattan, K. B., Pitts, W. M., Prasad, K. R., Sadek, F. H., & Nelson, H. E. (2008). ''Structural fire response and probable collapse sequence of World Trade Center Building 7'' (NIST NCSTAR 1-9, Vols. 1–2). National Institute of Standards and Technology.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nowlen, S. P., Najafi, B., Wyant, F. J., Forester, J., Kazarians, M., Kolaczkowski, A., Joglar, F., Funk, D., Hannaman, G., & Anoba, R. (2005b). ''EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Vol. 2. Detailed methodology'' (NUREG/CR-6850; EPRI 1011989). Electric Power Research Institute; U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.
* Scheffey, J. L., & Hickey, H. E. (2002). Foam system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-123–4-148). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)

[[Category:Fire Suppression]]

Available safe egress time

2026-11-02T17:13:50Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Available safe egress time''' ('''ASET''') is the time that people in a building have to escape safely during a fire. It runs from ignition until conditions in a given space become untenable, for example because smoke has made it too hard to see or the heat has become too great.<ref name="VR-7.1">Van Rossum (2027), §7.1.</ref><ref name="IFEG-0.5-2">National Research Council of Canada et al. (2005), p. 0.5-2 (Australia).</ref> ASET is one half of the central comparison in escape design. The other half is the [[required safe egress time]] (RSET), the time people need to notice the fire, respond and reach a place of safety. A design is acceptable when ASET is longer than RSET by a sufficient margin.<ref name="VR-7.1" />

[[File:fse-available-safe-egress-time-v1.svg|thumb|upright=1.5|alt=Timeline from ignition: a long ASET arrow above a chain of RSET steps (detection, alarm, recognition, response, movement) and a safety margin at the end.|ASET runs from ignition to the moment the first tenability limit is exceeded; RSET is the sum of the escape steps. The gap between them is the safety margin. Based on Van Rossum (2027), §7.1.]]

ASET describes the fire, not the people: it depends on how fast the fire grows, how quickly smoke fills the space and which limits are set for heat, smoke and toxic gases.<ref name="Purser-2-90">Purser (2002), p. 2-90.</ref> It is usually found with [[fire modelling]], ranging from simple zone models to computational fluid dynamics.<ref name="VR-8">Van Rossum (2027), §8 (Method for determining ASET).</ref> In New Zealand and Australia, the verification methods fix the tenability limits used to define ASET, measured 2 m above the floor. Visibility is usually the first limit to be exceeded.<ref name="MBIE-17">Ministry of Business, Innovation and Employment (2025b), p. 17.</ref><ref name="ABCB-2022-10">Australian Building Codes Board (2022), p. 10.</ref> Because many inputs are uncertain, a safety factor or safety margin is applied between ASET and RSET.<ref name="VR-7.1" />

== Definition and variants ==
The book ''Introduction to Fire Safety Engineering'' defines ASET as the time available for people to escape safely. The comparison with RSET can be made for a whole building or for a single room, and the available time must clearly be longer than the time needed.<ref name="VR-7.1" /> Escape is mostly prevented because hot smoke reduces visibility, although temperature and toxicity can also play a part.<ref name="VR-7">Van Rossum (2027), §7.</ref>

Purser describes the idea behind ASET as a race between two processes that run in parallel. The first is the developing hazard from the fire: ignition, fire growth and the spread of smoke and heat. Its assessment aims at the time when an occupant would receive an incapacitating exposure. The second is the escape process, from detection and warning to movement out of the building, which gives the time required for escape.<ref name="Purser-2-85">Purser (2002), p. 2-85.</ref> If the occupants are out before the fire becomes hazardous, with a reasonable margin of safety, the design can be considered to have succeeded.<ref name="Purser-2-85" /> The time available for escape depends on the fire dynamics and on the time to incapacitation of people exposed to smoke.<ref name="Purser-2-90" />

Written definitions differ slightly in their starting point. The International Fire Engineering Guidelines (2005) define ASET in the Australian version as the time between ignition and the onset of untenable conditions in a specific part of a building. The New Zealand version starts the clock at the start of established burning instead of ignition.<ref name="IFEG-0.5-2" /><ref name="IFEG-0.5-2-NZ">National Research Council of Canada et al. (2005), p. 0.5-2 (New Zealand).</ref> These guidelines also expand the abbreviation as available safe ''evacuation'' time; the meaning is the same.<ref name="IFEG-0.5-2" />

Current regulations tie the definition to a measurable criterion. In New Zealand, C/VM2 (second edition, 2025) defines ASET as the time between ignition of the design fire and the moment the first tenability criterion is exceeded in a specified room.<ref name="MBIE-17" /> The Australian Fire Safety Verification Method Standard 2022 uses the same wording.<ref name="ABCB-2022-10" /> ASET is therefore not one number for a building. It belongs to a particular space, fire location and set of criteria, and C/VM2 requires several fire locations to be checked to find the limiting case.<ref name="MBIE-18">Ministry of Business, Innovation and Employment (2025b), p. 18.</ref>

== Comparison with RSET ==
Many inputs of both ASET and RSET are uncertain, so a safety margin is important. The book gives two common forms of the comparison.<ref name="VR-7.1" /> The first multiplies RSET by a safety factor:

<math display="block">\mathrm{ASET} \geq f \cdot \mathrm{RSET}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\mathrm{ASET}</math> || available safe egress time, from ignition to the first exceeded tenability limit || s
|-
| <math>\mathrm{RSET}</math> || required safe egress time, from ignition until the occupants have left the space || s
|-
| <math>f</math> || safety factor (1.0 or more) || none
|}

The second form requires the difference ASET − RSET to be at least a fixed safety margin. Which margin is needed depends on several factors, of which the reliability of the inputs is probably the most important. The more reliable the input, the more certain the outcome and the smaller the margin needs to be.<ref name="VR-7.1" /> A sensitivity analysis shows whether variation in the input has a large effect on the results, and so whether the margin is sufficient.<ref name="VR-7.1" />

In the Netherlands, the standards NEN 6060 and NEN 6079 for large fire compartments include an ASET/RSET comparison with safety factors. For new buildings the factors range from 1.2 to 2.5. They are higher when RSET exceeds 3 minutes and when a rule-of-thumb method is used with nominal (50 %) rather than conservative (80 %) parameters. For existing buildings the factors range from 1.0 to 1.5.<ref name="NEN-VR-7.1">NEN 6060 and NEN 6079, as cited in Van Rossum (2027), §7.1.</ref> A conservative solution is possible only for ASET, because RSET can only be entered as a nominal value.<ref name="NEN-VR-7.1" />

[[File:fse-available-safe-egress-time-v3.svg|thumb|upright=1.5|alt=Grouped bar chart of Dutch ASET/RSET safety factors from 1.0 to 2.5, highest for new buildings, nominal parameters and RSET over 3 minutes.|Safety factors on RSET in the Dutch standards NEN 6060 and NEN 6079 for large fire compartments. Based on NEN 6060 and NEN 6079, as cited in Van Rossum (2027), §7.1.]]

The International Fire Engineering Guidelines describe the same comparison as a timeline approach. In an absolute evaluation, ASET should exceed RSET by a margin agreed in the fire engineering brief. In a comparative evaluation, the margin of the proposed design should be the same as or greater than that of a design meeting the prescriptive rules.<ref name="IFEG-1.2-22">National Research Council of Canada et al. (2005), p. 1.2-22.</ref> In Australia, the 2020 data sheets for the Fire Safety Verification Method take the same comparative view. The margin ASET − RSET for a [[performance-based design|performance solution]] should be at least equal to that of the reference building.<ref name="ABCB-D1-241">Australian Building Codes Board (2020a), Data Sheet D1, p. 241.</ref>

Where both designs have the same occupants, alarm and exits, the RSET can be assumed equal and only the ASET values need to be compared.<ref name="ABCB-H-63">Australian Building Codes Board (2020b), pp. 63–64.</ref> Such a comparison avoids the need to treat the variability of human behaviour explicitly.<ref name="ABCB-H-69">Australian Building Codes Board (2020b), p. 69.</ref> New Zealand's C/VM2 (second edition, 2025) simply requires ASET to be greater than RSET for the design scenarios concerned.<ref name="MBIE-14">Ministry of Business, Innovation and Employment (2025b), p. 14.</ref>

== Tenability criteria ==
ASET ends when the first limit for safe escape is exceeded. The limits are called [[tenability criteria]] and usually cover heat, [[visibility in smoke|visibility]] and toxic gases.<ref name="VR-7.7">Van Rossum (2027), §7.7.</ref> According to the book, such limits are not formally laid down anywhere, but several national and international sources give guidance.<ref name="VR-7.7" />

In the Netherlands, the explanatory notes to the building regulations give values for non-enclosed spaces through which an escape route runs. These are a smoke-free height of 2.5 m above the highest part of the escape route and a radiant heat flux below the smoke layer of at most 2 kW/m². When people escape through mixed smoke, the temperature should not exceed 45 °C and light-emitting objects should be visible from at least 30 m.<ref name="VR-7.7" /> The book calls these values strict. In the United Kingdom, BS 9999 lets visibility fall to 8 to 10 m, and in corridors 5 m is often accepted. A visibility of 30 m is only needed in large spaces.<ref name="VR-7.7" />

Heat limits follow from radiation and from hot gases. In Dutch practice 2.5 kW/m² is commonly used for escape and 1 kW/m² for long stays. A very conservative calculation shows that a smoke layer at about 200 °C radiates roughly this flux to a person below it, so the layer should be no hotter than that.<ref name="VR-7.7" /> In mixed hot gases, about 80 °C can be endured for a short escape, but for exposures over ten minutes in humid air the limit is about 60 °C. For toxic gases, the dose taken in during escape counts. With less than 30 to 60 seconds in smoke, toxicity is usually not decisive.<ref name="VR-7.7" />

In New Zealand, Building Code clause C4.3 sets three limits at 2.0 m above the floor. These are a fractional effective dose (FED) of carbon monoxide of 0.3, an FED of thermal effects of 0.3, and a visibility of 10 m, or 5 m in rooms smaller than 100 m².<ref name="MBIE-2013-28">Ministry of Business, Innovation and Employment (2013), p. 28.</ref><ref name="MBIE-17" /> An FED of 1.0 is the dose expected to stop a person of average susceptibility from escaping, which affects about half of the occupants. The threshold of 0.3 corresponds to about 11 % of the population being susceptible, and no threshold is safe for every occupant.<ref name="MBIE-2013-29">Ministry of Business, Innovation and Employment (2013), p. 29.</ref> Clause C4.4 drops the thermal and visibility limits in sprinklered firecells where no more than 1,000 occupants can be exposed.<ref name="MBIE-2013-28" />

[[File:fse-available-safe-egress-time-v2.svg|thumb|upright=1.5|alt=Room cross-section with a smoke layer, a person walking to an exit and a dashed line 2.0 m above the floor where tenability is checked.|Tenability is assessed 2.0 m above the floor: ASET ends when the smoke at that height exceeds the first limit for heat, carbon monoxide or visibility. Based on Ministry of Business, Innovation and Employment (2013), p. 28.]]

In Australia, the Fire Safety Verification Method uses an FED of thermal effects of 0.3 and the same visibility limits of 10 m and 5 m. The 5 m value also applies where the distance to an exit is 5 m or less.<ref name="ABCB-2022-10" /> The handbook adds that visibility should be calculated for back-lit exit signs and that smouldering fires need a carbon monoxide criterion.<ref name="ABCB-H-110">Australian Building Codes Board (2020b), p. 110.</ref> Purser's suggested limits point the same way: visibility of 5 m for small enclosures with short travel distances, and 10 m for large enclosures with long travel distances.<ref name="Purser-2-118">Purser (2002), p. 2-118.</ref>

== How ASET is determined ==
ASET is mostly calculated with a model that simulates the effects of a fire on a space, in particular the smoke layer and the smoke-free layer below it. The model shows how long it takes before conditions become too poor for escape.<ref name="VR-8" /> The limits must be chosen before the simulation is run.<ref name="VR-7.6">Van Rossum (2027), §7.6.</ref> In order of complexity, the book lists a steady-state zone model (NEN 6093), dynamic zone models such as OZone and CFAST, and computational fluid dynamics with FDS.<ref name="VR-8" /> The choice depends on whether the model can represent the key features of the scenario. Examples are a fire in an adjacent room, a post-flashover fire or a complex geometry.<ref name="VR-8" />

In a single room, the simplest picture is [[smoke filling]]: the smoke layer under the ceiling grows downwards until it reaches head height. A simple one-room smoke-filling computer program was itself called ASET. It calculates the thickness and temperature of the smoke layer and the concentrations of combustion products for a user-defined fire.<ref name="Iq-B-85">Iqbal and Salley (2004), p. B-85.</ref> For a room without ventilation and a constant heat release rate, the related ASET correlation predicts the height of the hot layer interface.<ref name="Stroup-4-2">Stroup et al. (2016), p. 4-2.</ref>

Verification methods add rules so that ASET is not overestimated. C/VM2 requires the fire to be placed away from walls and corners, with its base no higher than 0.5 m above the floor. A higher fire would be detected earlier and its smoke layer would descend more slowly, which would give a longer ASET.<ref name="MBIE-17" /> Several fire locations must be checked to find the lowest ASET for an escape route.<ref name="MBIE-18" /> In both C/VM2 and the Australian method, visibility is generally the first criterion to be exceeded unless an exception applies.<ref name="MBIE-17" /><ref name="ABCB-2022-10" />

Purser illustrates the order in which hazards arise with an analysis of a furniture fire in a room. Smoke obscuration was the first hazard and exceeded its limit after the second minute. Irritant gases followed in the third minute, heat crossed its limit during the fourth minute, and the radiant flux reached 2.5 kW/m² during the fifth.<ref name="Purser-2-133">Purser (2002), p. 2-133.</ref> Escape would become difficult in the third minute. Large-scale tests suggest that even a rapidly growing flaming fire leaves at least three minutes before conditions become untenable, and often much longer.<ref name="Purser-2-136">Purser (2002), p. 2-136.</ref>

== Worked example ==
The commentary to the first edition of C/VM2 (New Zealand, 2013) gives an example of a sprinklered hospital ward with 12 single patient rooms. A fire starts in a patient room of 8.0 m by 4.0 m with an open door; all other doors are closed. The corridor and nurses' area are modelled as one room of 5.0 m by 32 m, both rooms being 2.4 m high.<ref name="MBIE-2013-98">Ministry of Business, Innovation and Employment (2013), p. 98.</ref> The zone model results give smoke detection at 46 s and sprinkler activation at 112 s, which caps the fire at 565 kW.<ref name="MBIE-2013-98" />

In the patient room, visibility drops below 10 m after 23 s and the thermal FED exceeds 0.3 after 118 s. In the corridor this happens after 86 s and 628 s. Because the building is sprinklered, only the carbon monoxide FED counts. It exceeds 0.3 after 586 s in the patient room and after 863 s in the corridor, and these are the ASET values.<ref name="MBIE-2013-98" /><ref name="MBIE-2013-100">Ministry of Business, Innovation and Employment (2013), p. 100.</ref>

[[File:fse-available-safe-egress-time-v4.svg|thumb|upright=1.5|alt=Two timelines for a sprinklered hospital ward: visibility, thermal dose and carbon monoxide limits reached at different times, compared with RSET markers.|Times at which the tenability limits are exceeded in the fire room and the corridor of a sprinklered hospital ward, with the RSET of each space. Based on Ministry of Business, Innovation and Employment (2013), pp. 98 and 100.]]

All patients are assumed unable to walk, so staff must move them. The RSET follows from detection (46 s), alarm verification (30 s), the time for nurses to arrive (60 s) and 120 s per patient moved by two staff. This gives 256 s for the fire room and 856 s for the whole ward with four staff.<ref name="MBIE-2013-100" />

For the fire room the margin is 586 − 256 = 330 s, a factor of 2.29 on RSET. For the ward the margin is only 863 − 856 = 7 s, a factor of 1.01. The ward just meets the bare requirement that ASET exceeds RSET.<ref name="MBIE-2013-100" /> It would not meet a safety factor of 1.2, the lowest factor in the Dutch table for new buildings.<ref name="NEN-VR-7.1" /> The example also shows how strongly ASET depends on the criterion chosen: without the sprinkler exception, visibility would end the ASET in the corridor after 86 s.<ref name="MBIE-2013-98" />

== Limits and uncertainty ==
Tenability limits contain deliberate conservatism. One reason is that the endpoints cannot be measured precisely, because humans cannot be exposed to dangerous smoke in experiments. Another is the wide range of sensitivity in a mixed population. Purser names the elderly and people with asthma, about 15 % of children and 5 % of adults, as groups that are more sensitive to fire gases.<ref name="Purser-2-90" /> The C/VM2 commentary therefore notes that the FED limit may be set lower for vulnerable populations.<ref name="MBIE-2013-31">Ministry of Business, Innovation and Employment (2013), p. 31.</ref>

ASET is also limited by the fire model. The result holds only for the design fire, fire location and space that were analysed, and different fire locations can give different values.<ref name="MBIE-18" /> A model without the relevant features, such as a fire in an adjacent room or flashover, can give a misleading ASET.<ref name="VR-8" /> In the C/VM2 example, the choice of criterion alone changed the corridor ASET from 86 s for visibility to 863 s for carbon monoxide.<ref name="MBIE-2013-100" />

The criteria assume that people are in the smoke at a given height. In Purser's analysis the occupant's head was assumed to be in the smoke at all times. In a real room the hot layer might settle at 1 to 1.2 m above the floor, and a person lower down would be exposed later.<ref name="Purser-2-133" /> The book stresses that the limits must be judged in context. Navigating a corridor is easier than crossing a large hall, so a lower visibility can be accepted there.<ref name="VR-7.7" />

== See also ==
* [[Effects of heat on people]]
* [[Pre-movement time]]
* [[Evacuation modelling]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Purser, D. A. (2002). Toxicity assessment of combustion products. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-83–2-171). National Fire Protection Association; Society of Fire Protection Engineers.
* Stroup, D., Lindeman, A., McGrattan, K., Peacock, R., Overholt, K., Joglar, F., LeStrange, S., & Montanez, S. (2016). ''Verification and validation of selected fire models for nuclear power plant applications: Supplement 1'' (NUREG-1824, Supplement 1; EPRI 3002002182). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Fire scenario

2026-11-02T17:13:48Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''fire scenario''' is a description of how one particular fire could start and develop in a building, from ignition through growth, spread and decay to burnout, as modified by the building's fire safety system.<ref name="IFEG-053">National Research Council of Canada et al. (2005), p. 0.5-3 (Part 0, Australia).</ref> It is described by the times at which its key events occur, such as detection, sprinkler operation, flashover or the failure of a barrier.<ref name="IFEG-053" /> A fire scenario is broader than the fire alone. It includes the ignition source and the first item ignited, the arrangement of fuel and enclosures, the status of fire protection measures, the occupants and, where relevant, external conditions such as wind.<ref name="BU-231">Bukowski (2002), p. 5-231.</ref><ref name="IFEG-222">National Research Council of Canada et al. (2005), pp. 2.2-2–2.2-3.</ref>

Fire scenarios are the starting point of almost every fire safety analysis. Watts and Hall describe fire risk analysis as answering three questions, the first being what could happen. A sequence of events ending in a fire loss is called a scenario.<ref name="WH-2">Watts and Hall (2002), p. 5-2.</ref> Because a building can burn in countless ways, engineers analyse only a limited set of design fire scenarios.<ref name="IFEG-1227">National Research Council of Canada et al. (2005), p. 1.2-27.</ref> These are chosen to be severe but credible, and each is turned into a design fire for calculation.<ref name="IFEG-1227" /> Several regulations now prescribe a minimum set of design scenarios, such as the ten scenarios of New Zealand's Verification Method C/VM2.<ref name="CVM2-7">Ministry of Business, Innovation and Employment (2025b), p. 7.</ref>

== Definitions ==
The International Fire Engineering Guidelines (IFEG, 2005) define a fire scenario as the ignition, growth, spread, decay and burnout of a fire in a building as modified by its fire safety system.<ref name="IFEG-053" /> The glossary of NUREG-1805, a report of the U.S. Nuclear Regulatory Commission, gives a similar definition. There a fire scenario is a set of conditions that defines the development of fire and the spread of combustion products through a building or part of a building.<ref name="IS-F28">Iqbal and Salley (2004), p. F-28.</ref> Bukowski, writing about fire hazards in passenger rail, describes a fire scenario as the sequence of events that must occur to result in an uncontrolled fire.<ref name="BU-231" />

Three related terms are easily confused. A fire scenario is any possible course of a fire, whereas a design fire scenario is the subset selected for design purposes.<ref name="ME-162">Meacham (2002), p. 5-162.</ref> The IFEG define a design fire scenario as a fire scenario that is used as the basis for a [[design fire]].<ref name="IFEG-052">National Research Council of Canada et al. (2005), p. 0.5-2 (Part 0, Australia).</ref> A design fire in turn is a representation of a fire by the variation of its heat output with time.<ref name="IFEG-052" /> The scenario is therefore the story, and the design fire is the quantified heat release curve used to tell it in a calculation.

The NUREG-1805 glossary stresses the purpose of a design fire scenario. It is a set of conditions that describes the critical factors for determining the outcomes of trial designs.<ref name="IS-F21">Iqbal and Salley (2004), p. F-21.</ref> New Zealand's C/VM2 uses the shorter term design scenario. The first edition defined it as a specific scenario on which a deterministic fire safety engineering analysis is conducted.<ref name="MBIE13-9">Ministry of Business, Innovation and Employment (2013), p. 9.</ref>

== What a scenario describes ==
A fire scenario combines the fire with the building and the people in it. The IFEG list the factors that a scenario takes into account. These are the nature, quantity, arrangement and burning behaviour of combustibles, the geometry of the enclosure, and the number of enclosures and the connections between them.<ref name="IFEG-1227" /> The fire protection measures in the building and their effect on the fire complete the list.<ref name="IFEG-1227" />

A scenario represents one particular combination of outcomes for the types of fire, its development and the external environment.<ref name="IFEG-222" /> The fire may be smouldering, flaming without flashover, or flashover.<ref name="IFEG-222" /> Its development is influenced by the size and type of ignition source, the distribution and type of fuel, the fire load density, the location of the fire, ventilation, construction materials and air handling equipment.<ref name="IFEG-222" /> External conditions include the season, wind speed and direction, and the stack effect.<ref name="IFEG-222" />

The IFEG also define a scenario by the outcome of each of the six sub-systems of a fire safety system.<ref name="IFEG-223">National Research Council of Canada et al. (2005), p. 2.2-3.</ref> Typical alternatives are whether smoke management operates, whether doors are open or closed, and whether detectors and sprinklers succeed. Others are whether occupants are awake or asleep, and whether the fire service rescues and extinguishes successfully.<ref name="IFEG-223" /> A simple example with sprinklers and barriers gives three scenarios. In the first, sprinklers control the fire; in the second, sprinklers fail but the barriers contain the fire; in the third, both fail and the fire spreads to adjoining enclosures.<ref name="IFEG-223" />

[[File:fse-fire-scenario-v1.svg|thumb|upright=1.5|alt=Event tree: ignition, sprinklers succeed (scenario I) or fail; if they fail, barriers hold (scenario II) or fail (scenario III).|Event tree for a fire with sprinklers and barriers: each path from ignition to an outcome is one fire scenario. Based on National Research Council of Canada et al. (2005), pp. 2.2-3–2.2-4.]]

The people form part of the scenario as well. The IFEG describe occupant response scenarios in which people notice the fire directly, respond to an alarm, or respond later to new cues or to warnings from others.<ref name="IFEG-225">National Research Council of Canada et al. (2005), p. 2.2-5.</ref> Each occupant response scenario is combined separately with each fire scenario.<ref name="IFEG-226">National Research Council of Canada et al. (2005), p. 2.2-6.</ref> For escape analysis, ''Introduction to Fire Safety Engineering'' lists the possible or expected fire scenario among the aspects that determine whether people can escape safely.<ref name="VR-7">Van Rossum (2027), §7.</ref>

== Identifying potential scenarios ==
The first task is to find the potential fire scenarios. The IFEG suggest reviewing the information assembled for the fire engineering brief, especially the hazard analysis.<ref name="IFEG-1227" /> Other sources are the published literature, fire statistics, and the experience and knowledge of the team.<ref name="IFEG-1227" /> The Australian handbook for the Fire Safety Verification Method (FSVM) names several hazard identification techniques. These are checklists, what-if analysis, hazard identification (HAZID), hazard and operability analysis (HAZOP), failure mode and effects analysis, and reviews of the literature and of historic records.<ref name="HB-59">Australian Building Codes Board (2020b), p. 59.</ref> The handbook recommends a mix of structured and informal methods, so that free thinking can reveal significant events of low probability and high consequence.<ref name="HB-59" />

For a systematic search, the IFEG recommend [[event tree analysis]]. An event tree starts with an initiating event, such as ignition, and branches at each system or event into success or failure.<ref name="IFEG-224">National Research Council of Canada et al. (2005), p. 2.2-4.</ref> Each path through the tree, from the initiating event to a final event, defines one fire scenario.<ref name="IFEG-224" /> Where the probability of an event is not known, fault trees can be used to calculate it from the failure mechanisms of the system.<ref name="IFEG-224" />

The number of scenarios grows quickly. Meacham gives the example of a building with three compartmented rooms.<ref name="ME-162" /> If only the room of origin and the order of spread are considered, there are 15 possible paths of fire propagation. Adding simultaneous spread to the second and third rooms gives 18 scenarios.<ref name="ME-162" /> That count is reached before open or closed doors, the characteristics of the people present, or whether they are asleep are even considered.<ref name="ME-162" />

[[File:fse-fire-scenario-v4.svg|thumb|upright=1.5|alt=Plan of three rooms in a row with a list of the five spread paths for a fire starting in room 1; 15 paths in total.|A building of three rooms already has 15 possible paths of fire spread, or 18 with simultaneous spread. Based on Meacham (2002), p. 5-162.]]

== Selecting design fire scenarios ==
A fire engineering analysis can consider only a limited number of the potential scenarios.<ref name="IFEG-1227" /> The IFEG describe a three-step approach. The team first determines the potential fire scenarios, then selects the design fire scenarios, and finally specifies a schematic design fire for each.<ref name="IFEG-1227" /> Usually, the selected scenarios are severe, have a reasonable probability of occurrence and have a significant potential for loss.<ref name="IFEG-1227" /> Events with a very low probability should not be analysed unnecessarily. If their consequences could be very high, however, they deserve consideration at least in the sensitivity studies.<ref name="IFEG-1227" />

[[File:fse-fire-scenario-v2.svg|thumb|upright=1.5|alt=Funnel: many potential fire scenarios narrow to a few design fire scenarios, each turned into a design fire curve.|From potential fire scenarios to design fires in three steps. Based on National Research Council of Canada et al. (2005), p. 1.2-27.]]

Deliberately started fires are treated in the same way as accidental ones when design scenarios are chosen.<ref name="IFEG-1227" /> The IFEG recognise, however, that some large arson attacks and terrorist events may be beyond the capacity of the fire safety system and need other measures.<ref name="IFEG-1228">National Research Council of Canada et al. (2005), p. 1.2-28.</ref> The selection is made by the stakeholders in the fire engineering brief.<ref name="IFEG-1227" /> At that stage, the design fires are defined and described as far as possible without calculation, in a largely [[Qualitative design review|qualitative review]] of the design.<ref name="IFEG-1229">National Research Council of Canada et al. (2005), p. 1.2-29.</ref>

Some regulations prescribe a minimum set of design scenarios. In New Zealand, Verification Method C/VM2 (second edition, 2025) contains ten design scenarios that must each be satisfied separately.<ref name="CVM2-7" /> They range from a fire that blocks an exit and a smouldering fire to horizontal and vertical fire spread, firefighting operations and a challenging fire.<ref name="CVM2-8">Ministry of Business, Innovation and Employment (2025b), p. 8.</ref> The scenarios are intended to give a diverse range of fire events that will challenge the design and its fire safety systems.<ref name="CVM2-7" /> Many can be satisfied by inspection or by providing features such as fire separations or automatic alarms.<ref name="CVM2-7" /> Only a few need a comparison of the available safe egress time (ASET) with the required safe egress time (RSET), or other computational modelling.<ref name="CVM2-7" />

The challenging fire scenario of C/VM2 represents credible worst cases in normally occupied spaces. Design fires are placed in various locations in the building, using a single fire source for each location.<ref name="CVM2-56">Ministry of Business, Innovation and Employment (2025b), p. 56.</ref> In this scenario, active and passive fire safety systems are assumed to perform as designed.<ref name="CVM2-57">Ministry of Business, Innovation and Employment (2025b), p. 57.</ref> A separate scenario then assumes the failure of each key fire safety system in turn.<ref name="CVM2-58">Ministry of Business, Innovation and Employment (2025b), p. 58.</ref> The commentary to the first edition explains that the framework was modelled on structural design, which also specifies design events such as wind, earthquake and snow.<ref name="MBIE13-11">Ministry of Business, Innovation and Employment (2013), p. 11.</ref>

In Australia, the FSVM specifies a minimum of twelve design scenarios. Each is considered in one or more locations to compare the proposed design with a reference building that fully complies with the deemed-to-satisfy provisions.<ref name="HB-7">Australian Building Codes Board (2020b), p. 7.</ref> The handbook explains that prescribing scenarios reduces the risk that critical design scenarios are missed. It notes similar approaches in C/VM2, in NFPA 5000 in the United States and in ISO 16733-1.<ref name="HB-8">Australian Building Codes Board (2020b), p. 8.</ref>

Hazard identification may show that some prescribed scenarios do not apply. For example, the smouldering fire scenario near sleeping areas is not needed in an office without sleeping accommodation.<ref name="HB-60">Australian Building Codes Board (2020b), p. 60.</ref> Buildings with innovative or unusual features may need additional scenarios.<ref name="HB-60" />

== From scenario to design fire ==
Once a design fire scenario has been chosen, it must be quantified. The Australian handbook lists what is needed: a design fire, the status of active and passive fire protection, occupant characteristics where required, and acceptance criteria.<ref name="HB-59" /> The IFEG compare this step with structural engineering, where the loads must be specified before safety can be evaluated.<ref name="IFEG-1227" /> The validity of the whole fire engineering evaluation rests on the validity of the design fires.<ref name="IFEG-1227" />

The IFEG first sketch each design fire as a schematic heat release rate curve. A smouldering fire that does not become flaming typically has a maximum heat release rate below 5 kW.<ref name="IFEG-1229" /> For flaming fires, the heat release rate is generally assumed to grow with the square of time, reach a steady value limited by fuel or ventilation, and then decay.<ref name="IFEG-1230">National Research Council of Canada et al. (2005), p. 1.2-30.</ref> For a flashover fire, flashover is drawn as an instantaneous jump in heat release rate.<ref name="IFEG-1230" /> The curve is then adjusted for events such as windows breaking, smoke management operating, doors burning through, and suppression by sprinklers, occupants or the fire service.<ref name="IFEG-1230" />

[[File:fse-fire-scenario-v3.svg|thumb|upright=1.5|alt=Three heat release rate curves: a low smouldering curve, a non-flashover curve with growth, plateau and decay, and a flashover curve with a vertical jump.|Schematic design fires for smouldering, non-flashover and flashover fire scenarios. Based on National Research Council of Canada et al. (2005), pp. 1.2-29–1.2-30.]]

''Introduction to Fire Safety Engineering'' describes the t-squared growth curve as a widely used starting point in a fire scenario. Its four standard growth rates reach 1 MW after 75 s (ultra-fast), 150 s (fast), 300 s (medium) and 600 s (slow).<ref name="VR-8.6">Van Rossum (2027), §8.6.</ref> The book stresses that a temperature-time curve is not suitable for representing a fire in a room. The temperature is a result of a fire scenario, not an input to it.<ref name="VR-8.6" /> A study of fires in indoor car parks makes the same point, representing the car park fire scenario by a heat release rate curve rather than a temperature curve.<ref name="VG-24">DGMR (2015), as cited in Van der Graaf et al. (2021/2023), p. 24.</ref>

Fire models do not choose the scenario themselves. The book points out that a computational fluid dynamics model does not predict how a fire develops. The modeller must prescribe the fire scenario and the growth of the heat release rate.<ref name="VR-11.4">Van Rossum (2027), §11.4.</ref> Before any model is used, it must also be established that the scenario lies within the model's field of application.<ref name="VR-8.1">Van Rossum (2027), §8.1.</ref>

== Scenarios in risk analysis ==
A deterministic analysis calculates the outcome of each design fire scenario, but it does not indicate the probability of that outcome. A probabilistic analysis also assigns a probability to each scenario and combines the results.<ref name="IFEG-1222">National Research Council of Canada et al. (2005), p. 1.2-22.</ref> The IFEG note that the use of multiple scenarios combined by probabilistic techniques is the key feature of some risk methods.<ref name="IFEG-1222" /> Such methods need the [[reliability of fire protection systems]], because each branch of the event tree depends on whether a system works.<ref name="ME-162" /> Meacham distinguishes availability, meaning a system is ready to perform, from reliability, meaning that an available system actually functions.<ref name="ME-162" />

The risk-cost assessment model described by Meacham shows how scenarios are built up in practice. It uses six design fires in the room of origin: smouldering, flaming without flashover, and flashover, each with the room door open or closed.<ref name="ME-163">Meacham (2002), p. 5-163.</ref> The probability of each design fire is based on statistics. In Canada, 18% of apartment fires reach flashover, 63% are flaming fires without flashover, and 19% are smouldering fires.<ref name="ME-163" /> The number of fire scenarios is the product of the number of design fires, the number of apartment units, and whether the occupants are awake or asleep.<ref name="ME-163" /> For an illustrative building of 20 apartments, this already gives 6 × 20 × 2 = 240 scenarios.

[[File:fse-fire-scenario-v5.svg|thumb|upright=1.5|alt=Bar chart of Canadian apartment fires: flashover 18%, flaming without flashover 63%, smouldering 19%.|Share of apartment fires in Canada by type, used to give each design fire a probability. Based on Meacham (2002), p. 5-163.]]

Frequencies must be handled with care. Meacham warns that a single scenario with a frequency of 10−7 per year may suggest that fire is not a concern.<ref name="ME-166">Meacham (2002), p. 5-166.</ref> If 30 such scenarios are developed, however, their combined frequency is 3 × 10−6 fires per year.<ref name="ME-166" /> In a [[fire risk assessment]], the consequences of each scenario are therefore weighted by its likelihood and summed over all scenarios.<ref name="BU-231" /><ref name="ME-161">Meacham (2002), p. 5-161.</ref>

In fire probabilistic risk assessment of US nuclear power plants, a fire scenario is defined by the location and burning characteristics of the initiating fire.<ref name="SI-216">Siu (2002), p. 5-216.</ref> Scenarios can be defined broadly, for example as a fire anywhere in an area, or in detail with a precise location and initial size.<ref name="SI-217">Siu (2002), p. 5-217.</ref> Broad definitions are used where a conservative analysis is enough, and detailed ones where the area is risk-significant.<ref name="SI-217" />

== Limits and misconceptions ==
A design fire scenario is not a prediction of the fire that will occur. It is a deliberately chosen case for testing a design, and design fires are normally intended to be conservative simplifications.<ref name="IFEG-1229" /> The IFEG warn that fire and its effects are complex and variable, so a fire safety system may not cope with every possible scenario.<ref name="IFEG-014">National Research Council of Canada et al. (2005), p. 0.1-4 (Part 0, Australia).</ref> Owners, occupiers and approving authorities need to understand this limit when they assess a fire engineered solution.<ref name="IFEG-014" />

"Worst credible" is not the same as "worst possible". Selection balances severity against probability, and very unlikely events are usually examined only in sensitivity studies.<ref name="IFEG-1227" /> For the same reason, a scenario set chosen for one building cannot be transferred to another without checking. The Australian handbook stresses that practitioners must consider each project case by case and not rely only on tables of scenarios.<ref name="HB-55">Australian Building Codes Board (2020b), p. 55.</ref>

Scenarios can also be used without any probability. A Dutch study on indoor car parks with electric vehicles defined three realistic scenarios. These were a long-lasting battery fire in one car, an electric car exposed to radiation from a burning neighbour, and several burning battery packs, for example in a travelling fire.<ref name="VG-32">Van der Graaf et al. (2021/2023), pp. 32–33.</ref> The authors state that no statements can be made about the probability of these scenarios. They served only as a conceptual framework for choosing prevention and firefighting measures.<ref name="VG-32" /> Such qualitative scenarios support decisions in [[performance-based design]], but they do not by themselves quantify risk.

== See also ==
* [[Fire safety strategy]]
* [[Prescriptive design]]
* [[Fire safety objectives]]
* [[Acceptance criteria]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Bukowski, R. W. (2002). Fire hazard assessment for transportation vehicles. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-227–5-233). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Meacham, B. J. (2002). Building fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-153–5-175). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Siu, N. (2002). Fire risk assessment for nuclear power plants. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-214–5-226). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Safety Design & Regulation]]

Wet pipe sprinkler system

2026-11-02T17:13:45Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''wet pipe sprinkler system''' is an automatic [[sprinkler system]] in which the pipes are filled with water under pressure at all times. When the heat of a fire opens a sprinkler, water flows from it at once and keeps flowing until the system is shut off.<ref name="GJ-xix">Grill and Johnson (2005a), p. xix.</ref> Only the sprinklers that are heated enough to operate discharge water; the others stay closed.<ref name="FL-72">Fleming (2002), p. 4-72.</ref> The wet pipe system is by far the most common of the four basic sprinkler system types, which also include dry pipe, deluge and preaction systems.<ref name="FL-72" />

Wet pipe systems are described as the oldest and most reliable type of sprinkler system. They are simple, have few mechanical parts and need no human action to operate.<ref name="HO-14">Hopkins et al. (2005), p. 14.</ref> Most are designed to control a fire rather than to extinguish it: they limit its growth and keep it within the room of origin until the fire service completes extinguishment.<ref name="HO-15">Hopkins et al. (2005), p. 15.</ref> Their main limitation is that the water in the pipes must not freeze, so the protected spaces must be heated.<ref name="IS-10-1">Iqbal and Salley (2004), p. 10-1.</ref> For residential and general buildings, US statistics show that the systems operate in about 90 to 95 % of fires large enough to activate them, and are effective in 96 to 97 % of those cases.<ref name="AB-71">Australian Building Codes Board (2020a), Data Sheet C2, p. 71.</ref>

== How the system works ==
A sprinkler system consists of a water supply, a network of distribution pipes and individual sprinklers, supported by control valves, pumps and water flow alarms.<ref name="HO-13">Hopkins et al. (2005), p. 13.</ref> In a wet pipe system, the whole network from the supply to the sprinklers is permanently charged with water.<ref name="IS-10-1" /> An alarm check valve is fitted where the water enters from the supply. It has connections for local and remote alarms, and it acts as a non-return valve that lets water flow only towards the sprinklers.<ref name="IS-10-1" />

[[File:fse-wet-pipe-sprinkler-system-v1.svg|thumb|upright=1.5|alt=Building section: supply and alarm check valve at the bottom, water-filled riser and branch lines, one open sprinkler spraying onto a fire.|A wet pipe sprinkler system. All pipes are permanently filled with water under pressure; only the sprinkler heated by the fire opens. Based on Iqbal and Salley (2004), p. 10-1 and Hopkins et al. (2005), pp. 13 to 14.]]

Each [[sprinkler head]] is held closed by a heat-sensitive element. The most common elements are a soldered metal link, which melts at a set temperature, and a glass bulb filled with liquid, which expands and shatters the bulb.<ref name="IS-10-3">Iqbal and Salley (2004), p. 10-3.</ref><ref name="IS-10-4">Iqbal and Salley (2004), p. 10-4.</ref> When the element releases, water flows out of the orifice and strikes a deflector, which breaks it into a spray over the burning material.<ref name="IS-10-1" /> A sprinkler therefore acts both as a heat detector and as a nozzle that distributes water.<ref name="FL-85">Fleming (2002), p. 4-85.</ref> The spray sprinkler, introduced in the early 1950s, was a major step forward because it spread the water much more evenly than earlier designs.<ref name="HO-13" />

Operation is local. When a fire grows large enough, a sprinkler close to it opens and discharges water at a set rate. If the fire keeps growing, further sprinklers open, until the fire is controlled or extinguished or the water supply runs out.<ref name="HO-14" /> Water cools the burning materials as it evaporates. Sprinklers are designed to control a fire, but they often extinguish it, because the surrounding materials can no longer heat up to their ignition temperature.<ref name="IS-10-1" /> One gallon of water, about 3.8 litres, absorbs 2,586.5 kJ when it is heated from 21 °C and turned into steam at 100 °C.<ref name="FL-72" />

Most wet pipe systems in offices are control-mode systems. Their aim is to limit the heat release rate and keep the fire in the room or area of origin. Full extinguishment may need the fire service.<ref name="HO-15" /><ref name="HO-16">Hopkins et al. (2005), p. 16.</ref> A separate class of sprinklers is designed for early response and extinguishment, for example in storage buildings and dwellings.<ref name="HO-16" /> In general, an active extinguishing system intervenes while the fire is still growing, so the fire does not become fully developed.<ref name="VR-4.11">Van Rossum (2027), §4.11.</ref>

== Comparison with other system types ==
All sprinkler systems, whatever their purpose, can be placed in one of four basic types. These types differ in the most basic aspect: how water reaches the area of the fire.<ref name="FL-73">Fleming (2002), p. 4-73.</ref> In a wet pipe system, water is discharged as soon as the first sprinkler opens.<ref name="HO-15" />

[[File:fse-wet-pipe-sprinkler-system-v3.svg|thumb|upright=1.5|alt=Four pipe sketches: wet pipe full of water, dry pipe full of air, deluge with open heads and a detector valve, preaction with closed heads and a detector.|The four basic sprinkler system types: wet pipe, dry pipe, deluge and preaction. Based on Fleming (2002), pp. 4-72 to 4-73.]]

A [[dry pipe sprinkler system]] contains compressed air or nitrogen instead of water. A dry pipe valve holds the water back until an open sprinkler releases the air pressure.<ref name="IS-10-2">Iqbal and Salley (2004), p. 10-2.</ref> Such systems are used where water in the pipes could freeze, for example at loading docks, in cold stores and in attics.<ref name="FL-72" /><ref name="HO-14" /> The air must first escape and the water must then travel through the pipes, so the discharge is delayed. In the United States, NFPA 13 limits this delay to 60 seconds and increases the design area of dry systems by 30 % above that of wet systems.<ref name="HO-15" /> Alternating wet and dry periods also cause corrosion inside the pipes.<ref name="IS-10-2" />

Deluge systems use open sprinklers. A separate detection system opens a deluge valve, after which water flows from all sprinklers at once; they are used for rapidly spreading, high hazard fires.<ref name="FL-72" /> Preaction systems use closed sprinklers, but water enters the pipes only after a detection system has opened the valve. They are used where accidental discharge is a particular concern, such as in computer rooms.<ref name="FL-72" /><ref name="FL-73" /> A preaction system is more expensive and complicated than a wet pipe system.<ref name="IS-10-2" /> Where some freezing risk exists, a wet pipe system can be filled with an antifreeze solution, which is discharged first and is followed by water.<ref name="IS-F-4">Iqbal and Salley (2004), p. F-4.</ref>

The World Trade Center towers illustrate the usual pattern in office buildings. The office and support spaces were protected by wet pipe systems. Dry pipe and preaction systems covered less than 10 % of the floor area, mainly in computer rooms and other special hazard areas.<ref name="HO-14" />

== Design ==
In the United States and Canada, sprinkler systems are designed and installed to NFPA 13, which most building codes reference.<ref name="FL-73" /> In Australia, the effectiveness values of the national verification method assume wet pipe systems built and maintained to AS 2118 and AS 1851.<ref name="AB-70">Australian Building Codes Board (2020a), Data Sheet C2, p. 70.</ref> Fleming notes that calculation methods are well developed only for the flow of water through the pipes. The ability of a spray to suppress a fire cannot yet be calculated, so the fire suppression side of sprinkler design is empirical.<ref name="FL-73" />

Design starts with the hazard classification of the occupancy, which Fleming calls the most critical step. If the hazard is underestimated, the fire may overpower the sprinklers.<ref name="FL-73" /> NFPA 13 uses the classes light hazard, ordinary hazard groups 1 and 2, and extra hazard groups 1 and 2. Offices are light hazard, laundries ordinary hazard group 1 and flammable liquid spray booths extra hazard group 2.<ref name="HO-16" /><ref name="HO-17">Hopkins et al. (2005), p. 17.</ref> The class sets a minimum water density, a flow per unit area, over a design area in which the sprinklers are expected to open.<ref name="HO-16" />

The designer then shows by a [[hydraulic calculation]] that the pipes deliver this density to every sprinkler in the hydraulically most remote design area.<ref name="FL-74">Fleming (2002), p. 4-74.</ref> In one example from Fleming, a density of 6.1 mm/min over 221 m² is chosen. With each sprinkler covering 16.72 m², the design area contains 14 sprinklers.<ref name="FL-74" /> Before the 1970s, systems were sized from pipe schedule tables; area and density curves then made hydraulic calculation possible and allowed more economical systems on strong water supplies.<ref name="FL-73" />

=== Flow from a sprinkler ===
The flow through a sprinkler depends on the water pressure at its orifice. All properties of the orifice are lumped into one experimentally determined constant, the K-factor:<ref name="FL-75">Fleming (2002), p. 4-75.</ref>

<math display="block">Q = K \sqrt{P}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>Q</math> || flow rate from the sprinkler || L/min
|-
| <math>K</math> || K-factor of the sprinkler, set at listing || L/(min·bar1/2)
|-
| <math>P</math> || pressure at the sprinkler || bar
|}

The relation applies to a single sprinkler orifice and uses the pressure acting at the orifice.<ref name="FL-75" /> A standard sprinkler with a nominal orifice of ½ inch has a K-factor of 5.6 in US units, gallons per minute per psi1/2.<ref name="FL-75" /> In SI units this corresponds to about 80.7 L/(min·bar1/2). In Fleming's example, a sprinkler covers 15.6 m² at a density of 6.1 mm/min, so it needs at least 95.4 L/min.<ref name="FL-75" /> The pressure required is then (95.4/80.7)² ≈ 1.4 bar, the value Fleming gives.<ref name="FL-75" />

The calculation then works back towards the supply. Each step adds the friction and elevation losses, so the required pressure and flow grow towards the riser. Sprinklers nearer the supply discharge more than the minimum, because they receive a higher pressure.<ref name="FL-75" /> NFPA 13 also requires a pressure of at least 0.48 bar at the end sprinkler, to ensure a proper spray pattern.<ref name="FL-75" />

The result is compared with the [[fire water supply]]. A test of a public main gives a static pressure without flow and a residual pressure at a known flow. Between these points, a supply curve can be drawn, and the system demand, including an allowance for fire service hoses, must lie below it.<ref name="FL-81">Fleming (2002), p. 4-81.</ref> In the WTC towers, standard ½ inch sprinklers with a K-factor of 5.6 and a temperature rating of 165 °F (about 74 °C) were used in most areas.<ref name="EV-23">Evans et al. (2005), p. 23.</ref>

== Performance and reliability ==
Sprinkler systems are designed for a density over a design area that usually includes 10 to 15 sprinklers. In practice, incident data show that more than half of all fires are controlled or extinguished by one or two sprinklers.<ref name="HO-18">Hopkins et al. (2005), p. 18.</ref> A study of fires in New York City buildings from 1969 to 1978 gives the following distribution:<ref name="HO-19">Hopkins et al. (2005), p. 19.</ref>

[[File:fse-wet-pipe-sprinkler-system-v2.svg|thumb|upright=1.5|alt=Step chart of cumulative share of fires controlled against number of sprinklers, rising from 65 and 57 % at one sprinkler to 97 and 93 % at six.|Cumulative share of fires controlled or extinguished by a given number of operating sprinklers, New York City, 1969 to 1978. Based on Hopkins et al. (2005), p. 19.]]

{| class="wikitable"
|+ Cumulative share of fires controlled or extinguished, New York City, 1969–1978<ref name="HO-19" />
! Sprinklers operating !! High-rise buildings (%) !! Low-rise buildings (%)
|-
| 1 || 65.4 || 57.1
|-
| 2 or fewer || 84.5 || 74.3
|-
| 3 or fewer || 91.3 || 82.3
|-
| 6 or fewer || 97.3 || 92.6
|}

The Australian Building Codes Board splits sprinkler effectiveness into two parts. Reliability is the share of fires, large enough to activate a sprinkler, in which the system operates. Efficacy is the share of operating cases in which the system contains and controls the fire until the fire brigade arrives.<ref name="AB-69">Australian Building Codes Board (2020a), Data Sheet C2, p. 69.</ref><ref name="AB-70" /> Effectiveness is the product of the two.<ref name="AB-70" /> If efficacy is achieved, flashover is assumed to be prevented and the fire is unlikely to spread beyond its room.<ref name="AB-74">Australian Building Codes Board (2020a), Data Sheet C2, p. 74.</ref> Based on US statistics for wet pipe systems, the Board gives these typical design values:<ref name="AB-71" />

[[File:fse-wet-pipe-sprinkler-system-v5.svg|thumb|upright=1.5|alt=Grouped bars for residential and general buildings: reliability 95 and 90 %, efficacy 97 and 96 %, effectiveness 92 and 86 % with low-high whiskers.|Reliability, efficacy and effectiveness of wet pipe sprinkler systems used as typical values in Australian performance-based design. Based on Australian Building Codes Board (2020a), Data Sheet C2, p. 71.]]

{| class="wikitable"
|+ Typical values for wet pipe sprinkler systems in Australian performance-based design<ref name="AB-71" />
! Building use !! Reliability (%) !! Efficacy (%) !! Effectiveness, typical (low–high) (%)
|-
| Residential (NCC Classes 2, 3 and 4) || 95 || 97 || 92 (87–97)
|-
| General (NCC Classes 5, 6, 7a, 8 and 9) || 90 || 96 || 86 (81–91)
|}

Most recorded failures have causes outside the sprinklers themselves. In US data, a system that was shut off accounted for 59 % of failures to operate. Water that did not reach the fire caused 51 % of the cases in which an operating system was ineffective, and too little water caused 30 %.<ref name="AB-79">Australian Building Codes Board (2020a), Data Sheet C2, p. 79.</ref> Other causes include manual intervention, lack of maintenance, damaged components and a system unsuited to the fire.<ref name="AB-79" /> NUREG-1805 adds closed supply valves, empty tanks, pumps that fail to start, corroded or painted sprinklers and inadequate water supplies.<ref name="IS-10-2" />

A fault tree study of Australian high-rise offices found reliability values between 86.6 and 97.9 %. Isolation of sprinkler zones during tenancy changes and out-of-specification sprinklers were the main factors. A zone isolation valve on each floor reduced the failure probability to between 10 and 63 % of its original value.<ref name="AB-81">Australian Building Codes Board (2020a), Data Sheet C2, p. 81.</ref> In probabilistic risk assessment for US nuclear power plants, NUREG/CR-6850 cites an unreliability of 0.02 for wet pipe systems, against 0.05 for deluge and preaction systems.<ref name="NO-P6">Nowlen et al. (2005b), p. P-6.</ref> A wet pipe system needs no signal from a detection system to operate, so a detection failure does not affect its performance.<ref name="NO-P5">Nowlen et al. (2005b), p. P-5.</ref>

== Use in fire safety engineering ==
In performance-based design, a wet pipe system affects both the design fire for life safety and the fire load used for structural design. In New Zealand, C/VM2 (second edition, 2025) treats a fire in a sprinklered building as controlled at a constant heat release rate from the moment a sprinkler activates, for life safety analysis.<ref name="MB-20">Ministry of Business, Innovation and Employment (2025b), p. 20.</ref> Activation is calculated with prescribed sprinkler properties. A standard response sprinkler has a response time index of 135 m1/2s1/2, an activation temperature of 68 °C and a radial distance of 3.25 m. A quick response sprinkler has a response time index of 50 m1/2s1/2.<ref name="MB-27">Ministry of Business, Innovation and Employment (2025b), p. 27.</ref>

For structural design, C/VM2 assumes burnout of the firecell without sprinkler intervention. The fire load energy density may, however, be multiplied by 0.5 for a sprinklered firecell for several applications, including the fire duration.<ref name="MB-20" /><ref name="MB-23">Ministry of Business, Innovation and Employment (2025b), p. 23.</ref> The maximum fire resistance rating that needs to be specified is 180/180/180 for a sprinklered firecell, against 240/240/240 for an unsprinklered one.<ref name="MB-18">Ministry of Business, Innovation and Employment (2025b), p. 18.</ref> In Australia, the verification method compares a proposed design with a reference building. Its sprinkler effectiveness values are used with a sensitivity analysis between the low and high estimates.<ref name="AB-71" />

The collapse of World Trade Center 7 in 2001 shows the importance of the water supply. Sprinklers on floors 1 to 20 were supplied directly from the city main through an automatic fire pump. A loss of power or damage to the main could therefore interrupt the supply.<ref name="EV-31">Evans et al. (2005), p. 31.</ref> NIST concluded that if water had been available and the sprinklers had operated as designed, the fires would probably have been controlled and the collapse prevented.<ref name="NI-xxxvi">National Institute of Standards and Technology (2008), p. xxxvi.</ref>

== Limitations and misconceptions ==
A common misconception is that all sprinklers in a building open when a fire breaks out. In reality, only those exposed to enough heat operate and discharge water.<ref name="HO-14" /> A related misconception is that a sprinkler system is a fire suppression system. Hopkins and co-workers call this a misnomer for most office installations, which are designed for control.<ref name="HO-15" />

The water-filled pipes are both the strength and the weakness of the system. The pipes must be kept above freezing; in the United States, NFPA 13 requires at least 40 °F (about 4 °C) wherever wet pipe systems are installed.<ref name="HO-14" /> A broken sprinkler or a failed pipe or fitting releases water onto building contents that may be sensitive to water damage.<ref name="IS-10-1" /> Water is also unsuitable for materials that react with it, and burning hydrocarbons can float on water and keep burning.<ref name="FL-72" />

A controlled fire is not a clean fire. A fire held back by sprinklers burns less completely and produces more soot per kilogram of fuel. Because less fuel burns, the total amount of smoke still remains limited.<ref name="VR-6.2">Van Rossum (2027), §6.2.</ref> Finally, a sprinkler system can be overwhelmed or disabled. NIST therefore stresses that structures should be designed for fires in which sprinklers are absent, do not function or are overwhelmed.<ref name="NI-xxxvi" /> In the US data, a system that was shut off accounted for 40 % of all effectiveness failures, against 7 % for damaged system components.<ref name="AB-79" />

== See also ==
* [[Deluge system]]
* [[Sprinkler hazard classification]]
* [[Sprinkler design density]]
* [[Response time index]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* Grill, R. A., & Johnson, D. A. (2005a). ''Documentation of the fuel system for emergency power in World Trade Center 7'' (NIST NCSTAR 1-1J). National Institute of Standards and Technology.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2008). ''Final report on the collapse of World Trade Center Building 7'' (NIST NCSTAR 1A).
* Nowlen, S. P., Najafi, B., Wyant, F. J., Forester, J., Kazarians, M., Kolaczkowski, A., Joglar, F., Funk, D., Hannaman, G., & Anoba, R. (2005b). ''EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Vol. 2. Detailed methodology'' (NUREG/CR-6850; EPRI 1011989). Electric Power Research Institute; U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Fire Suppression]]

Flame detector

2026-11-02T17:13:43Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''flame detector''' is a fire detector that is activated by the electromagnetic radiation emitted by flames.<ref name="IS-F-30">Iqbal and Salley (2004), p. F-30.</ref> The common types sense ultraviolet (UV) or infrared (IR) radiation, and some units combine both.<ref name="IFEG-2.7-2">National Research Council of Canada et al. (2005), p. 2.7-2.</ref><ref name="Stroup-3-23">Stroup et al. (2016), p. 3-23.</ref> Unlike a [[smoke detector]] or a [[heat detector]], a flame detector does not wait for combustion products or hot gases to reach it.<ref name="SMC-4-4">Schifiliti et al. (2002), p. 4-4.</ref> It responds to radiation that travels in a straight line from the flame, so it must have a clear view of the fire.<ref name="SMC-4-30">Schifiliti et al. (2002), p. 4-30.</ref>

Flame detectors often start fast-response extinguishing or explosion suppression systems, which makes a reliable response within the required time essential.<ref name="SMC-4-30" /> The SFPE Handbook gives the example of infrared detectors that must find a 1 m² pool fire of jet fuel in a 20 m high aircraft hangar.<ref name="SMC-4-31">Schifiliti et al. (2002), p. 4-31.</ref> UV detectors give the fastest response of the detectors used to trigger suppression of gas explosions.<ref name="Zalosh-3-417">Zalosh (2002), p. 3-417.</ref> The range of a flame detector depends on the size of the fire, the distance, the angle to the detector's axis and the absorption of radiation by the air.<ref name="SMC-4-31" /> Designers therefore work with manufacturer curves of fire size against distance, which are based on the inverse square law.<ref name="SMC-4-30" /><ref name="SMC-4-31" /> For analysis and design, the International Fire Engineering Guidelines (2005) refer designers to the relevant standards, handbooks and manufacturer data.<ref name="IFEG-2.7-6">National Research Council of Canada et al. (2005), p. 2.7-6.</ref>

== How flame detectors work ==
Every fire detector senses one or more fire signatures. The SFPE Handbook lists these signatures as radiation, invisible and visible products of combustion, a rapid change in temperature and a high temperature.<ref name="SMC-4-4" /> Smoke detectors respond to the products of combustion and heat detectors respond to temperature.<ref name="SMC-4-4" /> Ultraviolet and infrared detectors are the only types in that list that respond to radiation.<ref name="SMC-4-4" /> A flame detector is therefore one part of a [[fire alarm system]] that can sense a fire at a distance, before any heat or smoke reaches the ceiling.<ref name="SMC-4-4" /><ref name="SMC-4-30" />

The radiation that a flame detector senses is the same [[thermal radiation]] that transfers heat from a fire. Radiation is emitted by the surface of a hot body, which can be a solid, a liquid or a gas body such as a flame or a smoke layer.<ref name="VR-5.4">Van Rossum (2027), §5.4.</ref> At the temperatures that matter in fires, most of this radiation lies in the infrared and part of it in the visible range; ultraviolet usually plays a small role in heat transfer.<ref name="VR-5.4" /> Gases between the fire and a receiving surface can absorb part of the radiation.<ref name="VR-5.4" />

Fire detection devices operate in only one of three bands of the spectrum.<ref name="SMC-4-30" /> The SFPE Handbook gives ultraviolet as 0.1 to 0.35 micrometres, visible light as 0.35 to 0.75 micrometres and infrared as 0.75 to 220 micrometres.<ref name="SMC-4-30" /> Its table of fire signatures assigns the ultraviolet detector to radiation with a wavelength of 1700 to 2900 ångström, that is 0.17 to 0.29 micrometres.<ref name="SMC-4-4" /> A sensor reacts only to specific wavelengths, so the fuel must emit radiation within the sensor's band.<ref name="SMC-4-30" /> An infrared detector with a single sensor tuned to 4.3 micrometres, the emission peak of carbon dioxide, cannot be expected to detect a fire of a fuel that contains no carbon.<ref name="SMC-4-30" />

[[File:fse-flame-detector-v1.svg|thumb|upright=1.5|alt=Horizontal wavelength bar divided into ultraviolet, visible and infrared bands, with the UV detector band and the 4.3 micrometre carbon dioxide peak marked.|The three bands used by flame detectors, with the ultraviolet detector band and the 4.3 µm carbon dioxide emission peak. Based on Schifiliti et al. (2002), pp. 4-4 and 4-30.]]

== Types and selection ==
The International Fire Engineering Guidelines name ultraviolet and infrared flame detectors as the two main examples.<ref name="IFEG-2.7-2" /> Combination ultraviolet/infrared optical flame detectors also exist; they were among the devices tested in large US Navy aircraft hangars, 15 m and 22 m high.<ref name="Stroup-3-23" /> Fires of jet fuel in those tests ranged from 100 kW to about 33 MW.<ref name="Stroup-3-23" />

Selecting a sensor depends on the fuel, the growth rate of the fire, the ambient conditions and the control or extinguishing function that the detector must start.<ref name="SMC-4-30" /> The designer must also consider absorption of radiation by the air, other sources of radiation in the area, the size of the spark, ember or fire to be detected, and the distance to the sensor.<ref name="SMC-4-30" /> In the hangar example of the SFPE Handbook, infrared detection at 4.3 micrometres suits the carbon-based jet fuel.<ref name="SMC-4-31" /> Infrared detectors were also expected to cover a larger area per device than ultraviolet detectors and so to cost less in that case.<ref name="SMC-4-31" />

Sources of radiation that are not fires can cause false alarms. Resistance to such false responses can be obtained by filtering, by using several sensors in one unit, such as a two- or three-channel detector, or by both.<ref name="SMC-4-31" /> A system can also use coincident detection, in which two or more detectors must operate before an alarm is accepted.<ref name="IFEG-2.7-2" /> In explosion suppression, the choice between pressure sensors and ultraviolet detectors depends on the required response time and on concerns about spurious activation and about the detectors' view being blocked.<ref name="Zalosh-3-417" />

== Detection distance and the inverse square law ==
The radiant power that reaches a sensor falls quickly as the distance to the fire grows. Standard practice for the design of radiant energy detection is based on curves of fire size against distance, which are derived from the inverse square law.<ref name="SMC-4-30" /> The SFPE Handbook gives the relation as:<ref name="SMC-4-31" />

<math display="block">S = \frac{k\,P\,e^{-\zeta d}}{d^{2}}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>S</math> || radiant power reaching the detector || W
|-
| <math>k</math> || proportionality constant for the detector || –
|-
| <math>P</math> || radiant power emitted by the fire || W
|-
| <math>\zeta</math> || extinction coefficient of air || m−1
|-
| <math>d</math> || distance between the fire and the detector || m
|}

The handbook prints the exponential term without its minus sign.<ref name="SMC-4-31" /> The minus sign used here follows from the handbook's own statement that humidity and dust absorb radiation and so limit the radiation that reaches the sensor.<ref name="SMC-4-31" /> Manufacturers test a sensor with several fuels and plot the normalised fire size against the normalised distance.<ref name="SMC-4-31" /> The resulting curve gives the largest distance at which the sensor consistently detects a fire of a given size, usually stated in square metres of burning area.<ref name="SMC-4-31" />

As a worked example, the hangar case in the SFPE Handbook uses a detector that sees a 1.0 m² fire at 40 m on its axis.<ref name="SMC-4-32">Schifiliti et al. (2002), p. 4-32.</ref> In clear air, with the absorption term equal to one, halving the distance to 20 m makes the received power four times larger, because it scales with one over the distance squared. The other way round, a fire that emits only half the radiant power reaches the same signal only at 40/√2, about 28 m. Real air absorbs part of the radiation, so these clear-air values are upper limits for the reach of the detector.<ref name="SMC-4-31" />

[[File:fse-flame-detector-v2.svg|thumb|upright=1.5|alt=Falling curve of relative radiant power against distance from 10 to 60 m, equal to 1 at 40 m and 4 at 20 m.|Relative radiant power at the detector against distance in clear air, set to 1 at 40 m. Halving the distance gives four times the power. Based on Schifiliti et al. (2002), pp. 4-31 to 4-32.]]

The handbook then applies typical reductions to the detection distance of the hangar detectors, which have a field of view of 45 degrees.<ref name="SMC-4-32" /> A 30% loss of sensitivity at the edge of the field of view reduces the distance for the 1.0 m² fire from 40 m to 28 m.<ref name="SMC-4-32" /> A further loss of 3% at an ambient temperature of 50 °C gives about 26.8 m, and another 3% at 90% relative humidity gives about 25.6 m.<ref name="SMC-4-32" /> The largest viewing distance in the layout at 45 degrees is 20 m, so the design is considered valid.<ref name="SMC-4-32" /> Had the distance dropped below 20 m, a different layout or different devices would have been needed.<ref name="SMC-4-32" />

[[File:fse-flame-detector-v3.svg|thumb|upright=1.5|alt=Hangar cross-section with flame detectors in the upper corners, their 45 degree fields of view and a 1 square metre pool fire on the floor.|Infrared flame detectors in an aircraft hangar: the detection distance for a 1.0 m² fire falls from 40 m on axis to about 25.6 m at the edge of the view, still above the 20 m needed. Based on Schifiliti et al. (2002), pp. 4-31 to 4-32.]]

== Design of a flame detection system ==
Fire detection systems serve goals of life safety, property protection, business protection and environmental protection.<ref name="SMC-4-1">Schifiliti et al. (2002), p. 4-1.</ref> Detectors may also start other fire protection systems, such as special extinguishing systems and smoke control systems.<ref name="SMC-4-2">Schifiliti et al. (2002), p. 4-2.</ref> When a flame detector activates a fast extinguishing or control system, such as an explosion suppression system, it must respond reliably within the required time.<ref name="SMC-4-30" />

The SFPE Handbook describes the design in a fixed order.<ref name="SMC-4-31" /> First a detector type is chosen that matches the fuel. Then the possible sources of interfering radiation are identified and a device is selected that resists them.<ref name="SMC-4-31" /> Next, the manufacturer's data give the mounting height and lateral distance limits for the critical fire size.<ref name="SMC-4-31" /> The layout must then account for all obstructions so that every part of the space is watched.<ref name="SMC-4-31" /> Finally, the reductions in sensitivity due to viewing angle, temperature and humidity are checked against the actual viewing distances.<ref name="SMC-4-31" /><ref name="SMC-4-32" />

The International Fire Engineering Guidelines describe a flame detector as a point that receives radiation from a flame with a specific flame temperature and emissivity.<ref name="IFEG-2.7-6" /> The intensity of the radiation it receives can be calculated with the methods for radiation from fires in the same guidelines.<ref name="IFEG-2.7-6" /> The sensitivity varies with the direction of the incoming radiation, and this off-axis sensitivity must be considered in design.<ref name="IFEG-2.7-6" /> A method for estimating flame height is also needed for the operation of flame detectors.<ref name="IFEG-2.4-6">National Research Council of Canada et al. (2005), p. 2.4-6.</ref> In New Zealand, the fixed detector criteria of C/VM2 (second edition, 2025) do not cover flame detectors.<ref name="MBIE25-27">Ministry of Business, Innovation and Employment (2025b), p. 27.</ref> Its table gives activation criteria only for heat detectors, smoke detectors and sprinklers.<ref name="MBIE25-27" />

== Applications ==
The sources describe flame detectors in large hangars, in process plants and in explosion suppression systems. The SFPE Handbook uses an aircraft hangar of 50 m by 80 m with a 20 m ceiling, where a fuel spill fire must activate a fixed suppression system.<ref name="SMC-4-31" /> In the chemical process industries, smaller leaks may be detected by gas or flame detectors, if these are installed.<ref name="Barry-5-184">Barry (2002), p. 5-184.</ref> A quantitative risk assessment should then establish the position, effectiveness and reliability of these detectors.<ref name="Barry-5-184" /> It should also describe in detail how gas and flame detection start the emergency shutdown system.<ref name="Barry-5-182">Barry (2002), p. 5-182.</ref>

In explosion suppression systems, detectors, a power supply and control system and fast extinguishing units work together.<ref name="Zalosh-3-417" /> Dust explosion applications usually use pressure sensors, while gas explosion applications can use pressure sensors or ultraviolet detectors.<ref name="Zalosh-3-417" /> Of these, ultraviolet detectors give the fastest response.<ref name="Zalosh-3-417" /> The control system includes interlocks that monitor the system itself, shut down processes and trigger alarms when the system activates.<ref name="Zalosh-3-417" />

Detectors in such plants may be installed where a flammable gas atmosphere can form, and one way to prevent an explosion there is to exclude ignition sources.<ref name="Beyler-2-181">Beyler (2002b), p. 2-181.</ref> Low-power equipment can be designed so that even its worst fault cannot release the minimum ignition energy of a given gas; such equipment is called intrinsically safe.<ref name="Beyler-2-181" /> Where this is not possible, the equipment can be housed in an explosion-proof enclosure that stops a flame from spreading out of it.<ref name="Beyler-2-181" /> The fuel oil tank room of World Trade Center 7 is an example. All its electrical devices were explosion proof, suitable for Class 1, Division 1, Group C in the [[hazardous area classification]] of the New York City building code.<ref name="GJ-27">Grill and Johnson (2005a), p. 27.</ref>

== Limits ==
A flame detector is mainly a line-of-sight device.<ref name="SMC-4-30" /> A fire hidden behind equipment, stored goods or structure may not be seen, which is why the layout must consider all obstructions.<ref name="SMC-4-31" /> Sensitivity decreases as the fire moves away from the optical axis: a fire detectable at a given distance on the axis may not be detectable at the same distance 30 degrees off the axis.<ref name="SMC-4-31" />

The surroundings also matter. Humidity and dust absorb radiation, and a higher ambient temperature can lower the relative sensitivity of the sensor.<ref name="SMC-4-31" /> Even a small decrease can affect the response to the expected fire.<ref name="SMC-4-31" /> Other radiation sources in the area, or air that absorbs a large fraction of the radiation, may change the type, location and spacing of the sensors.<ref name="SMC-4-30" />

The detector also has to match the fuel. A sensor that reacts only to a narrow band, such as the carbon dioxide peak at 4.3 micrometres, will not detect fuels that do not emit in that band.<ref name="SMC-4-30" /> Finally, the performance data of flame detectors come largely from manufacturers.<ref name="SMC-4-31" /> The International Fire Engineering Guidelines therefore advise taking relevant standards, handbooks and manufacturer data into account in analysis and design.<ref name="IFEG-2.7-6" />

== See also ==
* [[Deluge system]]
* [[Beam smoke detector]]
* [[Multi-sensor detector]]
* [[Detector activation time]]
* [[Aspirating smoke detection]]

== Notes ==
<references />

== References ==
* Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
* Beyler, C. L. (2002b). Flammability limits of premixed and diffusion flames. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-172–2-187). National Fire Protection Association; Society of Fire Protection Engineers.
* Grill, R. A., & Johnson, D. A. (2005a). ''Documentation of the fuel system for emergency power in World Trade Center 7'' (NIST NCSTAR 1-1J). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Stroup, D., Lindeman, A., McGrattan, K., Peacock, R., Overholt, K., Joglar, F., LeStrange, S., & Montanez, S. (2016). ''Verification and validation of selected fire models for nuclear power plant applications: Supplement 1'' (NUREG-1824, Supplement 1; EPRI 3002002182). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Zalosh, R. (2002). Explosion protection. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-402–3-421). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Detection & Alarm]]

BLEVE

2026-11-02T17:13:41Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''BLEVE''' ('''boiling liquid expanding vapour explosion''') is the sudden failure of a container holding a liquid above its normal boiling point, which then flashes into vapour while the container breaks apart.<ref name="IS-5-15">Iqbal and Salley (2004), p. 5-15.</ref><ref name="IS-F-8">Iqbal and Salley (2004), p. F-8.</ref> Many substances that are gases at ambient conditions, such as liquefied petroleum gas, are stored under pressure as a saturated mixture of liquid and vapour. A rupture of such a vessel releases the stored energy violently.<ref name="IS-5-15" /><ref name="Bey-3-302" /> The typical cause is a fire that heats the part of the shell above the liquid until the metal can no longer withstand the internal pressure.<ref name="IS-5-15" /><ref name="B-5-184">Barry (2002), p. 5-184.</ref>

A BLEVE threatens its surroundings in three ways: a pressure wave, fragments of the vessel that can be thrown considerable distances, and, if the contents are flammable, a short but very intense [[fireball]].<ref name="IS-F-8" /><ref name="IS-5-15" /> In quantitative risk assessment of process plants it is treated as a delayed event, typically following 8 to 15 minutes of fire exposure.<ref name="B-5-184" /> The radiation from the fireball can be estimated from the mass of fuel with a simple empirical relation, and the result is compared with damage thresholds for people and equipment.<ref name="IS-5-15" /><ref name="B-5-186">Barry (2002), p. 5-186.</ref>

== Definition and terminology ==
The acronym stands for "Boiling Liquid, Expanding Vapor Explosion", written with the American spelling in the source documents.<ref name="IS-5-15" /> NUREG-1805 quotes the definition of the National Fire Protection Association (NFPA): the failure of a major container into two or more pieces. The failure occurs at a moment when the contained liquid is above its boiling point at normal atmospheric pressure.<ref name="IS-5-15" /> The glossary of the same report, citing Nolan (2000), describes it as a catastrophic rupture of a pressurised vessel with the simultaneous ignition of the vaporising fluid.<ref name="IS-F-8" /> During the rupture a pressure wave may be produced, and a short, intense fireball occurs if the liquid is flammable.<ref name="IS-F-8" />

The two definitions differ in one respect. The NFPA wording centres on the failure of the container and the state of the liquid, and does not require ignition.<ref name="IS-5-15" /> The glossary wording includes ignition and the fireball that follows when the liquid is flammable.<ref name="IS-F-8" /> Barry uses the term "BLEVE fireball" for this outcome.<ref name="B-5-184" /> Zalosh notes that a rising fireball occurs in most BLEVEs of liquefied fuel gases exposed to severe fire.<ref name="Z-3-419">Zalosh (2002), p. 3-419.</ref>

The failure itself fits the class of physical [[explosion]]s in NUREG-1805, in which the high pressure is generated by mechanical means without any chemical change.<ref name="IS-15-2">Iqbal and Salley (2004), p. 15-2.</ref> Its two examples are the external heating of a tank, which raises the internal pressure until the tank fails, and the sudden release of a superheated liquid that flash-evaporates.<ref name="IS-15-2" /> A BLEVE of a fire-exposed tank combines both. The burning of the released vapour afterwards is a separate, chemical process. ''Introduction to Fire Safety Engineering'' makes the same distinction for ordinary fuels: evaporation is a physical change, and only the vapour formed can burn.<ref name="VR-3.5">Van Rossum (2027), §3.5.</ref>

== How a BLEVE develops ==
In the vessel, the liquid and the vapour above it are stored under pressure in a saturated state.<ref name="IS-5-15" /> External heating of the tank increases the internal pressure.<ref name="IS-15-2" /> Evaporation takes energy from the surroundings.<ref name="VR-3.5" /> The contained liquid therefore acts as a heat absorber, so the wetted part of the shell usually stays relatively cool.<ref name="IS-5-15" /> Only the surfaces around the internal vapour space are at risk.<ref name="IS-5-15" />

[[File:fse-bleve-v1.svg|thumb|upright=1.5|alt=Cross-section of a horizontal gas tank above a fire: the lower wall touches liquid and stays cooler, the upper wall touches vapour and overheats.|A liquefied gas vessel exposed to fire: the liquid cools the wetted shell, while the shell around the vapour space heats up and weakens. Based on Iqbal and Salley (2004), p. 5-15.]]

NUREG-1805, citing Nolan (1996), states that a BLEVE typically occurs in a metal container that has been overheated above 538 °C (1,000 °F).<ref name="IS-5-15" /> At that temperature the metal may no longer withstand the internal stress, and failure occurs.<ref name="IS-5-15" /> Most BLEVEs occur when containers are less than one half to one third full of liquid, that is, when a large part of the shell is not wetted.<ref name="IS-5-15" />

Barry describes the sequence as it is used in risk studies. A primary fire overheats the pressurised vessel; this raises the internal pressure and weakens the shell, until the shell bursts and releases its contents as a large and very intense fireball.<ref name="B-5-184" /> The BLEVE is generally considered a delayed event, with 8 to 15 minutes of fire exposure before failure.<ref name="B-5-184" /> Among the possible exposing fires, an impinging [[jet fire]] is especially severe, far more so than a pool fire.<ref name="Bey-3-302" /> Large-scale tests measured heat fluxes of up to 250 kW/m² for two-phase propane jets and 300 kW/m² for sonic natural gas jets.<ref name="Bey-3-302">Beyler (2002a), p. 3-302.</ref> On fully engulfed cylindrical targets the average heat flux was 150 kW/m² for two-phase liquefied petroleum gas (LPG) jets and 200 kW/m² for natural gas jets.<ref name="Bey-3-302" />

[[File:fse-bleve-v2.svg|thumb|upright=1.5|alt=Flowchart: fire exposure heats the vessel, pressure rises and the shell weakens, the vessel bursts, the liquid flashes, giving blast, fragments and a fireball.|Typical BLEVE sequence: a primary fire heats the vessel for several minutes until the shell bursts; the flashing liquid then causes a blast, fragments and, for flammable contents, a fireball. Based on Barry (2002), p. 5-184, and Iqbal and Salley (2004), pp. 5-15 and F-8.]]

Fire is not the only route to failure. A container can be damaged by impact, after which a crack grows because of the internal pressure, brittleness of the vessel material, or both.<ref name="IS-5-15" /> The metal can also be weakened by corrosion, internal overheating or manufacturing defects, until large cracks form or the vessel separates into two or more pieces.<ref name="IS-5-15" />

When the shell of a vessel with a pressurised cargo opens, part of the released liquid flash-evaporates and a vapour cloud forms suddenly.<ref name="Bey-3-302" /> The phase change produces blast waves whose energy equals the change in internal energy of the liquid and vapour.<ref name="IS-5-15" /> If the vapour puff is ignited immediately after it forms, it burns as a rising sphere, the fireball.<ref name="Bey-3-302" />

== Hazards: blast, fragments and fireball ==
The blast wave from a BLEVE is one example of a breached-vessel blast.<ref name="Z-3-419" /> For a vessel holding a gas that does not ignite, ideal blast wave correlations based on the expansion energy of the gas can be used. BLEVEs of liquefied gases need their own correlations for blast and fragments.<ref name="Z-3-419" /> Zalosh refers to the guidelines of the Center for Chemical Process Safety for these correlations; they are not reproduced in the SFPE handbook.<ref name="Z-3-419" />

Fragments are a hazard in their own right. NUREG-1805 lists the container becoming a projectile as the first hazard, before the fireball, and notes that pieces of the vessel may be thrown considerable distances.<ref name="IS-5-15" /><ref name="IS-F-8" /> In quantitative risk assessment, "tank or equipment rupture or fragmentation" appears as a separate outcome under explosion overpressures.<ref name="B-5-182">Barry (2002), p. 5-182.</ref>

For flammable contents, the fireball and its radiation are added to these hazards. The rapid failure of the container is followed by a fireball or major fire, which produces a powerful radiant heat flux.<ref name="IS-5-15" /> Fireballs from large accidents are usually of the order of 100 m in diameter and last only a few seconds, because the fuel mixes rapidly with the surrounding air.<ref name="Bey-3-303">Beyler (2002a), p. 3-303.</ref> During that short time a fraction of the combustion energy is radiated, and the harm to people and property depends on both the intensity and the duration of the radiation.<ref name="Bey-3-303" /> The maximum size is governed mainly by the mass of fuel released and vaporised and grows with the cube root of that mass; it does not depend on the storage pressure, provided the fuel vaporises.<ref name="Bey-3-306">Beyler (2002a), p. 3-306.</ref> For propane at or above about 20 °C, complete vaporisation is expected.<ref name="Bey-3-306" />

A small test shows the time scale. When a cylinder holding 7.5 kg of LPG burst, the fireball reached a diameter of about 10 m within roughly 0.2 s. It stayed between about 10 and 12 m until the record ends at 1.8 s.<ref name="Bey-3-307">Beyler (2002a), p. 3-307.</ref> The test went through three stages. First came rapid growth with a very bright flame, then a phase with little change in size in which the flame cooled from bright yellow to dull orange. In the final phase the fireball cooled rapidly and lifted off from the ground.<ref name="Bey-3-307" />

The fraction of the heat release that a fireball radiates is generally between 0.1 and 0.4.<ref name="Bey-3-307" /> Large fireballs rise during their lifetime: for very large masses, above about 30,000 kg, even pressurised releases are dominated by buoyancy, and the fireball lifts off from the ground.<ref name="Bey-3-307" /> The height of the fireball centre therefore matters when the distance to a target is worked out.<ref name="IS-5-16" />

== Estimating the radiation from a BLEVE fireball ==
NUREG-1805 names four parameters that are often used to judge a fireball's thermal radiation hazard: the mass of fuel involved and the fireball's diameter, duration and surface emissive power.<ref name="IS-5-15" /> For an object relatively far from the fireball, it gives a single empirical relation, after Hasegawa and Sato (1977) and Roberts (1982).<ref name="IS-5-15" /> In words: the radiant heat flux at the target grows with the mass of fuel vapour to the power 0.771. It falls with the square of the distance from the centre of the fireball.

<math display="block">\dot q''_\mathrm{r} = \frac{828\, m_\mathrm{F}^{0.771}}{R^2}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\dot q''_\mathrm{r}</math> || thermal radiation from the fireball received at the target || kW/m²
|-
| <math>m_\mathrm{F}</math> || mass of fuel vapour in the fireball || kg
|-
| <math>R</math> || distance from the centre of the fireball to the target || m
|}

The relation is dimensional: the constant 828 only gives the right result with the units in the table.<ref name="IS-5-15" /> The distance ''R'' is measured from the centre of the fireball, which lies above the ground; NUREG-1805 gives a separate method to find the fireball height from the volume of fuel vapour.<ref name="IS-5-16">Iqbal and Salley (2004), p. 5-16.</ref> The relation applies to targets "relatively distant" from the fireball.<ref name="IS-5-15" /> Close to a fire, point-source radiation methods overestimate the incident flux, because the near field depends strongly on flame size, shape and target orientation. NUREG-1805 presents them as suitable where a conservative estimate is acceptable.<ref name="IS-5-17">Iqbal and Salley (2004), p. 5-17.</ref>

'''Worked example.''' Take the LPG cylinder test above, with ''m''F = 7.5 kg.<ref name="Bey-3-307" /> The numerator is 828 × 7.50.771 ≈ 3,915. At ''R'' = 20 m from the centre of the fireball:

<math display="block">\dot q''_\mathrm{r} = \frac{828 \times 7.5^{0.771}}{20^2} = \frac{3915}{400} \approx 9.8\ \mathrm{kW/m^2}</math>

At 30 m the flux is about 4.4 kW/m², and at 40 m about 2.5 kW/m².<ref name="IS-5-15" /> The fluxes are loads on the target, so they are rounded up. The same relation shows that the flux falls to 12.5 kW/m² at about 18 m, to 4.0 kW/m² at about 32 m and to 1.6 kW/m² at about 50 m (distances rounded up).<ref name="IS-5-15" />

[[File:fse-bleve-v3.svg|thumb|upright=1.5|alt=Curve of radiant heat flux falling from 39 kW/m² at 10 m to about 1 kW/m² at 60 m, crossing threshold lines at 12.5, 4.0 and 1.6 kW/m².|Radiant heat flux from the fireball of 7.5 kg of LPG against distance from the fireball centre, with damage thresholds. Based on Iqbal and Salley (2004), p. 5-15, Beyler (2002a), p. 3-307, and Barry (2002), p. 5-186.]]

These values are read against damage criteria from observations of large fires. In the table used by Barry, 12.5 kW/m² causes first-degree burns in 10 s and 1 % lethality in 1 minute. A flux of 4.0 kW/m² causes pain if the exposure lasts longer than 20 s, and 1.6 kW/m² causes no discomfort even for long exposure.<ref name="B-5-186" /> The criteria are given for exposure times of 10 s or longer, while the fireball of the example lasted less than 2 s.<ref name="B-5-186" /><ref name="Bey-3-307" /> The comparison therefore shows the importance of duration: the relation gives the intensity, but the harm also depends on how long the exposure lasts.<ref name="Bey-3-303" />

== Incidents ==
Rail and road accidents with liquefied gases illustrate the scale of BLEVE fireballs. A report of the U.S. General Accounting Office cites a tractor-semitrailer carrying 34 m³ (9,000 gal) of LPG; about 2 minutes after the accident a fireball about 123 m in diameter was observed.<ref name="Bey-3-302" /> Its radiant heat burned several people, a house, several other buildings and about 12 acres of woods.<ref name="Bey-3-302" />

In Oneonta, New York, a freight train of 27 cars derailed; 7 of them carried liquid propane, 120 m³ (33,000 gal) in total.<ref name="Bey-3-302" /> Seconds after the derailment a large fireball erupted, believed to result from a tank car that split open.<ref name="Bey-3-302" /> The fireball then heated other tank cars, which failed in several BLEVEs.<ref name="Bey-3-302" /> The case shows how one fireball can heat neighbouring vessels until they fail in turn.<ref name="Bey-3-302" />

In Crescent City, Illinois, 15 cars of a freight train derailed, 10 of them each carrying 130 m³ (34,000 gal) of LPG.<ref name="Bey-3-302" /> A tank car that rode over the pile tore a hole in another car, and the released propane produced the first fireball.<ref name="Bey-3-302" /> Several later explosions followed over a period of hours and destroyed 24 dwellings and 18 businesses.<ref name="Bey-3-303" />

== Use in risk assessment and protection ==
In quantitative risk assessment for the process industries, a "fire ball (BLEVE)" is one of the radiant heat outcomes considered. The others are the [[pool fire]], the torch fire and the flash fire.<ref name="B-5-182" /> The generic event trees of the World Bank, as presented by Barry, lead to a fireball when a release is instantaneous and ignited immediately.<ref name="B-5-183">Barry (2002), p. 5-183.</ref> For pressurised storage and transport vessels, the typical failure case listed is "BLEVE (non-buried case only)", with a total rupture that is ignited.<ref name="B-5-185">Barry (2002), p. 5-185.</ref> The results are drawn as effect zones around the release point. An example shows fatality and injury radii for a BLEVE fireball next to the zones for a torch fire, a flash fire and a vapour cloud explosion.<ref name="B-5-186" /><ref name="B-5-187">Barry (2002), p. 5-187.</ref>

The mechanism shows where the weak spot of a vessel lies: the shell around the vapour space, which is not cooled by the liquid.<ref name="IS-5-15" /> For high-hazard locations where a fire develops quickly and reaches very high temperatures, for example with highly flammable fuels, NUREG-1805 notes that a [[deluge system]] is specified.<ref name="IS-F-59">Iqbal and Salley (2004), p. F-59.</ref> In such a system all sprinklers or spray nozzles are open, and water flows from all of them at once when a detection system opens the deluge valve.<ref name="IS-F-59" /><ref name="IS-10-2">Iqbal and Salley (2004), p. 10-2.</ref> The documents used here do not give water application rates for the cooling of vessels.

The BLEVE also appears in fire service procedures. A 1994 hazardous materials response plan of the Fire Department of the City of New York is reproduced in the NIST report on the World Trade Center response. It requires an evaluation before personnel are committed in potentially explosive situations.<ref name="LV-A">Lawson and Vettori (2005), Appendix A.</ref> Among these it lists gas release or over-pressurisation of containers, with BLEVE given in brackets.<ref name="LV-A" />

== Limits of the simple approach ==
The fireball relation above is an empirical correlation and treats the fireball as a point source.<ref name="IS-5-15" /><ref name="IS-5-17" /> It ignores the growth, rise and lift-off of the fireball, the angle between the target and the line of sight, and the absorption of radiation by the atmosphere. More detailed point-source and spherical models in the SFPE handbook include the radiative fraction, the atmospheric transmissivity and the geometry of the target.<ref name="Bey-3-307" /> These belong to the topic of fireball modelling rather than to the BLEVE itself.

The blast and fragment effects are harder to quantify with the documents used here. Zalosh only refers to separate guidelines for BLEVE blast and fragment correlations.<ref name="Z-3-419" /> The failure temperature and fill level in NUREG-1805 are typical values quoted from the literature, not limits. They describe where most BLEVEs occur, not a temperature or level below which a vessel is safe.<ref name="IS-5-15" /> Finally, Barry gives the 8 to 15 minutes of exposure as a general characterisation of the BLEVE as a delayed event, not as a calculated failure time for a particular vessel.<ref name="B-5-184" />

== See also ==
* [[Vapour cloud explosion]]
* [[Boilover]]
* [[Deflagration]]
* [[TNT equivalence]]

== Notes ==
<references />

== References ==
* Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
* Beyler, C. L. (2002a). Fire hazard calculations for large, open hydrocarbon fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-268–3-314). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Lawson, J. R., & Vettori, R. L. (2005). ''The emergency response operations'' (NIST NCSTAR 1-8). National Institute of Standards and Technology.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Zalosh, R. (2002). Explosion protection. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-402–3-421). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Explosions & Industrial Fires]]

Deluge system

2026-11-02T17:13:38Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''deluge system''' is a water-based fire suppression system in which every nozzle or sprinkler is permanently open, so that water discharges from all of them at once when a valve in the supply opens.<ref name="FL-72">Fleming (2002), p. 4-72.</ref> The valve, called a deluge valve, is opened by a separate fire detection system rather than by heat at the nozzles.<ref name="FL-72" /> The deluge system is one of the four basic types of [[sprinkler system]], next to the wet pipe, [[Dry pipe sprinkler system|dry pipe]] and preaction systems.<ref name="FL-72" /><ref name="IS-10-1">Iqbal and Salley (2004), p. 10-1.</ref>

Its purpose is to deliver the largest amount of water in the shortest time.<ref name="NO-F59">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-59.</ref> Deluge systems are used against rapidly spreading, high hazard fires, such as fires in highly flammable fuels that develop quickly and reach very high temperatures.<ref name="FL-72" /><ref name="NO-F59" /> Typical applications in the sources include aircraft hangars, transformers, turbine-generator bearings, fan units and water curtains at openings.<ref name="SC-106">Scheffey (2002), p. 4-106.</ref><ref name="IS-H11">Iqbal and Salley (2004), p. H-11.</ref><ref name="IS-H20" /><ref name="HO-132">Hopkins et al. (2005), p. 132.</ref><ref name="IS-F69" /> The price of this speed is a large water demand and extensive water damage, both when the system is needed and when it operates by mistake.<ref name="IS-10-2">Iqbal and Salley (2004), p. 10-2.</ref><ref name="IS-H5">Iqbal and Salley (2004), p. H-5.</ref> In nuclear power plant risk studies, a deluge system is assumed to fail on demand about five times in a hundred.<ref name="NS-P6">Parkinson et al. (1994), as cited in Nowlen et al. (2005b), p. P-6.</ref>

== How a deluge system works ==
A deluge system consists of a network of pipes with open sprinklers or spray nozzles, connected to a water supply through a deluge valve.<ref name="FL-72" /> The open nozzles contain no fusible elements that could hold back the water.<ref name="IS-10-2" /> Instead, the deluge valve holds the water back from the piping until it is activated.<ref name="FL-72" /> The deluge valve sits on the water supply side of the system and opens automatically when the detection system calls for water.<ref name="IS-10-2" /> Once it has opened, water flows into the pipes and discharges from every outlet on the system at the same time.<ref name="NO-F59" /> In the words of the definition, there are no closed sprinklers in a deluge system.<ref name="NO-F59" />

[[File:fse-deluge-system-v1.svg|thumb|upright=1.5|alt=Schematic: a water main feeds a deluge valve; above it a pipe network with open nozzles all spraying; a detector near the ceiling signals the valve.|Main parts of a deluge system. A separate detector opens the deluge valve, after which water flows from every open nozzle at once. Based on Fleming (2002), p. 4-72 and Iqbal and Salley (2004), p. 10-2.]]

The detection devices are installed in the same area as the nozzles.<ref name="NO-F59" /> In the nuclear guidance of the United States Nuclear Regulatory Commission (NRC), the system is described as typically actuated by heat detectors mounted above the open sprinklers.<ref name="IS-10-2" /> Most deluge systems can also be started by hand.<ref name="IS-10-2" /> A glossary in the same report defines the deluge sprinkler system simply as a system with open sprinklers or nozzles, in which all water flows out when the deluge valve actuates.<ref name="IS-F21">Iqbal and Salley (2004), p. F-21.</ref> The single word "deluge" is defined there as the immediate release of a commodity, usually a water spray released for fire suppression.<ref name="NO-F20">Nolan (2000), as cited in Iqbal and Salley (2004), p. F-20.</ref>

A water spray system works on the same principle. It is a fixed pipe system with spray nozzles chosen for a specific discharge and distribution over the surface or area to be protected.<ref name="NO-F59" /> Its control valve is opened automatically by fire detectors in the same area, or by hand.<ref name="NO-F59" /> In the United States, the National Fire Protection Association (NFPA) covers water spray systems in NFPA 15, ''Standard for Water Spray Fixed Systems for Fire Protection''. Foam-water sprinkler and spray systems fall under NFPA 16.<ref name="FL-73">Fleming (2002), p. 4-73.</ref> The same open-nozzle principle is used in total compartment application water mist systems, where water is discharged from all nozzles as soon as a control valve opens.<ref name="MB-327">Mawhinney and Back (2002), p. 4-327.</ref>

== Comparison with other sprinkler systems ==
All sprinkler systems can be placed in one of four basic types, which differ in how the water reaches the area of the fire.<ref name="FL-73" /> In a wet pipe system, the pipes are filled with water under pressure. Heat opens a single sprinkler, which then discharges water over its own area.<ref name="FL-72" /> In a dry pipe system, the pipes contain air or nitrogen under pressure, and water enters only after a sprinkler has opened and the air pressure has dropped.<ref name="FL-72" />

[[File:fse-deluge-system-v2.svg|thumb|upright=1.5|alt=Four panels: wet pipe, dry pipe, preaction and deluge systems, each with a ceiling pipe, four outlets and a fire under one outlet.|The four basic sprinkler system types. Only in the deluge system do all outlets discharge, because its nozzles are open. Based on Fleming (2002), pp. 4-72 to 4-73.]]

These two types use automatic sprinklers, which open one by one in response to heat. Only the sprinklers exposed to enough heat operate; the idea that all sprinklers open in a fire is a common misconception.<ref name="HO-14">Hopkins et al. (2005), p. 14.</ref> For a deluge system that idea is correct, because the whole system discharges together.<ref name="NO-F59" />

The preaction system is the closest relative. It is similar to a deluge system, but it uses automatic sprinklers instead of open ones.<ref name="FL-72" /> A separate detection system opens a deluge valve and fills the pipes, yet water flows only where heat has also opened a sprinkler.<ref name="FL-72" /> The NRC report describes it as a deluge system with closed heads, in which two conditions must be met before water is released.<ref name="IS-10-2" /> Preaction systems are used where accidental discharge of water is a special concern, as in valuable computer rooms.<ref name="FL-73" /> In the World Trade Center towers, the preaction systems used hydraulically operated differential diaphragm deluge valves with manual and electric detectors.<ref name="HO-50">Hopkins et al. (2005), p. 50.</ref>

{| class="wikitable"
|+ The four basic types of sprinkler system<ref name="FL-72" /><ref name="IS-10-2" />
! Type !! Sprinklers !! Pipes before operation !! What releases the water
|-
| Wet pipe || automatic (closed) || water under pressure || heat opens individual sprinklers
|-
| Dry pipe || automatic (closed) || air or nitrogen under pressure || an opened sprinkler lets the air escape, then the dry pipe valve opens
|-
| Preaction || automatic (closed) || usually air at a small pressure || detection opens the valve, heat opens individual sprinklers
|-
| Deluge || open || no water (held back by the deluge valve) || detection or manual action opens the deluge valve; all outlets discharge
|}

The difference also shows up in the hydraulic design. For pressure losses in steel pipe, the [[Hydraulic calculation|Hazen–Williams method]] assigns a roughness coefficient C of 120 to wet and deluge systems, but only 100 to dry and preaction systems.<ref name="FL-76">Fleming (2002), p. 4-76.</ref> A lower value means a rougher pipe and a larger friction loss for the same flow. The values are intended to simulate the expected interior roughness of aged pipe.<ref name="FL-76" />

== Detection and actuation ==
The nozzles of a deluge system cannot sense a fire, so the system relies on a separate detection system.<ref name="FL-72" /> The NRC report names heat detectors above the open sprinklers as the typical choice.<ref name="IS-10-2" /> In aircraft hangars, the systems were historically activated by rapid-response detection systems.<ref name="SC-106" /> The International Fire Engineering Guidelines list heat, smoke, gas sensing and flame detectors as the main detector types.<ref name="IFEG-2">National Research Council of Canada et al. (2005), p. 2.7-2.</ref> A [[flame detector]] is activated by electromagnetic radiation emitted by flames.<ref name="IS-F30">Iqbal and Salley (2004), p. F-30.</ref> It can be treated as a point that receives radiation from the flame, and its sensitivity can vary with the direction of the incoming radiation.<ref name="IFEG-6">National Research Council of Canada et al. (2005), p. 2.7-6.</ref>

A deluge valve can also be tied to more than one detector. In a coincident detection system, two or more detectors must operate before an alarm is recognised. These can be two similar detectors at different locations or two different types of detector.<ref name="IFEG-2" /> The air handling units of the World Trade Center towers used this approach. The first smoke detector to operate shut down the unit, and a second detector within the unit then activated its dedicated deluge system for a filter fire.<ref name="KG-39">Keough and Grill (2005), p. 39.</ref><ref name="KG-86">Keough and Grill (2005), p. 86.</ref>

False activation has been a serious problem in practice. All military service branches in North America experienced false activation of foam-water deluge systems over aircraft with open cockpits.<ref name="SC-110">Scheffey (2002), p. 4-110.</ref> The causes included lightning strikes, water hammer in aging underground mains, accidental releases by maintenance staff and vandalism. Accidental use of manual pull stations, failed pressure relief valves, roof leakage into overhead heat detection and false detector signals also played a part.<ref name="SC-110" />

Optical fire detectors were found to be prone to false alarms and were tested with fuels not typical of aviation.<ref name="SC-110" /> The concept that the U.S. Navy developed in response used low-level deluge nozzles for aqueous film-forming foam (AFFF), operated by improved optical detectors. These detectors were designed to be highly immune to false alarms and to detect fuel spill fires rapidly.<ref name="SC-111">Scheffey (2002), p. 4-111.</ref>

== Water demand ==
In a wet pipe system, water is supplied only for the sprinklers in a design area, since the others are not expected to open.<ref name="FL-74">Fleming (2002), p. 4-74.</ref> In a deluge system all outlets discharge together.<ref name="FL-72" /> The water supply must therefore feed the whole system at once. The minimum flow from one sprinkler is its area of coverage multiplied by the required density, the depth of water applied per minute.<ref name="FL-75">Fleming (2002), p. 4-75.</ref> Applied to a deluge system, the same product over the whole protected area gives the flow that the deluge valve must pass:

<math display="block">Q = D \times A</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>Q</math> || total water flow through the deluge valve || L/min
|-
| <math>D</math> || design density (application rate) || L/(min·m²)
|-
| <math>A</math> || total floor area covered by all open nozzles of the system || m²
|}

The relation assumes an even discharge over the area. In reality, each outlet flows in proportion to the square root of the pressure at that outlet, so nozzles near the supply discharge more than the most remote one.<ref name="FL-75" /> The result is therefore a lower bound on the flow, before any pipe friction is taken into account. The density itself is not derived from theory; sprinkler design criteria are empirical, because the mechanisms by which water suppresses fire have not been characterised well enough for calculation.<ref name="FL-73" />

=== Worked example ===
Factory Mutual Research Corporation tested deluge arrangements for aircraft hangars. One arrangement used standard sprinklers at a spacing of 12.1 m² per head, discharging 6.6 L/(min·m²) of AFFF solution at an end head pressure of 97 kPa.<ref name="SC-107">Scheffey (2002), p. 4-107.</ref> Each head therefore delivers at least 6.6 × 12.1 = 79.9 L/min. Assume a deluge zone of 40 such heads, covering 484 m². The deluge valve must then pass at least:

<math display="block">Q = 6.6\ \mathrm{L/(min \cdot m^2)} \times 484\ \mathrm{m^2} \approx 3{,}200\ \mathrm{L/min}</math>

This is about 53 L/s for the zone, before any friction losses are added. In the same test series, this arrangement extinguished an 83.6 m² JP-4 pool fire in 1 minute 45 seconds.<ref name="SC-107" /> A wet pipe system, by contrast, needs water only for the sprinklers in its design area, because the other sprinklers are not expected to open.<ref name="FL-74" />

== Applications ==
Aircraft hangars are the best documented application in the sources. Hangar protection aims to protect the aircraft and to prevent collapse of the roof, which is usually unprotected steel.<ref name="SC-106" /> Before foam was developed, hangars were protected by water deluge systems discharging about 10.4 L/(min·m²) (0.25 gpm/ft²), together with sloped floors and drains.<ref name="SC-107" /> Even then, protection of aircraft fuselages against burnthrough within about 1 minute could not be ensured.<ref name="SC-107" /> In one 121 m² JP-4 fuel fire, ceiling temperatures of 927 °C were recorded within 30 seconds of ignition, before the deluge system discharged.<ref name="SC-107" />

The introduction of [[firefighting foam]] changed this design. Protein foam systems discharging 8.2 L/(min·m²) were an improvement on water alone.<ref name="SC-107" /> With AFFF through standard sprinklers, the NFPA 409 rate for overhead deluge systems may be reduced from 8.2 to 6.6 L/(min·m²), which is 20 % less foam.<ref name="SC-108">Scheffey (2002), p. 4-108.</ref> NFPA 16 requires a minimum of 6.6 L/(min·m²) for deluge and closed-head AFFF systems. This gives a safety factor over the 4.1 L/(min·m²) at which AFFF from sprinklers is effective on pool fires.<ref name="SC-112">Scheffey (2002), p. 4-112.</ref>

[[File:fse-deluge-system-v3.svg|thumb|upright=1.5|alt=Bar chart of application rates: water deluge 10.4, protein foam 8.2, AFFF through standard sprinklers 6.6, AFFF effective on pool fires 4.1 L/min per m2.|Application rates for overhead hangar deluge protection with water, protein foam and AFFF. Based on Scheffey (2002), pp. 4-107, 4-108 and 4-112.]]

Tests with closed-head sprinklers instead of open heads showed that a large number of sprinklers opened during the fire tests, so that this alternative was not feasible for the hangar scenario.<ref name="SC-108" /> The later U.S. Navy concept combined a quick-response, closed-head wet pipe system in the hangar ceiling with low-level AFFF deluge nozzles for fuel spill fires.<ref name="SC-111" /> More generally, the foam-water deluge system is one type of fixed low-expansion foam system, used in bunds and on spills.<ref name="IS-B60">Iqbal and Salley (2004), p. B-60.</ref>

In nuclear power plants, deluge systems protect equipment such as transformers and turbine-generator bearings.<ref name="IS-H11" /><ref name="IS-H20" /> At North Anna Unit 2 in 1981, a main transformer failed and its oil burned. The deluge systems on two transformers activated, but the fire was too severe for the system and took the fire brigades about one hour to control.<ref name="IS-H11" /> The NRC drew the lesson that a fixed suppression system can be overwhelmed even when the fire starts in the equipment it protects.<ref name="IS-H11" /> At Salem Unit 2 in 1991, deluge systems on two generator bearing housings actuated as designed during a turbine failure. Together with other automatic systems, they extinguished some of the fires.<ref name="IS-H20">Iqbal and Salley (2004), p. H-20.</ref>

In buildings, deluge systems are used for local protection and for water curtains. A water curtain is a screen or wide-angle spray of water that protects exposures from radiated heat, smoke and flames, and may also be called a water screen.<ref name="IS-F69">Iqbal and Salley (2004), p. F-69.</ref> The New York City building code applied to the World Trade Center covered exit ramps in public garages. Their vehicle openings had to be protected by a deluge-type sprinkler water curtain of 3 gpm per linear foot (about 37 L/min per metre).<ref name="GJ-39">Grill and Johnson (2005b), p. 39.</ref> In the towers themselves, local application deluge systems protected the fan filters in the air handling units.<ref name="HO-59">Hopkins et al. (2005), p. 59.</ref><ref name="HO-132" />

Offshore platforms also use deluge nozzles. Their spray is much coarser than water mist, and in British Gas explosion tests such large-drop sprays increased turbulence and could make a gas deflagration worse.<ref name="MB-316">Mawhinney and Back (2002), p. 4-316.</ref> The editors of the industry report concluded that more work was needed with sprays of finer drop sizes.<ref name="MB-316" />

Deluge systems are specified where highly flammable fuels can produce a fast, very hot fire.<ref name="NO-F59" /> In industrial processes, many substances that are gases at ambient conditions are stored under pressure as a liquid with vapour above it.<ref name="IS-5-15">Iqbal and Salley (2004), p. 5-15.</ref> The failure of such a container while its liquid is above its normal boiling point is called a [[BLEVE]], a boiling liquid expanding vapour explosion. A BLEVE typically occurs in a metal container that has been overheated above 538 °C. The liquid absorbs heat, so the wetted part of the shell is usually not at risk; the danger lies in the surfaces of the vapour space.<ref name="IS-5-15" />

== Reliability and limitations ==
Unlike a wet pipe system, a deluge system needs both a detection signal and a deluge valve that opens before any water flows.<ref name="FL-72" /> Unreliability values from a 1994 data report by Parkinson and co-workers are recommended in the fire risk method for United States nuclear power plants. They are 0.05 for deluge or preaction sprinkler systems, compared with 0.02 for wet pipe systems.<ref name="NS-P6" /> These values do not include unavailability during maintenance, credit for manual actuation, dependent failures or plant-specific data.<ref name="NS-P6" /> Earthquakes add a further failure mode, because deluge trip valves may not withstand an earthquake and may open spuriously or jam shut.<ref name="NO-13-3">Nowlen et al. (2005b), p. 13-3.</ref>

[[File:fse-deluge-system-v4.svg|thumb|upright=1.5|alt=Bar chart of unreliability: wet pipe 0.02, carbon dioxide 0.04, halon 0.05, deluge or preaction 0.05.|Recommended unreliability values for automatic suppression systems in fire risk studies of US nuclear power plants. Based on Parkinson et al. (1994), as cited in Nowlen et al. (2005b), p. P-6.]]

The main disadvantage named in the NRC report is extensive water damage, because all of the open sprinklers discharge.<ref name="IS-10-2" /> Unwanted operation makes this worse. At Fort Calhoun in 2001, smoke from a fire in another area actuated a deluge water curtain in a stairwell. The water sprayed onto safety-related motor control centres and caused short circuits in safety-related circuits.<ref name="IS-H5" /> The curtain had been installed because there was no fixed fire barrier between the two fire areas.<ref name="IS-H5" />

A deluge system also depends on its water supply, which can be a shared weak point. At one chemical plant, the fire pump house was flooded three times in five years, and the pump controllers were shut off each time. The pumps thus formed a [[common cause failure]] for numerous deluge water spray systems.<ref name="BA-203">Barry (2002), p. 5-203.</ref> Finally, a deluge system may control a fire rather than extinguish it. The International Fire Engineering Guidelines distinguish three possible effects of suppression on a design fire: no effect, control and extinguishment.<ref name="IFEG-7">National Research Council of Canada et al. (2005), p. 2.7-7.</ref> The North Anna transformer fire shows that a system which operates can still be overwhelmed by the fire it is meant to protect against.<ref name="IS-H11" />

== See also ==
* [[Jet fire]]
* [[Wet pipe sprinkler system]]
* [[Water mist system]]
* [[Fire water supply]]

== Notes ==
<references />

== References ==
* Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
* Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-72–4-87). National Fire Protection Association; Society of Fire Protection Engineers.
* Grill, R. A., & Johnson, D. A. (2005b). ''Fire protection and life safety provisions applied to the design and construction of WTC 1, 2, and 7 and post-construction provisions applied after occupancy'' (NIST NCSTAR 1-1D). National Institute of Standards and Technology.
* Hopkins, M., Schoenrock, J., & Budnick, E. (2005). ''Fire suppression systems'' (NIST NCSTAR 1-4B). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Keough, R. J., & Grill, R. A. (2005). ''Fire alarm systems'' (NIST NCSTAR 1-4C). National Institute of Standards and Technology.
* Mawhinney, J. R., & Back, G. G., III. (2002). Water mist fire suppression systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-311–4-337). National Fire Protection Association; Society of Fire Protection Engineers.
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Nowlen, S. P., Najafi, B., Wyant, F. J., Forester, J., Kazarians, M., Kolaczkowski, A., Joglar, F., Funk, D., Hannaman, G., & Anoba, R. (2005b). ''EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Vol. 2. Detailed methodology'' (NUREG/CR-6850; EPRI 1011989). Electric Power Research Institute; U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research.
* Scheffey, J. L. (2002). Foam agents and AFFF system design considerations. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-88–4-122). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Suppression]]

Occupant load

2026-11-02T17:13:36Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

The '''occupant load''' is the greatest number of people likely to occupy a particular space within a building.<ref name="MBIE25-67">Ministry of Business, Innovation and Employment (2025b), p. 67.</ref> It is usually found by dividing the floor area by a fixed area per person for the activity in the space, or by counting the beds or fixed seats.<ref name="MBIE25-67" /> The number is the starting point for the design of the [[means of escape]], because building codes size exits and stairs to the number of people they must serve.<ref name="Averill-44">Averill et al. (2005), p. 44.</ref> In a fire engineering analysis it is also an input for the [[required safe egress time]]. A hand calculation of the evacuation time begins by establishing how many people are present in each space.<ref name="VR-7.5">Van Rossum (2027), §7.5.</ref><ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref>

Most [[building regulations]] express the design values as occupant load factors in square metres of floor area per person. In New Zealand, C/VM2 (second edition, 2025) lists factors from 0.4 m² per person for standing space to 10 m² per person for offices and 100 m² per person for bulk storage.<ref name="MBIE25-11">Ministry of Business, Innovation and Employment (2025b), pp. 11–12.</ref> These factors are design values rather than predictions. Field studies in Canadian office buildings found about 25 m² of gross rentable area per actual occupant, against the 9.3 m² per occupant assumed in traditional code rules.<ref name="Proulx-355">Proulx (2002), p. 3-355.</ref>

== Definition and related terms ==
C/VM2 (second edition, 2025) gives three ways to determine the occupant load of a space. The floor area can be divided by the area per person for the activity, the sleeping or care spaces can be counted, or the fixed seats can be counted.<ref name="MBIE25-67" /> In care and detention, a "bed" is any place where a person stays for treatment, including recliners, dentist chairs and treatment tables.<ref name="MBIE25-10">Ministry of Business, Innovation and Employment (2025b), p. 10.</ref> The first edition of C/VM2 used the term "occupant densities" for the same table of values.<ref name="MBIE13-83">Ministry of Business, Innovation and Employment (2013), p. 83.</ref>

In crowd movement research, density is the number of persons per unit area of walkway, such as 2.0 persons/m².<ref name="Proulx-345">Proulx (2002), p. 3-345.</ref> It is often expressed by its inverse, the area per person, which for the same crowd is 0.5 m² per person.<ref name="Proulx-345" /> An occupant load factor in m² per person is therefore an inverse density for a whole space. Nelson and Mowrer use density in a different sense, as the degree of crowding in an evacuation route at a given moment.<ref name="NM-369">Nelson and Mowrer (2002), p. 3-369.</ref> The two ideas are linked but not the same, as the section on crowd density below explains.

In Australia, the data sheets for the Fire Safety Verification Method speak of "occupant numbers and densities".<ref name="ABCB-255">Australian Building Codes Board (2020a), Data Sheet D1, p. 255.</ref> They take the values from Table D1.13 of the National Construction Code (NCC) Volume One, which gives the area required per person for each use.<ref name="ABCB-255" /> Where that table does not cover a use, the authority, the fire engineer and other stakeholders agree a value that is consistent with it.<ref name="ABCB-255" />

The use of a space matters a great deal. In the building codes discussed in the NIST report on the World Trade Center, office space was designed for 100 ft² (about 9.3 m²) per occupant.<ref name="Averill-19">Averill et al. (2005), p. 19.</ref> Restaurants, bars and function rooms were assembly space at 15 ft² (about 1.4 m²) per occupant.<ref name="Averill-19" /> A typical office floor in the towers therefore had a design load of 365 to 390 people, while the restaurant floors had more than 1,000 each.<ref name="Averill-19" />

== How the occupant load is calculated ==
For a space with one activity, the occupant load follows from the floor area and the occupant load factor for that activity.<ref name="MBIE25-10" /> The factor is the floor area allowed for each person. Dividing the area by this factor gives the number of people:
<math display="block">N = \frac{A}{f}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>N</math> || occupant load || persons
|-
| <math>A</math> || floor area of the part of the building containing the activity || m²
|-
| <math>f</math> || occupant load factor for the activity || m²/person
|}

The relation applies to one activity at a time. A floor with offices and a conference room, for example, is split into parts with different factors.<ref name="MBIE13-87">Ministry of Business, Innovation and Employment (2013), p. 87.</ref> It does not apply to fixed seating, beds or sleeping spaces, which are counted directly.<ref name="MBIE25-67" /> The area used is the total enclosure floor area, including the area taken up by internal partitions and permanent fixtures.<ref name="MBIE13-85">Ministry of Business, Innovation and Employment (2013), p. 85.</ref> The factors already allow for the part of the floor occupied by furniture, partitions, fixtures and equipment.<ref name="MBIE25-10" />

C/VM2 adds several rules for applying the factors. If a space may be used for different activities, the activity with the greatest occupant load governs.<ref name="MBIE25-10" /> If an activity is not listed, the nearest reasonable description is used.<ref name="MBIE25-10" /> People involved in more than one activity are counted once. No load is assigned to exitways, lift lobbies or sanitary facilities, because the people using them are already counted elsewhere.<ref name="MBIE25-10" /> For fixed seating, extra standing occupants may be added on adjoining floor area, provided the escape route is not obstructed.<ref name="MBIE25-10" />

Codes also differ in which floor area they use. The New York City Building Code calculates the occupant load on a net basis, subtracting shafts, storage rooms and stairs from the gross floor area.<ref name="Averill-47">Averill et al. (2005), p. 47.</ref> The logic is that no egress capacity is needed for floor area where nobody will be.<ref name="Averill-44" /> The International Building Code and NFPA 101 use the gross floor area.<ref name="Averill-47" /> For a tower floor of the World Trade Center, the net basis gave a design load of 390 people, the gross basis 428.<ref name="Averill-47" />

[[File:fse-occupant-load-v2.svg|thumb|upright=1.5|alt=Two identical square office floor plans; left the whole floor is shaded as gross area, right the core shafts and stairs are left white as excluded area.|Gross and net floor area of the same office floor: a net basis leaves out shafts, storage rooms and stairs, so the design occupant load is lower. Based on Averill et al. (2005), pp. 44 and 47.]]

== Typical occupant load factors ==
The table lists a selection of the occupant load factors in C/VM2 (second edition, 2025), from the most crowded uses to the least crowded.<ref name="MBIE25-11" />

{| class="wikitable"
! Activity !! Occupant load factor (m²/person)
|-
| Standing space || 0.4
|-
| Bar standing areas || 0.5
|-
| Dance floors; stadiums and grandstands || 0.6
|-
| Space with loose seating || 0.8
|-
| Bar sitting areas; lobbies and foyers || 1
|-
| Space with loose seating and tables || 1.1
|-
| Dining, restaurant and cafeteria spaces || 1.25
|-
| Exhibition areas and trade fairs || 1.4
|-
| Classrooms || 2
|-
| Meeting rooms || 2.5
|-
| Retail spaces and malls || 3.5
|-
| Art galleries and museums || 4
|-
| Office spaces || 10
|-
| Parking buildings and garages || 50
|-
| Bulk storage including racks and shelves || 100
|}

The range covers more than two orders of magnitude. Standing space at 0.4 m² per person corresponds to 2.5 persons per square metre, while a parking building is assumed to hold one person per 50 m².<ref name="MBIE25-11" /> Bleachers, pews and bench seating are not given an area but a length of 0.45 m of bench per person.<ref name="MBIE25-11" /> For assembly occupancies, Proulx regards seat counts and an area of 0.3 to 0.6 m² per person in waiting spaces as a reasonable basis for the maximum population.<ref name="Proulx-355" />

[[File:fse-occupant-load-v1.svg|thumb|upright=1.5|alt=Horizontal bar chart of occupant load factors on a log scale, from 0.4 m² per person for standing space to 100 m² per person for bulk storage.|Selected occupant load factors in C/VM2 (second edition, 2025): the design floor area per person ranges from 0.4 m² for standing space to 100 m² for bulk storage. Based on Ministry of Business, Innovation and Employment (2025b), pp. 11–12.]]

== Worked examples ==
The commentary to the first edition of C/VM2 works through a function centre and hotel. The occupant loads are based on the floor area of each space and the occupant densities of C/VM2.<ref name="MBIE13-188">Ministry of Business, Innovation and Employment (2013), p. 188.</ref> The grand ballroom measures 50 m by 20 m, so its floor area is 1,000 m². With 1 m² per person, the ballroom holds 1,000 people, for whom four exits are provided where two are required.<ref name="MBIE13-190">Ministry of Business, Innovation and Employment (2013), p. 190.</ref>
<math display="block">N = \frac{1000\ \mathrm{m^2}}{1\ \mathrm{m^2/person}} = 1000\ \mathrm{persons}</math>
The restaurant of 420 m² is assessed at 1.1 m² per person. The division gives 381.8, and the commentary uses 382 people.<ref name="MBIE13-190" /> The meeting rooms, 564 m² at 2.5 m² per person, give 225.6 and are taken as 226 people.<ref name="MBIE13-190" /> Hallways and corridors count as spaces of intermittent use, so their occupants are included in the loads of the rooms they serve.<ref name="MBIE13-190" />

[[File:fse-occupant-load-v3.svg|thumb|upright=1.5|alt=Plan of a ballroom, meeting rooms and a restaurant, each with its floor area, area per person and resulting number of occupants.|Occupant loads in the commentary example: floor area divided by the area per person for each room. Based on Ministry of Business, Innovation and Employment (2013), p. 190.]]

These numbers then set requirements for the escape routes. The example separates the exits by at least 8 m when up to 250 occupants use the escape routes, and by at least 20 m above that number.<ref name="MBIE13-190" /> The second edition keeps a similar rule for spaces with more than 250 people that escape through only two routes.<ref name="MBIE25-37">Ministry of Business, Innovation and Employment (2025b), p. 37.</ref> Where a primary entrance can be identified, C/VM2 (2025) assigns half of the occupant load of a space to that entrance.<ref name="MBIE25-32">Ministry of Business, Innovation and Employment (2025b), p. 32.</ref> For the restaurant this is 191 people. A self-closing door leaf is limited to 50 people per minute, so one leaf would need about 3.8 min to pass them.<ref name="MBIE25-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref>

A second example shows the effect of the factor itself. An office floor of 1,400 m² at the C/VM2 office factor of 10 m² per person gives a design load of 140 people.<ref name="MBIE25-11" /> Proulx describes 70 able-bodied persons on a floor of about 1,400 m² as a fairly high population for an office building.<ref name="Proulx-355" /> At the observed 25 m² per occupant, the same floor would hold 56 people.<ref name="Proulx-355" /> The design value is thus two to two and a half times the population found in the Canadian studies.

[[File:fse-occupant-load-v4.svg|thumb|upright=1.5|alt=Column chart for a 1,400 m² office floor: 151 people at 9.3 m² per person, 140 at 10 m², 70 in a busy floor and 56 at 25 m² per person.|Number of people on a 1,400 m² office floor for different areas per person: design values give two to two and a half times the observed populations. Based on Ministry of Business, Innovation and Employment (2025b), p. 12, and Proulx (2002), p. 3-355.]]

== Use in egress design ==
Building codes largely relate the required egress capacity to the size and use of the area served.<ref name="Averill-44" /> The width and number of stairs are specified so that they equal or exceed the number of occupants on a floor. Generally the floor with the largest occupant load dictates the design of the whole egress system.<ref name="Averill-44" /> Under the 1968 New York City Building Code, 365 people on a World Trade Center floor needed six and a half units of exit width at 60 people per unit of 22 in.<ref name="Averill-45">Averill et al. (2005), p. 45.</ref> That gave an allowable floor load of 390 people.<ref name="Averill-45" />

The number of stairs also depends on the occupant load. The same code required at least two stairs for fewer than 500 people and four stairs for more than 1,000.<ref name="Averill-45" /> The International Building Code of 2000 requires 0.005 m of stair width per occupant in a sprinklered building, which gives 2.2 m for a floor of 429 people.<ref name="Averill-46">Averill et al. (2005), p. 46.</ref> NFPA 101 (2003) requires 0.0076 m per occupant, or 3.3 m in total for the same floor.<ref name="Averill-46" /> The occupant load is thus the quantity that turns a code's [[exit capacity]] per person into a required width in metres.

Several thresholds in C/VM2 (second edition, 2025) are expressed directly as occupant loads. A manual alarm instead of automatic detection is only permitted in high spaces where fewer than 50 occupants are awake and familiar with their surroundings.<ref name="MBIE25-28">Ministry of Business, Innovation and Employment (2025b), p. 28.</ref> Doors on escape routes must open in the direction of escape unless no more than 50 occupants use them.<ref name="MBIE25-32" /> The design scenario in which a fire blocks an exit applies to escape routes serving more than 50 people, and single escape routes may serve up to 50 people.<ref name="MBIE25-37" />

== Use in evacuation time calculations ==
C/VM2 requires the occupant loads in a calculation of the required safe egress time to be determined with its occupant load factors.<ref name="MBIE25-26" /> In the ''Introduction to Fire Safety Engineering'', the first of eight steps in a hand calculation is to establish the number of people in each relevant space.<ref name="VR-7.5" /> The next steps assign these people to the exits and find the exit that takes longest to clear. In the book's example, 300 people are divided equally over four stairs, so each stair serves 75 people.<ref name="VR-7.5" /> A stair of 1.6 m passes 1.2 persons per second, and the 75 people need about 63 s to pass it.<ref name="VR-7.5" />

The Australian data sheets for the Fire Safety Verification Method list occupant numbers and densities as the first occupant characteristic in an evacuation analysis.<ref name="ABCB-252">Australian Building Codes Board (2020a), Data Sheet D1, p. 252.</ref> Maximum numbers are typically assumed in deterministic analyses. Variations with time of day and season can be included in probabilistic analyses.<ref name="ABCB-252" /> The fire safety strategy must state the maximum number of occupants assumed for the building and its enclosures.<ref name="ABCB-255" />

In [[evacuation modelling]] with computer software, the geometry of the building is built first and then filled with people, often with several occupant profiles.<ref name="VR-10.1">Van Rossum (2027), §10.1.</ref> The NIST study of the World Trade Center towers used such models to test the effect of the population. In the simulations, each additional 1,000 occupants added about 5.2 min to the total evacuation time of a tower.<ref name="Averill-162">Averill et al. (2005), p. 162.</ref>

== Design values and actual populations ==
Code values and real populations can differ widely. In Canadian office buildings, Proulx found about 25 m² of gross rentable area per actual occupant, compared with the 9.3 m² per occupant of traditional codes and standards.<ref name="Proulx-355" /> Codes can therefore misjudge the actual number of occupants by a factor of more than 2.<ref name="Proulx-355" /> Counts of workspaces also tend to overestimate office populations, sometimes by 15 percent.<ref name="Proulx-355" />

For the World Trade Center towers, NIST counted offices, cubicles and other workstations on typical floor plans.<ref name="Averill-239">Averill et al. (2005), p. 239.</ref> This gave an average of 14 m² per person and 19,800 people per tower, or 11 m² per person and 25,500 people when visitors were included.<ref name="Averill-239" /> The Port Authority later estimated that the population of a tower would not have exceeded 20,000.<ref name="Averill-239" /> About 8,800 people were present in each tower on 11 September 2001.<ref name="Averill-xxxv">Averill et al. (2005), p. xxxv.</ref> Without building damage, the simulated evacuation of 8,800 people took at least 52 to 71 min, and that of 19,800 people 92 to 112 min.<ref name="Averill-xxxv" />

C/VM2 allows a lower occupant load than its table gives, but the basis must be substantiated to the building consent authority.<ref name="MBIE25-10" /> The document warns that designing for a reduced load can severely restrict future uses of the building and make later upgrades of the means of escape expensive.<ref name="MBIE25-10" /> Conversely, if the maximum number of people will be higher than the table gives, the higher number must be used and justified.<ref name="MBIE25-10" />

== Occupant load and crowd density ==
A common misunderstanding is to treat the occupant load factor as the crowd density in the escape routes. The factor describes how many people a whole space may hold, averaged over its area. Crowd density describes how closely people stand while they move or queue.<ref name="NM-369" /> Nelson and Mowrer base the density of the first exit element, such as an aisle or corridor, on all the occupants it serves, unless their dispersion is known.<ref name="NM-369" />

Crowd density controls walking speed. Below about 0.54 persons/m², people move at their own pace, and above about 3.8 persons/m² movement stops until the crowd thins out.<ref name="NM-370">Nelson and Mowrer (2002), p. 3-370.</ref> The ''Introduction to Fire Safety Engineering'' gives about 1.25 m/s for unhindered walking on a level floor at fewer than 1 person/m².<ref name="VR-7.4">Van Rossum (2027), §7.4.</ref> For flow through doors, C/VM2 assumes a density of 1.9 persons/m² near the constriction.<ref name="MBIE25-33">Ministry of Business, Innovation and Employment (2025b), p. 33.</ref> In crowd disasters, densities approach a critical value of about 8 persons/m², at which no space remains between people.<ref name="Proulx-345" />

The occupant load therefore answers only the first question of an evacuation analysis: how many people must leave. The time needed to pass an exit follows from this number and the flow capacity of the usable width.<ref name="Proulx-347">Proulx (2002), p. 3-347.</ref> The total evacuation time also contains the delay before people start to move, which the occupant load does not describe.<ref name="Proulx-347" />

== See also ==
* [[Specific flow]]
* [[Walking speed]]
* [[Travel distance]]
* [[Evacuation strategy]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Ministry of Business, Innovation and Employment. (2013). ''Commentary for Building Code clauses C1–C6 and Verification Method C/VM2''.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* Nelson, H. E., & Mowrer, F. W. (2002). Emergency movement. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-367–3-380). National Fire Protection Association; Society of Fire Protection Engineers.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Fire safety strategy

2026-11-02T17:13:34Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''fire safety strategy''' is the overall plan that describes how a building will achieve its fire safety objectives, setting out the philosophy and approach adopted to reach the required level of performance.<ref name="FSVM-7">Australian Building Codes Board (2022), p. 7.</ref> It brings together the measures that warn people, allow safe evacuation, restrict the spread of fire and control or extinguish a fire, both active and passive.<ref name="IFEG-053">National Research Council of Canada et al. (2005), p. 0.5-3 (Part 0, Australia).</ref> The strategy also covers the protection of neighbouring property, the plan for firefighting, and the management needed to keep the measures working.<ref name="HB-28">Australian Building Codes Board (2020b), p. 28.</ref><ref name="HB-29">Australian Building Codes Board (2020b), p. 29.</ref>

The strategy sits between the objectives and the detailed analysis. The Australian handbook for the Fire Safety Verification Method (FSVM) describes a logical sequence for fire safety engineering. It runs from objectives and performance requirements, through the building concept and the fire safety strategy, to hazard identification, scenario development, analysis and compliance.<ref name="HB-20">Australian Building Codes Board (2020b), p. 20.</ref> Development is iterative: a trial strategy is analysed, and it is revised until it meets the agreed criteria.<ref name="IFEG-113">National Research Council of Canada et al. (2005), p. 1.1-3.</ref> At the end of design the strategy is documented, preferably in a fire safety handbook that guides construction, commissioning, management and maintenance for the life of the building.<ref name="HB-31">Australian Building Codes Board (2020b), p. 31.</ref>

[[File:fse-fire-safety-strategy-v1.svg|thumb|upright=1.5|alt=Flowchart from fire safety objectives through the fire safety strategy, scenarios and analysis to compliance, with a loop back to the strategy.|The fire safety strategy in the sequence of fire safety engineering; if the analysis does not show compliance, the strategy is modified and analysed again. Based on Australian Building Codes Board (2020b), p. 20 and National Research Council of Canada et al. (2005), p. 1.1-3.]]

== Definition and scope ==
In Australia, the Fire Safety Verification Method Standard 2022 requires the performance-based design brief (PBDB) to cover the fire safety strategy for the building. That strategy must outline the philosophy and approach used to achieve the required level of performance.<ref name="FSVM-7" /> The same clause requires the strategy to pay particular attention to the evacuation strategy and to the management regimes that are necessary.<ref name="FSVM-7" /> A fire safety strategy is therefore more than a list of installations: it is the reasoning that explains why a particular combination of measures will work in a particular building.

The International Fire Engineering Guidelines (IFEG, 2005) use the related term fire safety system. A fire safety system is one or any combination of the methods used in a building to warn people and provide for safe evacuation. The same definition adds the methods used to restrict the spread of fire and to control or extinguish it.<ref name="IFEG-053" /> It includes both active systems, such as alarms and sprinklers, and passive systems, such as fire-resisting construction.<ref name="IFEG-053" /> A fire safety system that is proposed for assessment with fire engineering techniques is called a trial design.<ref name="IFEG-055">National Research Council of Canada et al. (2005), p. 0.5-5 (Part 0, Australia).</ref> In practice, the strategy is the explanation, the system is the set of measures, and a trial design is one candidate version of that system.

The strategy is a product of [[fire safety engineering]]. ''Introduction to Fire Safety Engineering'' defines that discipline as the application of all fire-related fields of knowledge to achieve the relevant fire safety goals, while taking the specific features of the project into account.<ref name="VR-1.1">Van Rossum (2027), §1.1.</ref> The phrase "project-specific" is important. The IFEG point out that fire engineering is specific to the building, its occupants and its site.<ref name="IFEG-037">National Research Council of Canada et al. (2005), p. 0.3-7 (Part 0, Australia).</ref> Buildings that look similar often differ in details that affect fire safety, so a strategy accepted for one building is not normally a precedent for another.<ref name="IFEG-037" />

== Objectives as the starting point ==
A strategy can only be judged against what it is meant to achieve, so design starts with the [[fire safety objectives]]. The Australian handbook states that the design process normally begins by defining the relevant objectives, both regulatory and other, and then develops the strategy.<ref name="HB-20" /> In the National Construction Code (NCC) of Australia, the broad fire safety objectives are life safety of occupants, protection of other property, facilities for firefighting and fire safety during construction.<ref name="HB-27">Australian Building Codes Board (2020b), p. 27.</ref> Protection of the subject building itself and its contents is addressed only to a limited extent for some building classes.<ref name="HB-27" />

Clients and other stakeholders often have further objectives. The IFEG list examples such as limiting damage to the structure and contents, maintaining business continuity, protecting corporate image and heritage, and limiting the release of hazardous materials.<ref name="IFEG-032">National Research Council of Canada et al. (2005), p. 0.3-2 (Part 0, Australia).</ref> Non-fire objectives also shape the strategy, for example extensive natural lighting, an open-plan layout, new materials, sustainability, future flexibility and low life-cycle costs.<ref name="IFEG-033">National Research Council of Canada et al. (2005), p. 0.3-3 (Part 0, Australia).</ref> The IFEG recommend that all these objectives, together with the mandatory requirements, are taken into account in one integrated and cost-effective fire safety system.<ref name="IFEG-033" />

The Dutch context shows how objectives can differ between regulations. ''Introduction to Fire Safety Engineering'' lists possible goals such as preserving the building and its contents, business continuity, limiting reputational damage, limiting casualties and societal impact, and protecting the environment.<ref name="VR-1">Van Rossum (2027), §1.</ref> In the Netherlands, the explanatory notes to the building regulations (Besluit bouwwerken leefomgeving, Bbl) name only two goals: preventing casualties and preventing spread of fire to another plot.<ref name="VR-1.2">Van Rossum (2027), §1.2.</ref> Preserving the building is explicitly not a goal of the Bbl.<ref name="VR-1.2" /> A strategy that must also protect the building or its operations therefore has to go beyond the regulatory minimum.

== Building blocks of a strategy ==
Engineers use several frameworks to make sure a strategy covers every route by which a fire can cause harm. The [[fire safety concepts tree]] of the US National Fire Protection Association (NFPA 550) divides fire safety into two main branches: prevent fire ignition, and manage fire impact.<ref name="ME-158">Meacham (2002), p. 5-158.</ref> The idea is that one or the other must succeed for the objectives to be met.<ref name="ME-158" /> Managing fire impact divides further into managing the fire and managing the exposed people or property.<ref name="ME-158" /> Watts describes how a list of strategies can be taken from a cut set of the tree. His examples are ignition prevention, limitation of combustibles, compartmentation, detection and alarm, suppression, and protection of exposed people or things.<ref name="WA-137">Watts (2002), p. 5-137.</ref>

[[File:fse-fire-safety-strategy-v2.svg|thumb|upright=1.5|alt=Logic tree: fire safety objectives met by preventing fire ignition or managing fire impact; managing impact splits into managing the fire and the exposed.|Top branches of the fire safety concepts tree: either ignition is prevented or the impact of the fire is managed. Based on Meacham (2002), p. 5-158.]]

The IFEG divide a building's fire safety system into six sub-systems for analysis.<ref name="IFEG-112">National Research Council of Canada et al. (2005), p. 1.1-2.</ref> Sub-system A covers fire initiation and development, B smoke development and spread, and C fire spread and its impact on the structure.<ref name="IFEG-112" /> Sub-system D covers detection, warning and suppression, E occupant evacuation, and F intervention by the fire services.<ref name="IFEG-112" /> For each sub-system the guidelines list typical measures. Examples include limiting ignition sources and following procedures for hot work, smoke barriers and mechanical smoke management, fire-resisting barriers and dampers, detectors and sprinklers, evacuation plans and signage, and fire service access and water supplies.<ref name="IFEG-1210">National Research Council of Canada et al. (2005), p. 1.2-10.</ref>

[[File:fse-fire-safety-strategy-v3.svg|thumb|upright=1.5|alt=Section of a three-storey building with six tags A to F marking ignition control, smoke venting, fire-resisting floor, sprinkler, exit stair and fire engine.|The six sub-systems of a fire safety system, each with one typical measure. Based on National Research Council of Canada et al. (2005), pp. 1.1-2 and 1.2-10.]]

Two of these building blocks appear in many strategies. [[Fire compartmentation]] is a fire protection strategy in which a building is subdivided into compartments separated from one another by fire-resistant barriers.<ref name="IS-F16">Iqbal and Salley (2004), p. F-16.</ref> [[Smoke control]] limits the development and spread of smoke, using measures such as smoke barriers, natural smoke venting and mechanical smoke management.<ref name="IFEG-112" /><ref name="IFEG-1210" /> The chosen combination depends on the building. Watts and Hall stress that the relative value of a measure is not constant, because it changes with the other features of the building.<ref name="WH-5">Watts and Hall (2002), p. 5-5.</ref>

== Main parts of the strategy ==
The Australian handbook divides the strategy for NCC compliance into three parts. The strategy for life safety must pay close attention to the [[evacuation strategy]] and to the management regimes needed to achieve the outcome.<ref name="HB-28" /> Documenting both in the design brief is meant to reduce inadvertent non-compliance and to secure the expected reliability of fire protection systems.<ref name="HB-28" /> It also ensures that the evacuation strategy for all occupants, including people with disabilities, becomes part of the fire safety management of the building.<ref name="HB-28" />

The strategy for protecting other property limits the risk of fire spread between buildings. The handbook lists five common means: separation distances, material controls that limit combustibility, protection of openings, fire-resistant construction, and limits on the number, size and arrangement of openings.<ref name="HB-28" /> In Australia, two verification methods, CV1 and CV2, set a maximum radiant heat flux for spread between buildings on adjoining allotments and on the same allotment.<ref name="HB-29" />

The third part is the firefighting strategy. It describes how the fire brigade will be notified of a fire and its location, gain access to the site, and receive communication facilities on arrival.<ref name="HB-29" /> The brigade should also receive documentation of the strategy and an appropriate set-up area for firefighting, search and rescue.<ref name="HB-29" /> In Australia, this strategy is developed with the fire authority, using the Fire Brigade Intervention Model where appropriate.<ref name="HB-29" />

The handbook adds individual and societal risk as a cross-cutting consideration. Individual risk is the frequency at which a person may be expected to suffer a given level of harm from a fire in the building.<ref name="HB-22">Australian Building Codes Board (2020b), p. 22.</ref> Societal risk relates frequency to the number of people who suffer a specified level of harm. Both must be considered, so that the proposed design is at least as safe as a reference building that complies with the deemed-to-satisfy (DTS) provisions.<ref name="HB-22" />

== Development during design ==
Design is iterative. A typical project moves from a feasibility study through schematic design and design development to design documentation.<ref name="HB-20" /> The Australian handbook notes significant advantages when a fire safety engineer is involved throughout these stages, because synergies can be used and practical, cost-effective strategies developed.<ref name="HB-20" /> The IFEG make the same point: the benefits of fire engineering are greatest when it is involved early in the design process.<ref name="IFEG-035">National Research Council of Canada et al. (2005), p. 0.3-5 (Part 0, Australia).</ref>

During design, the team proposes one or more trial designs. A fire safety system should be developed with other factors in mind, such as aesthetics, cost, ease of everyday use, speed of construction and maintenance.<ref name="IFEG-1211">National Research Council of Canada et al. (2005), p. 1.2-11.</ref> Trial designs may include measures not required by the prescriptive provisions to compensate for features that do not comply.<ref name="IFEG-1211" /> They should also include redundancies to compensate for possible failure of components.<ref name="IFEG-1211" /> If analysis shows that a trial design is unsatisfactory, the analysis is repeated on a different trial design until an acceptable one is found.<ref name="IFEG-113" />

New Zealand's Verification Method C/VM2 (second edition, 2025) illustrates this loop in four steps.<ref name="CVM2-9">Ministry of Business, Innovation and Employment (2025b), p. 9.</ref> A concept design first sets out occupant characteristics and the preferred active and passive systems. A fire engineering brief (FEB) then records the inputs and stakeholder considerations.<ref name="CVM2-9" /> In the third step, the trial design is evaluated against ten design scenarios and revised until all ten are satisfied, before the design is documented in step four.<ref name="CVM2-9" /> The IFEG warn that the evaluation of the fire safety strategy should be finished before the design or the cost estimate of the project is complete.<ref name="IFEG-113" />

[[File:fse-fire-safety-strategy-v4.svg|thumb|upright=1.5|alt=Four boxes: concept design, fire engineering brief, trial design tested against ten scenarios with a revise loop, design documents.|Example design process of C/VM2 (New Zealand, 2025): the trial design is revised until all ten design scenarios are satisfied. Based on Ministry of Business, Innovation and Employment (2025b), p. 9.]]

The strategy is shaped by the way it will be analysed.<ref name="HB-20" /> In the Netherlands, rules can be met in a rule-based way or in a risk-based way. A risk-based alternative to the Bbl is usually called equivalence and must reach at least the same level of fire safety as the original prescribed measure.<ref name="VR-1" /> In Australia, the FSVM uses the DTS provisions to define tolerable risk levels, because the performance requirements are generally not quantified.<ref name="HB-28" />

== Documentation and the building life cycle ==
At the end of the design stage, the Australian handbook expects the strategy to be documented in enough detail to start the assessment.<ref name="HB-31" /> The document should contain a summary of the objectives and an overview of the philosophy and approach. It should also include drawings with the fire safety requirements highlighted, the occupant and building characteristics, and the evacuation strategy.<ref name="HB-31" /> It further lists the physical fire safety measures with their operation and expected efficacy and reliability, the management measures, and an implementation plan stating who is responsible.<ref name="HB-31" /> Finally, it states the actions needed to keep the strategy effective through the life of the building.<ref name="HB-31" />

The handbook calls it good practice to consolidate the strategy into a draft fire safety handbook for the building.<ref name="HB-31" /> That handbook gives special attention to [[fire safety management]], such as maintenance of fire protection measures and the evacuation strategy.<ref name="HB-31" /> The final performance-based design report in Australia includes such a handbook, with the procedures needed to achieve the required reliability of fire protection systems.<ref name="HB-125">Australian Building Codes Board (2020b), p. 125.</ref>

A strategy only works if it survives construction and decades of use. The IFEG note that real-life fire safety depends heavily on construction, commissioning, management and use, and maintenance, not only on the approved design.<ref name="IFEG-1232">National Research Council of Canada et al. (2005), p. 1.2-32.</ref> Evacuation procedures and training must be consistent with the design, for example the method of warning and whether evacuation is staged or horizontal.<ref name="IFEG-036">National Research Council of Canada et al. (2005), p. 0.3-6 (Part 0, Australia).</ref> Limits on fuel loads and on the use of escape routes also need to be recorded.<ref name="IFEG-036" /> Alterations and changes of use can compromise the original design and may require a new evaluation.<ref name="IFEG-037" />

For each fire safety feature, the Australian handbook raises a set of life-cycle questions.<ref name="HB-23">Australian Building Codes Board (2020b), p. 23.</ref> These ask how the feature achieves its objectives, and how it will be built safely, commissioned and verified. They also ask whether it creates a hazard during occupation, how it will be maintained or replaced, and how hazards are avoided during renovation or demolition.<ref name="HB-23" />

== Redundancy and limits of protection ==
No strategy removes fire risk completely. The IFEG state that absolute or 100% safety is not attainable, and that a fire safety system may not cope with all possible scenarios.<ref name="IFEG-014">National Research Council of Canada et al. (2005), p. 0.1-4 (Part 0, Australia).</ref> The Australian handbook adds that, as zero risk is approached, the strategy starts to conflict with the function and use of the building.<ref name="HB-28" />

Because the performance requirements of the Australian building code are not quantified, the IFEG recommend building redundancies into the fire safety system.<ref name="IFEG-024">National Research Council of Canada et al. (2005), p. 0.2-4 (Part 0, Australia).</ref> Such redundancies compensate for uncertainty in the methods and data, and they should be examined in sensitivity studies.<ref name="IFEG-024" /> The Fire Safety Evaluation System, a US method for showing equivalence to the NFPA 101 Life Safety Code in certain institutional occupancies, uses the same idea.<ref name="WA-129">Watts (2002), p. 5-129.</ref> It relies on the simultaneous use of three strategies, containment, extinguishment and people movement, so that the failure of one device does not cause a major loss.<ref name="WA-132">Watts (2002), pp. 5-132–5-133.</ref>

C/VM2 in New Zealand turns redundancy into an explicit test. Its design scenario RC assumes that each key fire safety system fails in turn and requires the design to still allow people to escape.<ref name="CVM2-58">Ministry of Business, Innovation and Employment (2025b), p. 58.</ref> The scenario applies where such a failure could expose more than 150 people to untenable conditions, with lower thresholds for sleeping and care occupancies.<ref name="CVM2-58" /> Sprinkler systems and automatic fire alarms installed to a recognised standard are considered reliable enough to be exempt.<ref name="CVM2-58" />

The Dutch situation shows a further limit. ''Introduction to Fire Safety Engineering'' explains that Dutch regulations set measurable limits on means, such as travel distance in metres or compartment area in square metres. These limits do not correspond directly to quantified goals.<ref name="VR-1.2" /> How many casualties are acceptable, or how large the risk of spread to the neighbouring plot may be, is not laid down.<ref name="VR-1.2" /> A strategy that departs from the rules must therefore agree in advance which performance or residual risk is acceptable, often by comparison with a reference situation that complies with the regulations.<ref name="VR-1.2" />

== See also ==
* [[Qualitative design review]]
* [[Performance-based design]]
* [[Prescriptive design]]
* [[Fire scenario]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* 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, Office of Nuclear Reactor Regulation.
* Meacham, B. J. (2002). Building fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-153–5-175). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Watts, J. M., Jr. (2002). Fire risk indexing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-125–5-142). National Fire Protection Association; Society of Fire Protection Engineers.
* Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.

[[Category:Fire Safety Design & Regulation]]

Stay-put strategy

2026-11-02T17:13:32Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''stay-put strategy''' is an approach to a fire emergency in which occupants remain where they are, instead of leaving the building when the alarm sounds. Engineering guidance describes the same idea as "protect in place", "defend in place", "stay in place" or "shelter in place".<ref name="IFEG282">National Research Council of Canada et al. (2005), p. 2.8-2.</ref><ref name="SBE4-2">Schifiliti et al. (2002), p. 4-2.</ref><ref name="VM2-34">Ministry of Business, Innovation and Employment (2025b), p. 34.</ref> The ''International Fire Engineering Guidelines'' describe protect in place as a more recent development, used particularly in fully sprinklered residential buildings. Occupants are encouraged to remain where they are rather than escape through corridors or stairs that may be filled with smoke.<ref name="IFEG282" />

Staying in place is only as safe as the place itself. In New Zealand, C/VM2 (second edition, 2025) requires such a place inside a building to be separated from the fire, protected by sprinklers and large enough for its occupants, with a way out.<ref name="VM2-68">Ministry of Business, Innovation and Employment (2025b), p. 68.</ref> Evidence from real fires shows both sides of the approach. In a fire in a 25-storey apartment building, almost half of the residents who tried to leave were turned back by smoke in corridors and stairs.<ref name="Bryan334">Proulx (1998), as cited in Bryan (2002), pp. 3-334–3-335.</ref> At the World Trade Center in 2001, advice to stay in place and await rescue was not the preferable course of action.<ref name="Av108">Averill et al. (2005), p. 108.</ref>

== Terminology ==
The sources group the strategy with several related ideas, and the names overlap. The ''International Fire Engineering Guidelines'' (IFEG, 2005) separate three approaches that differ from a traditional full evacuation. These are staged evacuation in high-rise buildings and hospitals, the fire safe refuge, and protect in place.<ref name="IFEG282" /> A fire safe refuge is a special fire compartment where occupants go to wait for rescue by the fire service, rather than evacuate.<ref name="IFEG282" /> In protect in place, occupants remain where they are.<ref name="IFEG282" />

Other documents use "defend in place" in a wider sense. The design guidance of the Australian Building Codes Board speaks of "evacuation and / or defend in place strategies".<ref name="HB117">Australian Building Codes Board (2020b), p. 117.</ref> The NIST report on occupant behaviour at the World Trade Center calls a [[phased evacuation]] "also known as defend-in-place". In that scheme, occupants of the fire floor and the floors around it go down to three floors below the fire, while everybody else stays.<ref name="AvXXXV">Averill et al. (2005), p. xxxv.</ref> The SFPE Handbook chapter on detection systems lists "stay-in-place or defend-in-place strategies" next to partial evacuation and relocation strategies.<ref name="SBE4-2" /> In the narrow sense used in this article, a stay-put strategy keeps most occupants where they are, while a phased evacuation moves floors out in a planned order.

== Place among evacuation strategies ==
A stay-put strategy is one of the options in a building's [[evacuation strategy]], the plan that states who leaves, when and by which route. The IFEG notes that traditional practice has been to start evacuating occupants in response to fire alarms, following an evacuation management plan.<ref name="IFEG182">National Research Council of Canada et al. (2005), p. 1.8-2.</ref> In [[high-rise building]]s with emergency warning and intercommunication systems, trained personnel may manage the evacuation. Floors furthest from the fire are first put on alert and are evacuated progressively only if the fire continues to develop.<ref name="IFEG282" /> In hospitals, nursing staff first move patients horizontally to an adjacent smoke zone or fire compartment. Patients are moved further only if the threat keeps growing.<ref name="IFEG282" />

Staying put is thus the far end of a scale that runs from full evacuation through staged and horizontal movement. The IFEG itself is limited to designs in which occupants are evacuated to a place of safety. It describes the newer occupant management concepts as "still developing".<ref name="IFEG182" /> Some occupants cannot leave at all without help. Iqbal and Salley note that the US model building codes place much stricter requirements on institutional occupancies, such as hospitals, nursing homes and penal institutions. A defend-in-place strategy is necessary there because the occupants cannot flee the structure without assistance.<ref name="IS-B33">Iqbal and Salley (2004), p. B-33.</ref>

[[File:fse-stay-put-strategy-v1.svg|thumb|upright=1.5|alt=Horizontal scale with five building icons, from full evacuation on the left to protect in place on the right.|Evacuation approaches described in the guidance, from moving everybody out to keeping occupants where they are. Based on National Research Council of Canada et al. (2005), pp. 1.8-2 and 2.8-2.]]

== Conditions for staying in place ==
A person can only stay safely where fire and smoke cannot reach for the whole duration of the fire. This is why the strategy rests on [[fire compartmentation]]: the division of a building into separate fire compartments. In the Netherlands, the sub-goals behind the building regulations include maintaining compartmentation, limiting the spread of fire and smoke, and keeping escape routes usable.<ref name="VR1.2">Van Rossum (2027), §1.2.</ref> The New Zealand Verification Method C/VM2 (second edition, 2025) states the conditions most precisely. A place of safety inside a building must meet four requirements:<ref name="VM2-68" />
* its fire separations have enough fire resistance to withstand burnout at the point of the fire source;
* the building is protected by an automatic sprinkler system complying with NZS 4541 or NZS 4515;
* the place is designed for the intended number of persons;
* it has enough means of escape for those persons to reach a safe place outside the building.

The first condition means that the walls and floors around the waiting place must hold until the fire has consumed all its fuel. The second condition matches the IFEG remark that protect in place is used particularly in fully sprinklered residential buildings.<ref name="IFEG282" /> The fourth condition keeps a way out open in case the situation changes.<ref name="VM2-68" />

[[File:fse-stay-put-strategy-v2.svg|thumb|upright=1.5|alt=Cross-section of three floors with a burning flat, fire-resisting walls and floors, sprinkler heads, residents in a neighbouring flat and a protected stair.|Conditions for a place of safety inside a building in New Zealand's C/VM2 (second edition, 2025), shown for a flat where residents stay during a fire elsewhere. Based on Ministry of Business, Innovation and Employment (2025b), p. 68.]]

Smoke is the second threat. ''Introduction to Fire Safety Engineering'' reports Dutch tests in residential buildings in which carbon monoxide concentrations far from the fire were very high while visibility was still more than 20 m.<ref name="VR6.1">Brandweeracademie (2020), as cited in Van Rossum (2027), §6.1.</ref> The reason is that soot particles settle and travel less far through a building than the gases do.<ref name="VR6.1" /> A corridor that looks clear can therefore still be dangerous. This matches the reasoning of the IFEG, in which occupants stay rather than pass through corridors or stairs that may be filled with smoke.<ref name="IFEG282" />

== Evidence from real fires ==
Studies of fires in high-rise residential buildings show why moving through a building can be riskier than staying. Proulx surveyed the residents of a 25-storey apartment building after a fire, with 137 questionnaires returned. Of those residents, 68 % were over 60 years of age.<ref name="Bryan334" />

In total, 114 occupants (83 %) tried to evacuate, and 96 of them (84 %) moved through smoke.<ref name="Bryan334" /> Only 61 of the 114 succeeded. The other 53 were stopped by smoke in the corridors or stairs. Of these, 29 went back to their own apartments and 24 sought refuge in other apartments.<ref name="Bryan334" />

[[File:fse-stay-put-strategy-v3.svg|thumb|upright=1.5|alt=Stacked bars: of 114 residents who tried to leave, 61 succeeded, 29 returned to their own flat and 24 sheltered in another flat.|Outcome for the 114 residents who tried to evacuate during a fire in a 25-storey apartment building. Based on Proulx (1998), as cited in Bryan (2002), pp. 3-334–3-335.]]

In two high-rise residential fires in Canada studied by Proulx, six people died in stairwells in one case and one person died in the suite of origin in the other.<ref name="P3-349">Proulx (2002), p. 3-349.</ref> In the first fire, the alarm was not audible in many apartments, and residents learned of the fire from others. In the second, every suite had a sounder and occupants waited for instructions from the voice communication system.<ref name="P3-349" /> In a night-time high-rise residential fire studied by Brennan, only about half the occupants actually evacuated.<ref name="Bren">Brennan (1998), as cited in Proulx (2002), p. 3-349.</ref>

Occupants also choose to stay of their own accord. Hall and Ahrens point out that occupants may defend in place when they should not, or evacuate when staying would be safer. They stress that neither is panic or irrational behaviour, but a realistic picture of how people respond to fire.<ref name="HA5-74">Hall and Ahrens (2002), p. 5-74.</ref> Bryan describes "convergence clusters", in which occupants gather in particular rooms that they select as refuges.<ref name="Bryan328">Bryan (2002), p. 3-328.</ref>

In the MGM Grand Hotel fire, guests gathered in rooms with balconies because these offered ventilation, less smoke, better visibility and a way to communicate.<ref name="Bryan328" /> The clusters held between 3 and 35 people. They usually lasted until help arrived or until fire or rescue personnel told the occupants to leave.<ref name="Bryan328" />

== Information, management and design ==
A stay-put strategy relies on information reaching the occupants. Schifiliti, Meacham and Custer note that in stay-in-place, defend-in-place and partial evacuation strategies, the detection system does more than raise an alarm. It provides information about the location and extent of the fire, after which instructions are given to the people concerned.<ref name="SBE4-2" />

A drill in a 21-storey office building in Ottawa in 1971 shows how much this depends on the people giving the instructions. The selective procedure could have ended after 6 minutes, once the presumed fire floor and two adjacent floors were cleared.<ref name="P3-358">Proulx (2002), p. 3-358.</ref> However, a wrongly set switch delayed the first announcement. The message that followed, "Please evacuate. Other floors stand by", was ambiguous.<ref name="P3-358" />

In Australia, the Fire Safety Verification Method (FSVM) requires the fire safety strategy to pay particular attention to the evacuation strategy and the management regimes it needs.<ref name="FSVM7">Australian Building Codes Board (2022), p. 7.</ref> The FSVM handbook asks that the report of the performance-based design brief describe the evacuation and defend-in-place strategies for all people. It must also describe how these will remain effective throughout the life of the building.<ref name="HB117" /> For an ultra-high-rise building, the handbook lists additional protect-in-place and partial evacuation strategies among the enhancements that may be appropriate.<ref name="HB49">Australian Building Codes Board (2020b), p. 49.</ref>

In New Zealand, C/VM2 (second edition, 2025) has a short section on "delayed evacuation strategies". Buildings whose occupants are required to stay in place, or are evacuated to a place inside the building, must meet the definition of a place of safety.<ref name="VM2-34" /> The examples given are people who are detained or who are undergoing treatment, such as in an operating theatre, a hyperbaric chamber or a dialysis unit.<ref name="VM2-34" /> A comment in C/VM2 adds that such spaces usually have climate control, so smoke detection and the smoke control of the air handling system need special care.<ref name="VM2-34" />

== Limits and risks ==
Guidance names several situations in which the strategy may not work as planned. The Australian data sheets warn that phased evacuation strategies may be sensitive to rapid fire spread over several floors through the outside wall, a [[facade fire]].<ref name="DS250">Australian Building Codes Board (2020a), p. 250.</ref> Depending on the building, this scenario may need to be analysed, and the strategy should state what to do when fire spreads quickly to several floors.<ref name="DS250" /> The same data sheets require a check of the design in which the automatic alarm or intercommunication system fails.<ref name="DS250" /> In small and medium-sized multi-residential buildings, alarms are often simple and the emergency organisation informal, so occupant behaviour varies more.<ref name="DS250" />

At the World Trade Center in 2001, stay-in-place advice did not match the situation. Emergency dispatchers and 9-1-1 operators generally told occupants to shelter in place and await rescue. This was in line with existing protocols, but it was not the preferable course of action that day.<ref name="Av108" /> Information about where the aircraft had struck was generally not passed on to trapped occupants.<ref name="Av145">Averill et al. (2005), p. 145.</ref> Simulations by NIST showed that a phased evacuation of the fire floor and the floors around it would have taken 4 to 11 minutes. A full evacuation of a tower with 19,800 occupants would have taken 92 to 112 minutes.<ref name="AvXXXV" />

After the investigation, NIST recommended that occupants be trained for a range of scenarios, including both evacuation and shelter in place. It also recommended that tall buildings be designed for timely full evacuation when needed.<ref name="NIST215">National Institute of Standards and Technology (2005), p. 215.</ref> NIST noted that current strategies and laws generally require occupants with reduced mobility to shelter in place and await assistance. It recommended giving them a means of self-evacuation where possible, for example protected lifts.<ref name="NIST216">National Institute of Standards and Technology (2005), p. 216.</ref> NIST also found it difficult to predict how long people would be willing to wait in a refuge area before entering a stair.<ref name="NIST216" />

== See also ==
* [[Progressive horizontal evacuation]]
* [[Evacuation of people with disabilities]]
* [[Human behaviour in fire]]
* [[Protected escape route]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Australian Building Codes Board. (2020b). ''Handbook: Fire Safety Verification Method'' (Print version 1.1).
* Australian Building Codes Board. (2022). ''Fire Safety Verification Method Standard 2022''.
* Averill, J. D., Mileti, D. S., Peacock, R. D., Kuligowski, E. D., Groner, N., Proulx, G., Reneke, P. A., & Nelson, H. E. (2005). ''Occupant behavior, egress, and emergency communications'' (NIST NCSTAR 1-7). National Institute of Standards and Technology.
* Bryan, J. L. (2002). Behavioral response to fire and smoke. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-315–3-341). National Fire Protection Association; Society of Fire Protection Engineers.
* Hall, J. R., Jr., & Ahrens, M. J. (2002). Data for engineering analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 5-65–5-78). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Institute of Standards and Technology. (2005). ''Final report on the collapse of the World Trade Center towers'' (NIST NCSTAR 1).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Proulx, G. (2002). Movement of people: The evacuation timing. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-342–3-366). National Fire Protection Association; Society of Fire Protection Engineers.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Human Behaviour & Evacuation]]

Stair pressurisation

2026-11-02T17:13:29Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

'''Stair pressurisation''' is a [[smoke control]] method in which fans blow outside air into a stair enclosure, so that the air pressure in the stair is slightly higher than on the floors around it. Air then leaks out of the stair through the gaps around closed doors, and this outward flow stops smoke from leaking in.<ref name="Klo-278">Klote (2002), p. 4-278.</ref><ref name="Iq-F61">Iqbal and Salley (2004), p. F-61.</ref> The main goal is a smoke-free escape route; a second goal is a smoke-free staging area for firefighters.<ref name="Klo-286">Klote (2002), p. 4-286.</ref> It is a positive pressure system, and building codes in the United States require it in high-rise buildings.<ref name="Evans-70">Evans et al. (2005), p. 70.</ref><ref name="FS-10">Ferreira and Strege (2005), p. 10.</ref>

The system has to work against the natural forces that move smoke through a building, above all the stack effect in tall shafts. A pressure difference that is too small lets smoke in, while one that is too large makes the doors too heavy to open.<ref name="Klo-281">Klote (2002), p. 4-281.</ref> For closed doors, minimum design values of about 5 to 25 Pa are suggested, depending on the fire and on sprinklers. In the United States, the Life Safety Code limits the force needed to open an escape door to 133 N.<ref name="Klo-285">Klote (2002), p. 4-285.</ref> Open doors are the weak point. When many people leave a building at once, many stair doors stand open and the pressure in the stair collapses, as studies of the World Trade Center towers showed.<ref name="FS-60">Ferreira and Strege (2005), p. 60.</ref>

== Purpose and principle ==
Smoke is a major threat in building fires, and it often travels far from the fire itself. Stairs and lift shafts frequently fill with smoke, which blocks or slows evacuation. In the late 1960s the idea of pressurising stairs to keep smoke out began to attract attention. The term smoke control was later coined for systems that use pressure produced by mechanical fans to limit smoke movement.<ref name="Klo-274">Klote (2002), p. 4-274.</ref> A glossary of the US Nuclear Regulatory Commission describes stair pressurisation as increasing the air pressure in stair wells, usually with fan systems, to provide a refuge area from fire and smoke.<ref name="Iq-F61" /> The book ''Introduction to Fire Safety Engineering'' likewise notes that a smoke management system can keep an area free of smoke by overpressure.<ref name="VR-6.6">Van Rossum (2027), §6.6.</ref>

Smoke control rests on two principles. First, airflow by itself can control smoke if its average velocity is high enough. Second, a pressure difference across a barrier, such as a wall with a door, can stop smoke from crossing it.<ref name="Klo-278" /> Pressurisation creates small airflows through the cracks around closed doors and through gaps in the construction, and these flows keep smoke out of the high-pressure side.<ref name="Klo-278" />

[[File:fse-stair-pressurisation-v1.svg|thumb|upright=1.5|alt=Cross-section of a five-storey building with a pressurised stair; air leaks out of the stair at each closed door and keeps smoke on the fire floor out.|A fan supplies outside air to the stair. The higher pressure in the stair drives air out through the gaps around the closed doors, which keeps smoke on the fire floor out of the stair. Based on Klote (2002), pp. 4-278 and 4-286.]]

For a closed door the velocity in the gaps is hard to determine, so the pressure difference is the quantity used in design. For an open door the air velocity is the relevant quantity.<ref name="Klo-279">Klote (2002), p. 4-279.</ref> A pressurised stair therefore has two tasks: keep a positive pressure across each closed stair door and, ideally, push enough air through the door on the fire floor when it is opened.<ref name="Klo-286" />

The World Trade Center investigation describes stair pressurisation as a positive pressurisation system that limits smoke spread into the stairways of buildings such as [[High-rise building|high-rise buildings]]. Its aim is to keep the primary escape routes of the building clear of smoke.<ref name="Evans-70" /> Positive pressure systems also have drawbacks. Too much pressure, or unbalanced pressure, may push smoke from the fire zone into other parts of the building, and such systems do not remove smoke from the building.<ref name="FS-10" /> A pressurised stair is a special form of [[protected escape route]]: the walls and doors form the barrier, and the fans make that barrier work even when it leaks.<ref name="Klo-279" />

== Forces that the system must overcome ==
The main forces that drive smoke through a building are the stack effect, the buoyancy of hot smoke, expansion, wind and the ventilation system.<ref name="Klo-274" /><ref name="VR-6.6" /> The [[stack effect]] is the vertical movement of a gas caused by a difference in density, as in a chimney.<ref name="VR-6.6" /> In cold weather the warm air inside a building rises through shafts such as stairs and lift shafts. Klote gives the example of a building 60 m tall with the neutral plane at mid-height. With −18 °C outside and 21 °C inside, the stack effect alone gives a pressure difference of up to 55 Pa between a shaft and the outside.<ref name="Klo-275">Klote (2002), p. 4-275.</ref>

Hot smoke adds its own buoyancy. In a fire room at 800 °C, the pressure difference 1.52 m above the neutral plane is about 13 Pa. In a series of full-scale fire tests, the largest pressure difference measured across the wall of the burn room was 16 Pa, at the ceiling.<ref name="Klo-276">Klote (2002), p. 4-276.</ref> Wind can produce much larger pressures on the outside of a building. A wind of 15.6 m/s with a pressure coefficient of 0.8 gives about 117 Pa. Its effect inside a tight building with closed windows is small, but it becomes important once a window in the fire room breaks.<ref name="Klo-277">Klote (2002), p. 4-277.</ref>

In field tests of smoke control systems, pressure differences across closed doors often fluctuated by as much as 5 Pa, probably because of the wind. A stair pressurisation system must therefore produce a pressure difference large enough not to be overcome by these fluctuations, by the stack effect, by smoke buoyancy or by the wind.<ref name="Klo-281" /> Weather matters too. A smoke control system has no thermal time lag, so it meets the full stack effect of the moment it starts. If the outside temperature is below the winter temperature used in design, stack effect problems may result.<ref name="Klo-285" />

== Airflow through door gaps ==
The amount of air a fan must supply follows from the leakage of the stair. Air flows from the stair through every gap, and the flow through each gap grows with the square root of the pressure difference across it. Klote takes a flow exponent of 0.5 as reasonable for all flow paths except extremely narrow cracks.<ref name="Klo-280">Klote (2002), p. 4-280.</ref> With a flow coefficient of 0.65 and an air density of 1.2 kg/m³, the volume flow through a gap is:<ref name="Klo-280" />

<math display="block">\dot{V} = K_f\, A \sqrt{\Delta P}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\dot{V}</math> || volume flow of air through the gap || m³/s
|-
| <math>K_f</math> || coefficient, 0.839 for a flow coefficient of 0.65 and an air density of 1.2 kg/m³ || m/(s·Pa0.5)
|-
| <math>A</math> || flow area of the gap (leakage area) || m²
|-
| <math>\Delta P</math> || pressure difference across the gap || Pa
|}

The flow coefficient depends on the shape of the gap and is generally between 0.6 and 0.7.<ref name="Klo-280" /> A closed door with a crack area of 0.01 m² leaks about 0.013 m³/s at 2.5 Pa and 0.073 m³/s at 75 Pa.<ref name="Klo-281" /> Leakage areas depend mainly on workmanship, for example on how well a door is fitted. For the walls of an average stairwell, Klote gives a leakage area of 0.11 × 10−3 times the wall area, excluding the gaps around doors.<ref name="Klo-282">Klote (2002), p. 4-282.</ref> For an open stair door, a flow area of half the door area is recommended, because the flow pattern through a real doorway is complex.<ref name="Klo-282" /> In New Zealand, C/VM2 (second edition, 2025) likewise treats an open egress door as half its width for smoke flow calculations.<ref name="MBIE-15">Ministry of Business, Innovation and Employment (2025b), p. 15.</ref>

[[File:fse-stair-pressurisation-v2.svg|thumb|upright=1.5|alt=Line chart: airflow through a door gap against pressure difference, two rising curves that flatten, for gaps of 0.01 and 0.024 square metres.|Air leakage through the gap of a closed door against the pressure difference, for gap areas of 0.01 m² and 0.024 m². Based on Klote (2002), pp. 4-280 and 4-288.]]

A worked example uses the stair of Klote's design example: 20 floors, each with a leakage area of 0.030 m² between stair and floor, made up of the door gap and the stair wall.<ref name="Klo-288">Klote (2002), p. 4-288.</ref> The pressure difference is taken as 20 Pa, the bottom value of that example. One floor then leaks 0.839 × 0.030 × √20 = 0.11 m³/s, and 20 floors leak about 2.3 m³/s.

This simple estimate assumes the same pressure difference on every floor. Klote's full analysis includes the stack effect for an outside temperature of −10 °C and a stair at 21 °C. The pressure difference then rises from 20 Pa at the bottom of the stair to 106 Pa at the top, and the fan must supply 3.9 m³/s.<ref name="Klo-288" /> Hand calculations of this kind assume no vertical leakage between floors. For vertical leakage and open stair doors, computer analysis is recommended.<ref name="Klo-287">Klote (2002), p. 4-287.</ref>

[[File:fse-stair-pressurisation-v3.svg|thumb|upright=1.5|alt=Line chart: pressure difference between stair and building rising in a straight line from 20 Pa at the bottom to 106 Pa at the top of a 66 m stair.|Pressure difference between a pressurised 20-storey stair and the building, with −10 °C outside and 21 °C in the stair. Based on Klote (2002), p. 4-288.]]

The result is sensitive to leakage. If the leakage area per floor in the same example were 0.050 m² instead of 0.030 m², the required supply would be 6.5 m³/s. In practice these areas are difficult to estimate and even more difficult to control. A fan with an adjustable drive is one way to match the supply to the real leakage of the finished building.<ref name="Klo-288" /> Network computer programs such as CONTAM model each stair as a vertical series of spaces, one per floor, and can include wind, stack effect and open doors.<ref name="Klo-290">Klote (2002), p. 4-290.</ref>

== Design criteria ==
Klote names five groups of design parameters for smoke control: leakage areas, weather data, pressure differences, airflow and the number of open doors.<ref name="Klo-285" /> The minimum pressure difference follows from the criterion that no smoke should leak in during the evacuation. To withstand an intense fire next to a door, a minimum of 20 to 25 Pa is suggested, slightly above the 16 Pa measured in fire tests. For hot smoke from a remote fire, 15 to 20 Pa is suggested. In a sprinklered building the smoke is cooled, and 5 to 10 Pa is suggested to allow for pressure fluctuations.<ref name="Klo-285" />

The maximum pressure difference is set by the force needed to open a door. A person's physical condition largely determines what force is reasonable, so the limit is hard to define. The US Life Safety Code (NFPA 101), as cited by Klote, limits the force to open any door in a means of escape to 133 N.<ref name="Klo-285" /> The total opening force is the force of the door closer plus the force needed to overcome the pressure difference. The closer alone usually needs more than 13 N and sometimes as much as 90 N. For a door 2.13 m high and 0.91 m wide, a pressure difference of 75 Pa with a closer force of 53 N gives a total of 133 N.<ref name="Klo-282" />

Where stair doors are [[fire door|fire doors]] with self-closers, C/VM2 (New Zealand, second edition, 2025) assumes them to be closed unless occupants are using them. During egress it treats each door as open for three seconds per person, or for the whole queuing time when people queue.<ref name="MBIE-15" /> In Australia, the 2020 data sheets for the Fire Safety Verification Method note that pressurisation can push opening forces above the permitted limits and can stop fire and smoke doors from closing automatically.<ref name="ABCB-132">Australian Building Codes Board (2020a), Data Sheet C4, p. 132.</ref>

Open doors need an air velocity criterion. Klote estimates about 4 m/s to stop smoke from an intense fire of 2.4 MW and 1.5 m/s for a fire of 125 kW. Where sprinklers cool the smoke, 0.25 to 1.25 m/s may be used.<ref name="Klo-286" />

In Australia, the deemed-to-satisfy provisions of the National Construction Code reference AS 1668.1. According to the 2020 data sheets, this standard requires an airflow of at least 1 m/s through each open doorway to the fire floor. That applies when all doors to that floor are open together with the main discharge doors of all fire-isolated exits.<ref name="ABCB-127">Australian Building Codes Board (2020a), Data Sheet C4, p. 127.</ref> For a door of 0.91 m by 2.13 m, 1 m/s means 1.94 m³/s through a single doorway. That is about half the supply in Klote's whole closed-door example. Because such flows are large, pressure relief is needed, so that excessive pressures do not make doors hard to open or stop them from closing.<ref name="ABCB-127" />

How many doors will be open at the same time depends on the use of the building. In a densely occupied building it is very likely that all doors will be opened during evacuation. With a staged evacuation plan, a refuge area concept or sparse occupation, only a few doors may be open at once.<ref name="Klo-286" /> Designing for all doors open makes sure the system always works, but it probably adds to the cost.<ref name="Klo-286" />

== System types ==
Stair pressurisation systems are divided into single and multiple injection systems. A single injection system supplies air at one point, most often at the top of the stair.<ref name="Klo-286" /> If smoke reaches the fan intake, the system can blow smoke into the stair it is meant to protect. Automatic shutdown should therefore be considered for that event. In tall stairs, single injection can fail when a few doors near the injection point are open, because all the supply air escapes through them. This failure is especially likely with injection at the bottom while the ground floor door is open.<ref name="Klo-286" />

Multiple injection systems supply air at several points over the height of the stair, with the fan at ground level or on the roof. The supply duct can run in a separate shaft or inside the stair itself. In that case the duct must not obstruct the evacuation.<ref name="Klo-287" /> An alternative is to divide a tall stair into compartments of a few floors, each with its own injection point. The benefit is lost when the doors between compartments are open. Compartmentation is therefore unsuitable for densely occupied buildings where everyone leaves by the stair at once, but it can work with a staged evacuation.<ref name="Klo-287" />

[[File:fse-stair-pressurisation-v4.svg|thumb|upright=1.5|alt=Three stair cross-sections: one with a single roof fan, one with a duct feeding air at several floors, one divided into sections each with its own supply.|Three ways to supply air to a tall stair: single injection at the top, multiple injection through a duct, and a stair divided into compartments. Based on Klote (2002), pp. 4-286 and 4-287.]]

Stair pressurisation is often combined with other measures. Lift shafts can be pressurised in the same way, and the same analysis applies with the lift shaft in place of the stair.<ref name="Klo-289">Klote (2002), p. 4-289.</ref> A protected lift shaft matters for the fire service, because firefighters use lifts with special keys for rescue and to bring equipment to the fire floor.<ref name="Klo-289" /> This is the role of a [[firefighting lift]]. A building with only stair pressurisation can still suffer smoke spread through floors and other shafts. Zone smoke control, in which the fire floor is exhausted and other floors are supplied with air, is meant to limit that spread.<ref name="Klo-289" />

== Reliability and testing ==
A smoke control system must be tested after installation, because many problems have been found in practice. Klote mentions fans running backwards, fans without electrical power and controls that did not work. The first test is functional and checks each component. The second checks the performance of the system as a whole, usually by measuring pressure differences across doors and, where needed, air velocities through open doors.<ref name="Klo-290" /> Chemical smoke from smoke candles is not recommended for performance tests, because it lacks the buoyancy of real hot smoke.<ref name="Klo-290" /> Pressure difference testing is therefore the most practical performance test.<ref name="Klo-291">Klote (2002), p. 4-291.</ref>

The International Fire Engineering Guidelines (2005) state that for smoke control, greater complexity means lower reliability.<ref name="IFEG-3.5-3">National Research Council of Canada et al. (2005), p. 3.5-3.</ref> The Australian data sheets of 2020 summarise studies of stair pressurisation in service. Audits of Australian buildings found that the fans failed to start on alarm in 10 % of office buildings and 8 % of retail premises.<ref name="ABCB-137">Australian Building Codes Board (2020a), Data Sheet C4, p. 137.</ref> A detailed study by Fazio treated door forces above 110 N or air velocities below 1 m/s as failure. On that basis the effectiveness for airflow was between 30 % and 63 % and that for door forces only 8 % to 10 %.<ref name="ABCB-137" /> The data sheets note that smoke may still be held back at lower velocities, and that some people can open doors with forces above 110 N.<ref name="ABCB-137" />

Fazio attributed these low values to systems that were never commissioned, were not maintained, or had faults that were not repaired.<ref name="ABCB-138">Australian Building Codes Board (2020a), Data Sheet C4, p. 138.</ref> The data sheets adopt a typical reliability of 50 % for Australian fire-isolated exit pressurisation systems, with a low value of 30 % and a high value of 60 %. For comparison, they give 70 % for lift shaft pressurisation.<ref name="ABCB-134">Australian Building Codes Board (2020a), Data Sheet C4, p. 134.</ref> They add that most of the data predate the current design and servicing standards, so newer systems may perform better.<ref name="ABCB-134" />

== Limits in tall buildings and mass evacuation ==
The World Trade Center towers in New York illustrate the limits of the method. Between 1976 and 1979 the Port Authority tested a prototype system in one stair of WTC 1, the north tower. The criteria of New York City Local Law 5 were 12.5 Pa across closed stair doors with three doors open, 25 Pa with all doors closed, and a maximum door opening force of 156 N.<ref name="FS-59">Ferreira and Strege (2005), p. 59.</ref> With three doors open the pressure differences were generally low, and with all doors closed the opening forces approached the limit.<ref name="FS-59" /> Sprinklered existing buildings were later exempted, and the towers were never fitted with stair pressurisation.<ref name="FS-60" />

After the bombing of 1993, many stair doors were held open by people leaving the building. This increased the airflow up the stairs caused by the stack effect, and much smoke spread through the stairs.<ref name="FS-59" /> The stairs met the code requirements for exit capacity, but they could not take all occupants quickly, so the doors stayed open.<ref name="FS-59" />

A later computer study found stair pressurisation feasible with the doors closed. It also found that relief venting was needed at the top of each stair section to keep door forces acceptable under stack effect. With several doors open a slightly positive pressure could be kept, but not with all doors open, as in a mass evacuation.<ref name="FS-60" /> A review committee concluded that stair pressurisation would not work in a mass evacuation with more than three stair doors open at once. It also found that the New York code lacked rules for periodic testing and maintenance.<ref name="FS-xxxiv">Ferreira and Strege (2005), p. xxxiv.</ref>

For the events of 11 September 2001, the investigation concluded that stair pressurisation would have been ineffective. The stairs ran through 110 storeys and changed position through transfer corridors, and the full evacuation of the undamaged floors would have lowered the stair pressure below the required value.<ref name="FS-111">Ferreira and Strege (2005), p. 111.</ref> In the impact zone the stairs were damaged and smoke entered them directly. Stair pressurisation is designed to keep smoke out of a stair, and its supply air would have been ineffective in diluting such large amounts of smoke.<ref name="FS-111" /> At the time, the main US building codes required stair pressurisation in new high-rise buildings.<ref name="FS-109">Ferreira and Strege (2005), p. 109.</ref> Full-scale tests in the 1970s had shown such systems to be effective against unsprinklered test fires.<ref name="Evans-71">Evans et al. (2005), p. 71.</ref>

== See also ==
* [[Evacuation lift]]
* [[Positive pressure ventilation]]
* [[Network model]]

== Notes ==
<references />

== References ==
* Australian Building Codes Board. (2020a). ''Data sheets: Fire Safety Verification Method'' (Handbook annex, Print version 1.1).
* Evans, D. D., Peacock, R. D., Kuligowski, E. D., Dols, W. S., & Grosshandler, W. L. (2005). ''Active fire protection systems'' (NIST NCSTAR 1-4). National Institute of Standards and Technology.
* Ferreira, M. J., & Strege, S. M. (2005). ''Smoke management systems'' (NIST NCSTAR 1-4D). National Institute of Standards and Technology.
* 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, Office of Nuclear Reactor Regulation.
* Klote, J. H. (2002). Smoke control. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-274–4-291). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Smoke Control]]

Water mist system

2026-11-02T17:13:27Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''water mist system''' is a fixed [[fire suppression system]] that discharges water as a very fine spray. In fire protection, water mist means a spray with no drops larger than 1.0 mm, and with much more of its water in very fine drops than a standard sprinkler produces.<ref name="MB-311">Mawhinney and Back (2002), p. 4-311.</ref> Splitting water into fine drops increases the surface available for heat transfer, so each kilogram absorbs heat faster and less water is needed to suppress a fire.<ref name="MB-311" /> Water mist acts mainly by cooling the flame and the hot gases, by diluting the oxygen with steam and by wetting the fuel. Blocking radiation and effects on the flame chemistry play a smaller part.<ref name="MB-313">Mawhinney and Back (2002), p. 4-313.</ref>

The technology grew rapidly in the 1990s, driven by the phase-out of halon extinguishing gases and by new rules for sprinklers on passenger ships.<ref name="MB-312">Mawhinney and Back (2002), p. 4-312.</ref> It is used both as a halon alternative for flammable liquid hazards in machinery spaces and as a sprinkler equivalent for ordinary combustibles.<ref name="MB-313" /> Unlike sprinklers, water mist systems cannot yet be designed from first principles. Each design has to be linked to full-scale fire tests of the specific manufacturer's equipment, because nozzles that pass the same test can use very different amounts of water.<ref name="MB-320">Mawhinney and Back (2002), p. 4-320.</ref><ref name="MB-325">Mawhinney and Back (2002), p. 4-325.</ref>

== Definition and drop sizes ==
The term water mist was adopted in the early 1990s by the committee responsible for NFPA 750, the US standard for water mist systems. It separates this technology from fixed water spray systems under NFPA 15 and from sprinkler systems under NFPA 13.<ref name="MB-311" /> Strictly speaking, a fire protection mist is not a true mist. In the scientific sense, a mist lies between an aerosol, with particles of about 5 µm, and a fog, with droplets of 10 to 100 µm. Droplets smaller than about 20 µm stay airborne for a long time.<ref name="MB-311" /> A fire protection mist contains a range of drop sizes, many in the true mist range and some considerably larger.<ref name="MB-311" />

The advantage of fine drops was known long before the term existed. Research in the 1950s showed good results in cooling and extinguishing diffusion flames with sprays of mean drop diameter below 0.3 mm. Smaller drops have more surface area per unit mass and settle more slowly, so they have more time to absorb heat and evaporate.<ref name="MB-311" /> For decades this knowledge was not applied widely. High operating pressures, small orifices that could become blocked by corrosion products and doubts about maintenance were technical concerns. Standard sprinklers and halon gases were also cheaper alternatives.<ref name="MB-311" />

A single number such as a mean diameter says little about the spread of drop sizes in a spray. NFPA 750 therefore describes a spray by a curve of cumulative volume against drop diameter.<ref name="MB-317">Mawhinney and Back (2002), p. 4-317.</ref> Two points on that curve are commonly quoted. The value Dv0.90 is the diameter below which 90 % of the water volume is contained, and Dv0.50 is the diameter that divides the volume into two equal halves.<ref name="MB-317" /> In measurements on four commercial nozzles, even the coarsest spray carried at least 50 % of its volume in drops smaller than 300 µm, and another spray carried about 80 %.<ref name="MB-318">Mawhinney and Back (2002), p. 4-318.</ref>

== Development and applications ==
Three events in the 1980s made fine sprays economically attractive. After an aircraft fire at Manchester in 1984, an international programme developed a cabin spray system. It extended tenable conditions by 7 minutes using about 10 litres of water, but regulators judged the cost per life saved too high to make it mandatory.<ref name="MB-312" /> The Montreal Protocol of 1987 began the phase-out of ozone-depleting halons, which released research funds for alternative agents, water among them.<ref name="MB-312" />

The third event was a ruling of the International Maritime Organization (IMO). After several fires with many deaths on Scandinavian ferries, it required sprinklers on all ships carrying 35 or more passengers from 1995.<ref name="MB-312" /> Ship designers regard sprinklers as heavy, and extra weight on upper decks can affect stability. Water mist promised the same protection with less water and less weight.

As a result, water mist has two main fields of use. Where flammable liquids are present, or where clean gaseous agents were chosen to avoid water damage, it serves as a halon alternative. Where ordinary cellulosic materials are protected, it serves as an alternative to a [[sprinkler system]].<ref name="MB-313" /> By 2000 the marine sector was the strongest market. Land-based niches included machinery modules, turbine enclosures, metro station plant rooms and clean-room wet benches.<ref name="MB-335">Mawhinney and Back (2002), p. 4-335.</ref> A water mist system installed in a London hotel instead of sprinklers offered smaller pipes, which eased retrofitting in a heritage building, and a lower water demand.<ref name="MB-335" />

Other guidance refers to water mist only briefly. The International Fire Engineering Guidelines (2005) state that the effect of water mist and gaseous systems should be assessed from their design specifications and the relevant standards, and do not cover these calculations themselves.<ref name="IFEG-2.7-8">National Research Council of Canada et al. (2005), p. 2.7-8.</ref> Dutch guidance on car parks with electric vehicles names a sprinkler or water mist system as an effective way to keep a fire limited to a small number of cars.<ref name="VG-36">Van der Graaf et al. (2021/2023), p. 36.</ref>

== How water mist suppresses fire ==
Mawhinney and co-workers describe three primary and two secondary [[extinguishing mechanisms]] for water mist. The primary ones are cooling of the gas phase, oxygen depletion with dilution of the fuel vapour, and wetting and cooling of the fuel surface. The secondary ones are attenuation of radiation and kinetic effects on the flame chemistry.<ref name="MB-313" /> All mechanisms usually act to some degree, but their relative importance differs between liquid and solid fuels.<ref name="MB-313" />

[[File:fse-water-mist-system-v1.svg|thumb|upright=1.5|alt=Cross-section of a room with ceiling nozzles spraying fine mist onto a pool fire; labels mark flame cooling, steam diluting oxygen and wetting of the fuel.|The three primary mechanisms of water mist: gas phase cooling, oxygen depletion and dilution by steam, and wetting and cooling of the fuel surface. Radiation attenuation acts as a secondary mechanism. Based on Mawhinney and Back (2002), pp. 4-313 to 4-314.]]

Gas phase cooling removes heat from the flame and the hot gases as the drops evaporate. Fine drops evaporate faster, so they extract more heat. If the flame temperature falls below a limiting value of about 1,600 K (1,326 °C) for diffusion flames, [[flame extinction]] follows.<ref name="MB-313" /> Laboratory work on small counterflow flames showed that water mist can be more effective than Halon 1301 on a mass basis. That holds only if nearly all the water reaches the flame, and in full-scale compartment fires the efficiency is far lower.<ref name="MB-313" />

When a drop turns into steam, its volume increases by more than three orders of magnitude. Close to the flame, this expansion can disrupt the flow of air into the flame. In a closed room, the steam together with the oxygen consumed by the fire lowers the oxygen concentration of the whole space.<ref name="MB-314">Mawhinney and Back (2002), p. 4-314.</ref> If it drops below the limiting oxygen concentration, about 13 % for most hydrocarbon fuels, the fire goes out.<ref name="MB-314" /> In full-scale water mist tests in obstructed machinery spaces, the oxygen concentration at extinction averaged 14.5 % for heptane spray fires and 13.5 % for pool fires.<ref name="BE-184">Beyler (2002b), p. 2-184.</ref>

For solid fuels and for liquids with a flash point above room temperature, wetting and cooling of the fuel surface is often the dominant mechanism. It reduces the release of fuel vapour until the mixture above the surface is too lean to burn.<ref name="MB-314" /> Water mist and steam also reduce the radiant heat reaching nearby objects, which helps to prevent fire spread to unburnt fuel.<ref name="MB-314" /> Fire models such as the Fire Dynamics Simulator therefore calculate the absorption and scattering of thermal radiation by water droplets.<ref name="MG-13">McGrattan et al. (2026), p. 13.</ref><ref name="MG-74">McGrattan et al. (2026), p. 74.</ref>

The enclosure itself strengthens these effects. The first contact with mist cools the hot, oxygen-poor gases in the upper layer, and the spray drives them down to the seat of the fire.<ref name="MB-314" /> Some manufacturers found that switching the spray off and on again, known as cycling, speeds up extinguishment in enclosures. Liu and co-workers showed that cycling extinguished fires faster and with less water than continuous discharge.<ref name="MB-315">Mawhinney and Back (2002), p. 4-315.</ref> Not every effect is favourable. A brief flare-up of the flame has been measured when mist first reaches it.<ref name="MB-314" />

=== Water needed for cooling ===
Wighus related the cooling capacity of a spray to the size of the fire with the spray heat absorption ratio (SHAR). It is the rate at which evaporating water absorbs heat divided by the heat release rate of the fire:<ref name="WI-313">Wighus (1995), as cited in Mawhinney and Back (2002), p. 4-313.</ref>

<math display="block">\mathrm{SHAR} = \frac{\dot{Q}_\mathrm{w}}{\dot{Q}_\mathrm{f}}</math>

{| class="wikitable"
|+ Symbols
! Symbol !! Description !! Unit
|-
| <math>\mathrm{SHAR}</math> || spray heat absorption ratio || –
|-
| <math>\dot{Q}_\mathrm{w}</math> || rate of heat absorbed by evaporating water || kW
|-
| <math>\dot{Q}_\mathrm{f}</math> || heat release rate of the fire || kW
|}

For an unconfined propane flame with optimum application, a ratio as low as 0.3 was enough. Values around 0.6 were found for more realistic machinery space conditions, where small flames survive in shielded corners and cause re-ignition.<ref name="WI-313" /> The ratio is well below 1 because the flame only has to be cooled to its limiting temperature, not to ambient temperature.<ref name="MB-313" /> Heating water from 298 K to 373 K takes 0.32 kJ/g and evaporating it another 2.26 kJ/g, so each gram that turns into steam absorbs 2.58 kJ.<ref name="TE-100">Tewarson (2002), p. 3-100.</ref>

As an example, take the 750 kW pan fire used in tests by the National Research Council Canada in a 60 m³ enclosure.<ref name="MB-327">Mawhinney and Back (2002), p. 4-327.</ref> With a SHAR of 0.3, the evaporating water must absorb 0.3 × 750 = 225 kW. Divided by 2.58 kJ/g, this gives about 88 g/s, or 5.3 litres per minute of water that actually evaporates in the flame. With the more realistic ratio of 0.6, the amount doubles to about 175 g/s, or 10.5 litres per minute.<ref name="WI-313" /><ref name="TE-100" /> Typical water mist nozzles deliver 5 to 80 litres per minute each.<ref name="MB-329">Mawhinney and Back (2002), p. 4-329.</ref>

[[File:fse-water-mist-system-v5.svg|thumb|upright=1.5|alt=Bars of 5.3 and 10.5 l/min evaporated water for a 750 kW fire at SHAR 0.3 and 0.6, beside a band of 5 to 80 l/min per nozzle.|Minimum water that must evaporate in the flame to cool a 750 kW fire, for spray heat absorption ratios of 0.3 and 0.6, compared with the flow of a single water mist nozzle. Based on Mawhinney and Back (2002), pp. 4-313, 4-327 and 4-329, and Tewarson (2002), p. 3-100.]]

The result is a lower bound, not a design flow. In real systems, the share of the mist that reaches the flame and evaporates there is almost unpredictable and cannot be controlled.<ref name="MB-313" /> The SHAR is therefore mainly of interest as a building block for computer models of suppression.<ref name="MB-313" /> Measured mist concentrations needed around a diffusion flame range from 100 to 200 g/m³, and 50 g/m³ has extinguished heptane flames. In a real machinery space, such local concentrations are extremely difficult to predict or control.<ref name="MB-314" />

== Spray characteristics ==
A spray is fully described by its drop size distribution, cone angle, discharge velocity, mass flow rate and momentum. Together these govern nozzle spacing and the maximum ceiling height.<ref name="MB-317" /> Listed nozzles usually produce solid cones of 90° or 120°.<ref name="MB-317" /> The mass flow of a nozzle depends on the water pressure and the total area of its orifices.<ref name="MB-317" />

Drop size alone does not decide which spray is better. In one comparison, a nozzle with about 80 % of its volume in drops below 300 µm discharged only 5 litres per minute. A coarser nozzle discharged 12 kg/min and therefore delivered more fine water in total, about 6 kg/min against 4 kg/min.<ref name="MB-318" /><ref name="MB-319">Mawhinney and Back (2002), p. 4-319.</ref> Discharge rate, cone angle and velocity also affect how well the mist mixes with the fire gases.<ref name="MB-319" />

Spray momentum combines the mass of the drops and of the air they drag along with their velocity. Its direction matters. A spray directed against the rising fire plume penetrates the flame and carries steam to the seat of the fire. A spray moving in the same direction as the plume mixes poorly, and the steam is carried away from the fuel.<ref name="MB-317" /> Mist does not spread through a room as a gas does. Its concentration can fall by more than half within one metre sideways from the spray pattern, so higher-performance systems use high-velocity sprays to mix the mist through the space.<ref name="MB-320" />

Additives can improve performance. Alkali salts and the surfactants of foam concentrates help to extinguish fires, and inert gases mixed with the mist support oxygen depletion in enclosed spaces.<ref name="MB-319" /> In the United States, the Environmental Protection Agency accepted water mist as a halon alternative in occupied spaces only with potable water or natural seawater. Anyone proposing an additive must show that it poses no health risk.<ref name="MB-319" />

== System types ==
Water mist systems are classified in three ways: by the way they are applied, by how the spray is produced and by the operating pressure.<ref name="MB-327" /> NFPA 750 distinguishes total compartment application, local application and zoned application systems.<ref name="MB-327" /> A total compartment system has open nozzles throughout the room, all of which discharge when a single control valve opens, in the manner of a deluge system.<ref name="MB-327" /> NUREG-1805 uses the term deluge water mist system for a system whose open nozzles discharge mist simultaneously.<ref name="IS-F21">Iqbal and Salley (2004), p. F-21.</ref>

[[File:fse-water-mist-system-v3.svg|thumb|upright=1.5|alt=Three room plans: nozzles discharging over the whole room, nozzles around one machine only, and nozzles discharging only in the zone around the fire.|Total compartment application (left), local application (centre) and zoned application (right). Based on Mawhinney and Back (2002), pp. 4-327 to 4-328.]]

Total compartment systems benefit from the enclosure, which captures heat, confines steam and recirculates oxygen-depleted gases. They are nevertheless not "total flooding" systems in the sense used for gases. A fixed quantity of mist does not put out a fire; the mist must be discharged continuously long enough to gain control. Unlike gaseous systems, which need a tight room, these systems can work with fairly large openings.<ref name="MB-327" /> Their main drawback is a high total water demand, because many nozzles each discharging a little still add up to a lot of water.<ref name="MB-327" />

A local application system sprays directly onto an object or hazard, for example a diesel engine or a turbine on an open floor. With little confinement of heat or steam, it relies almost entirely on gas phase cooling and wetting of the fuel.<ref name="MB-328">Mawhinney and Back (2002), p. 4-328.</ref> A zoned system protects only part of an enclosure, which usually reduces the water flow to a quarter or a third of a total compartment system. A system of individually heat-activated nozzles, as on ships, is in effect a zoned system in which each nozzle acts as detector and valve.<ref name="MB-328" />

The design of a water mist system for a 9 km cable [[tunnel fire|tunnel]] in Auckland shows why zoning can be necessary. Opening all nozzles over the full length would have required very large mains and booster pumps. Heat-activated nozzles were not suitable either, because the ventilation air at 5 m/s would carry the heat away from the nozzles nearest the fire.<ref name="MB-328" /> The chosen solution released the system in zones upstream and downstream of the fire. That made reliable detection and location of the fire the critical part of the design.<ref name="MB-328" />

By spray generation, single-fluid nozzles atomise water alone, either by forcing it through small orifices at high speed or by striking it against a surface. Twin-fluid nozzles add compressed gas as an atomising medium at the nozzle. They can work at low water pressure and have larger orifices that block less easily, but they need separate piping and storage for gas and water.<ref name="MB-329" /> Single-fluid nozzles are more widely used.<ref name="MB-329" />

By pressure, NFPA 750 defines low-pressure systems below 12 bar, intermediate-pressure systems between 12 and 34 bar and high-pressure systems above 34 bar.<ref name="MB-330">Mawhinney and Back (2002), p. 4-330.</ref> Low-pressure systems can follow standard sprinkler practice, with conventional centrifugal fire pumps.<ref name="MB-330" /> High-pressure systems are widely installed on cruise ships and in machinery spaces. They use stainless steel tubing and positive displacement pumps, or gas-driven cylinders whose pressure falls from about 140 bar to 30 to 50 bar over a 10-minute discharge.<ref name="MB-331">Mawhinney and Back (2002), p. 4-331.</ref> Mawhinney and Back report that acceptable suppression can be achieved with both high- and low-pressure systems; economic and engineering factors decide which suits an application.<ref name="MB-330" />

[[File:fse-water-mist-system-v2.svg|thumb|upright=1.5|alt=Pressure bars: sprinklers 0.5 to 7 bar, low-pressure mist below 12 bar, intermediate 12 to 34 bar, high-pressure 34 to about 140 bar.|Operating pressure ranges of standard sprinklers and of low-, intermediate- and high-pressure water mist systems. Based on Mawhinney and Back (2002), pp. 4-330 to 4-331.]]

== Design, testing and acceptance ==
There is no theoretical basis yet for designing the drop sizes, concentrations and distribution that a water mist system must deliver. Design parameters are extrapolated from large-scale tests case by case.<ref name="MB-320" /> NFPA 750 therefore requires every design to be linked to formal fire tests.<ref name="MB-325" /> Testing has shown that performance depends on the specific equipment, especially the nozzle. Two systems that pass the same test may differ in average water flux density by a factor of three.<ref name="MB-325" /> Generic design criteria that do not depend on the equipment, such as a nominal flux density for a given fire scenario, cannot yet be set.<ref name="MB-325" />

A listed system has had its design criteria validated through fire tests by a recognised laboratory. The engineer must still check that the real application is close to the tested conditions and that the pass criteria match the needs of the owner.<ref name="MB-325" /> The IMO tests for cabins and public spaces require only that the fire is controlled for ten minutes. The IMO machinery space test requires all fires to be extinguished, including small hidden fires in the bilge.<ref name="MB-325" /> A pre-engineered system has been tested for a hazard of limited size, such as a turbine enclosure, and needs no hydraulic calculations if the installer stays within its limits. Extrapolation beyond the listing turns it into an engineered system.<ref name="MB-329" />

Models can help to bound performance. A quasi steady-state zone model by Back and co-workers predicts the temperature and oxygen concentration in a machinery space from the fire size, room geometry, vent area and water flow. It assumes extinction when the oxygen falls to the limiting value, typically taken as 14 %.<ref name="MB-321">Mawhinney and Back (2002), p. 4-321.</ref><ref name="MB-322">Mawhinney and Back (2002), p. 4-322.</ref> The model agreed well with tests in rooms of 100 to 3,000 m³, but it cannot distinguish between systems with the same flow and different sprays.<ref name="MB-322" /> Computational fluid dynamics models show promise, yet still lack validated methods for drop transport and extinction.<ref name="MB-324">Mawhinney and Back (2002), p. 4-324.</ref>

Acceptance testing is more demanding than for sprinklers. NFPA 750 recommends a full discharge test where possible, especially for gas-driven systems with declining pressure.<ref name="MB-333">Mawhinney and Back (2002), p. 4-333.</ref> In one test of a twin-fluid system designed for six discharge cycles, all six discharges could be seen and heard. The pressure records showed that the compressed gas had run out before the end of the fifth cycle.<ref name="MB-333" />

== Limitations and misconceptions ==
A common misconception is that a water mist system fills a room like a gas. Mist drops settle and deposit on surfaces, and the mist must be generated and distributed continuously.<ref name="MB-320" /><ref name="MB-327" /> Effectiveness falls when the room is very large compared with the fire, because the enclosure effects become weak and the mist must then be projected to the seat of the fire.<ref name="MB-327" /> For the same reason, Dutch guidance notes that water mist is less effective in open car parks, where wind and weather reduce its action.<ref name="VG-65">Van der Graaf et al. (2021/2023), p. 65.</ref>

Some effects of mist can be harmful. Releasing mist into a very hot, tight room can cool the gases so fast that a strong negative pressure develops. At the National Research Council Canada, a wall of a 60 m³ test enclosure deflected inward and tore open along the ceiling line. That happened when mist was injected after a 750 kW pan fire had burnt for 90 seconds.<ref name="MB-327" /> The effect of water sprays on gas explosions is mixed. Some fine-mist tests reduced the overpressure by about 15 %, while coarser deluge sprays in a closed compartment increased it, because the spray made the gas mixture more turbulent.<ref name="MB-315" /><ref name="MB-316">Mawhinney and Back (2002), p. 4-316.</ref>

Water mist cannot reach every seat of fire. Dutch guidance on car parks states that neither sprinklers nor mist can extinguish a fire inside the battery pack of an electric vehicle, because the water cannot enter the pack. A system can still limit fire growth and spread to nearby vehicles and structures.<ref name="VG-37">Van der Graaf et al. (2021/2023), p. 37.</ref> In general, an active suppression system acts during the growth phase and prevents the fire from becoming fully developed.<ref name="VR-4.11">Van Rossum (2027), §4.11.</ref>

== See also ==
* [[Extinguishing agent]]
* [[Critical water application rate]]
* [[Gaseous extinguishing system]]
* [[Deluge system]]

== Notes ==
<references />

== References ==
* Beyler, C. L. (2002b). Flammability limits of premixed and diffusion flames. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 2-172–2-187). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* Mawhinney, J. R., & Back, G. G., III. (2002). Water mist fire suppression systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-311–4-337). National Fire Protection Association; Society of Fire Protection Engineers.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Tewarson, A. (2002). Generation of heat and chemical compounds in fires. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-82–3-161). National Fire Protection Association; Society of Fire Protection Engineers.
* Van der Graaf, J., Hessels, T., & Rosmuller, N. (2023). ''Fire safety of indoor car parks accommodating electrically powered vehicles''. Nederlands Instituut Publieke Veiligheid. (Original work published 2021)
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Fire Suppression]]

Beam smoke detector

2026-11-02T17:13:25Z

Lisa: Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text

A '''beam smoke detector''' is a [[smoke detector]] that sends a beam of light across the protected space to a receiver and gives an alarm when smoke weakens that beam.<ref name="IFEG-2.7-2">National Research Council of Canada et al. (2005), p. 2.7-2.</ref><ref name="IS-11-5">Iqbal and Salley (2004), p. 11-5.</ref> The sources also call it a projected beam smoke detector or an optical obscuration (beam) smoke detector.<ref name="IFEG-2.7-2" /><ref name="IS-11-5" /> It works on the photoelectric light obscuration principle, but the light path spans the area instead of lying inside a small housing.<ref name="IS-11-5" /> Because detection takes place along the whole path, it is classed as a line detector rather than a spot detector.<ref name="IS-11-5" />

Beam detectors are mostly used to protect large open areas and spaces with high ceilings.<ref name="IS-11-5" /> They can also be mounted below the ceiling of a tall space, so that they still see the smoke when it stops rising before it reaches the roof.<ref name="Milke-4-305">Milke (2002b), p. 4-305.</ref> According to the SFPE Handbook, the projected-beam detector is the only commercially available smoke detector that senses the attenuation of a light beam.<ref name="SMC-4-22">Schifiliti et al. (2002), p. 4-22.</ref> This makes its response easier to relate to the optical density of smoke calculated by fire models than the response of ionisation or light-scattering detectors.<ref name="SMC-4-22" /><ref name="SMC-4-23">Schifiliti et al. (2002), p. 4-23.</ref> The sensitivity is usually set as a total obscuration, typically between 20% and 70% of the beam.<ref name="MBIE25-28">Ministry of Business, Innovation and Employment (2025b), p. 28.</ref> Engineers convert this setting into an optical density at alarm, which depends on the length of the beam.<ref name="MBIE25-28" />

== How it works ==
A beam smoke detector has two main parts: an emitter that produces the light beam and a receiver that measures how much light arrives.<ref name="IS-11-3">Iqbal and Salley (2004), p. 11-3.</ref><ref name="McG-109">McGrattan et al. (2026), p. 109.</ref> The US fire alarm code NFPA 72, as summarised in NUREG-1805, describes the obscuration principle as a light source focused on a photosensitive sensor.<ref name="IS-11-3" /> When smoke particles enter the light path, part of the light is scattered and part is absorbed, so less light reaches the sensor.<ref name="IS-11-3" /> The detector processes this loss of signal and gives an alarm when it meets preset criteria.<ref name="IS-11-3" /> In the projected beam version the same principle is used, but the beam crosses the protected area.<ref name="IS-11-5" />

[[File:fse-beam-smoke-detector-v1.svg|thumb|upright=1.5|alt=Cross-section of a hall with an emitter on one wall, a receiver on the other and a smoke plume crossing the light beam between them.|A beam smoke detector: smoke between emitter and receiver weakens the beam; the alarm follows when the received light falls below the setting. Based on Iqbal and Salley (2004), pp. 11-3 and 11-5.]]

The quantity that matters is therefore the fraction of the original light that still reaches the receiver.<ref name="SMC-4-22" /> Researchers often call the loss of light obscuration, although Schifiliti, Meacham and Custer note that attenuation is the more accurate word.<ref name="SMC-4-22" /> The light may be absorbed, reflected or refracted by the smoke.<ref name="SMC-4-22" /> Percent obscuration is 100 times one minus the ratio of the received light to the light received without smoke.<ref name="SMC-4-22" /> The optical density expresses the same loss on a logarithmic scale, and per metre of path it has the unit m−1.<ref name="SMC-4-22" /><ref name="VR-6.4">Van Rossum (2027), §6.4.</ref>

This principle sets the beam detector apart from the two most common smoke detectors. Ionisation detectors sense smoke through a drop in the electrical conductance of ionised air in a small chamber.<ref name="IS-11-4">Iqbal and Salley (2004), p. 11-4.</ref> Light-scattering photoelectric detectors place the sensor out of the direct beam and respond to light that smoke scatters onto it.<ref name="IS-11-3" /> Neither type works by light attenuation, so their response cannot be related directly to measured optical density.<ref name="SMC-4-22" /> A scattering detector may, for example, respond at 0.029 m−1 to smoke from a smouldering cotton lamp wick but only at 0.15 m−1 to smoke from a kerosene fire.<ref name="SMC-4-23" />

Spot detectors, by contrast, cover a limited area each. NUREG-1805 gives up to 84 m² per spot detector, depending on the ceiling surface and the room height.<ref name="IS-11-5" /> A beam detector senses smoke continuously along its whole path, which is why it is listed among the line detectors.<ref name="IS-11-5" /> Like other detectors, it is normally part of a wider [[fire alarm system]] that warns occupants, calls the fire brigade and can start other building systems.<ref name="IS-11-1">Iqbal and Salley (2004), p. 11-1.</ref> Some designs use coincident detection, in which two or more detectors must operate before the system accepts an alarm.<ref name="IFEG-2.7-2" />

== Where beam detectors are used ==
NUREG-1805 states that projected beam photoelectric detectors are most often applied for large open area protection and are also useful in areas with high ceilings.<ref name="IS-11-5" /> In such spaces the actuation time of ceiling-mounted spot smoke detectors may be unacceptably long, and a projected beam detector may then be the preferable means of detection.<ref name="IS-11-6">Iqbal and Salley (2004), p. 11-6.</ref> Full-scale tests in two US Navy aircraft hangars, 15 m and 22 m high, included projected beam smoke detectors alongside spot detectors, flame detectors and sprinklers.<ref name="Stroup-3-23">Stroup et al. (2016), p. 3-23.</ref>

The main physical reason for using beams in tall spaces is [[smoke stratification]]. Smoke rises only while the plume is warmer, and therefore lighter, than the surrounding air.<ref name="Milke-4-304">Milke (2002b), p. 4-304.</ref> When the temperature in the plume falls to that of the surrounding air, the smoke stops rising and stays at that height, whatever the height of the ceiling.<ref name="SMC-4-27">Schifiliti et al. (2002), p. 4-27.</ref> This is a concern for low-energy fires and for rooms or volumes with very high ceilings.<ref name="SMC-4-27" /> In one SFPE example, a 1 kW fire of overheated cable insulation in a 7 m high air-conditioned room lifts its smoke only to about 6 m.<ref name="SMC-4-27" /> With detectors on the ceiling only, the design objective may then not be achieved.<ref name="SMC-4-27" />

Atria are the classic case. Milke reports that the floor-to-ceiling temperature difference in some atria with glazed roofs is of the order of 50 °C, because the sun heats the air under the glazing.<ref name="Milke-4-304" /> Smoke from a growing fire then stratifies below this warm layer.<ref name="Milke-4-304" /> For [[atrium smoke control]], Milke therefore advises a detector other than a ceiling-mounted one, such as a projected beam detector, to start the smoke management system.<ref name="Milke-4-305" /> The beams should be placed below the height of stratification so that they cross the rising plume.<ref name="Milke-4-305" /> Once the smoke management system runs, the warm air layer is extracted and the smoke can reach the ceiling.<ref name="Milke-4-305" />

[[File:fse-beam-smoke-detector-v2.svg|thumb|upright=1.5|alt=Atrium cross-section: a warm air layer under the glazed roof stops the smoke plume halfway; a beam below that height crosses the plume.|Intermediate stratification in an atrium: the smoke stops below a warm layer under the glazed roof, so a beam is placed below that height to intercept the plume. Based on Milke (2002b), pp. 4-304 to 4-305.]]

Large storage spaces are another application. The SFPE Handbook uses a mattress store of 50 m by 75 m with a 10 m high ceiling as a worked example for a projected beam detector.<ref name="SMC-4-26">Schifiliti et al. (2002), p. 4-26.</ref> This places the method in the setting of a [[warehouse fire]], where a large open volume dilutes the early smoke.<ref name="SMC-4-26" />

== Sensitivity and optical density at alarm ==
A beam detector does not respond to a fixed smoke concentration. It responds when the total loss of light along the beam reaches its setting, and that setting can usually be adjusted.<ref name="SMC-4-23" /> In New Zealand, C/VM2 (second edition, 2025) states that response sensitivities are typically in the range of 20% to 70% total obscuration.<ref name="MBIE25-28" /> This corresponds to a ratio of received light to clear-air light of 0.3 to 0.8.<ref name="MBIE25-28" /> To use such a setting in a fire model, the engineer needs the average optical density of smoke along the beam at the moment of alarm.<ref name="MBIE25-28" />

If the smoke is spread evenly along the beam, the optical density at alarm follows from the light ratio and the path length.<ref name="MBIE25-28" /><ref name="SMC-4-22" /> A longer beam passes through more smoke, so the same obscuration setting is reached at a lower optical density per metre.<ref name="MBIE25-28" /> The relation, as given in C/VM2, is:<ref name="MBIE25-28" />

<math display="block">D_\mathrm{u} = -\frac{1}{L}\,\log_{10}\!\left(\frac{I}{I_0}\right) \quad \text{with} \quad \frac{I}{I_0} = \frac{100 - \mathrm{OBS}}{100}</math>

{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>D_\mathrm{u}</math> || optical density of smoke per unit length at alarm || m−1
|-
| <math>L</math> || path length of the beam, from emitter to receiver || m
|-
| <math>I</math> || light intensity at the receiver with smoke present || any consistent unit
|-
| <math>I_0</math> || light intensity at the receiver without smoke || same unit as <math>I</math>
|-
| <math>\mathrm{OBS}</math> || total obscuration setting for alarm || %
|}

The relation assumes a uniform optical density along the whole beam.<ref name="MBIE25-28" /> Where the smoke is not uniform, for example where a narrow plume crosses the beam, the path can be divided into sections.<ref name="MBIE25-28" /> The optical density of each section times its length is then summed over all sections.<ref name="MBIE25-28" /> The logarithm is to base 10, in line with the SFPE definition of optical density.<ref name="SMC-4-22" /> The International Fire Engineering Guidelines (2005) give the same principle and treat the detector as activated once the optical density in the smoke layer exceeds the value at alarm.<ref name="IFEG-2.7-6">National Research Council of Canada et al. (2005), p. 2.7-6.</ref>

As a worked example, C/VM2 takes a beam with a path length of 10 m and an alarm setting of 50% total obscuration.<ref name="MBIE25-28" /> The light ratio is then (100 − 50)/100 = 0.5. The optical density at alarm is −(1/10) × log10(0.5) = 0.030 m−1.<ref name="MBIE25-28" /> For comparison, C/VM2 uses 0.097 m−1 at alarm for spot smoke detectors.<ref name="MBIE25-27">Ministry of Business, Innovation and Employment (2025b), p. 27.</ref>

The same calculation shows the effect of the setting and the beam length. For the 10 m beam, a setting of 20% gives about 0.0097 m−1 and a setting of 70% gives about 0.052 m−1. For a 50 m beam at 50%, the optical density at alarm falls to about 0.0060 m−1, one fifth of the value for the 10 m beam. These values follow directly from the relation above, but they hold only for smoke spread evenly along the beam.<ref name="MBIE25-28" />

[[File:fse-beam-smoke-detector-v3.svg|thumb|upright=1.5|alt=Line chart of optical density at alarm against obscuration setting for beams of 10 m, 50 m and 100 m; longer beams alarm at lower optical density.|Optical density of evenly spread smoke at alarm against the obscuration setting, for three beam lengths. Based on Ministry of Business, Innovation and Employment (2025b), p. 28.]]

The SFPE Handbook example of the mattress store works in the other direction.<ref name="SMC-4-26" /> The design goal is to detect a flaming fire before about 350 g of polyurethane has burned, with the smoke assumed to be mixed evenly through the 37,500 m³ room.<ref name="SMC-4-26" /> Using a mass optical density of 0.22 m²/g, the smoke gives about 0.002 m−1.<ref name="SMC-4-26" /> Over a 75 m beam, 70.8% of the light still arrives, which is a total obscuration of 29.2%.<ref name="SMC-4-26" /> The detector, adjustable from 20% to 70% in steps of 10%, would therefore have to be set at about 30% or less.<ref name="SMC-4-26" />

== Use in performance-based design ==
In New Zealand, C/VM2 (second edition, 2025) makes the detection time the first term of the required safe egress time, before notification, pre-travel activity and travel.<ref name="MBIE25-26">Ministry of Business, Innovation and Employment (2025b), p. 26.</ref> The same document requires the detection time of an automatic warning system to be found by fire modelling.<ref name="MBIE25-27" /> This can be an algorithm that includes a ceiling jet correlation, or a computational fluid dynamics (CFD) model that solves the velocity, temperature and smoke concentration directly.<ref name="MBIE25-27" /> Detectors must be simulated with the values of the C/VM2 detector table, whatever make and model will actually be installed.<ref name="MBIE25-27" /> For projected beam detectors, that table gives no single number: the optical density at alarm follows from the beam path length and the design obscuration setting.<ref name="MBIE25-27" />

Two modelling approaches are described in the SFPE Handbook.<ref name="SMC-4-23" /> A zone model calculates one optical density for a uniform hot layer and multiplies it by the path length of the beam.<ref name="SMC-4-23" /> This assumes that the smoke is spread evenly along the beam, which may not be valid.<ref name="SMC-4-23" /> A field model calculates the optical density in many points or segments between emitter and receiver and sums the products of optical density and segment length.<ref name="SMC-4-23" /> The Fire Dynamics Simulator (FDS) follows this second approach.<ref name="McG-109" /> The user sets the positions of emitter and receiver and the total obscuration for alarm, and FDS integrates the obscuration over the path from the soot concentration in each grid cell.<ref name="McG-109" />

A CFD study of a large atrium building, reported in the SFPE Handbook, shows how a beam can start a smoke control system in practice.<ref name="CK-3-212">Cox and Kumar (2002), p. 3-212.</ref> The building was 150 m long, 62 m wide and 52 m high, with a central atrium 40 m high.<ref name="CK-3-212" /> Extract fans with a capacity of 165 m³/s were to be started by a beam detector at first-floor level.<ref name="CK-3-212" /> The model predicted that this beam would detect a fire on the atrium floor after 65 s, at a heat release rate of 200 kW.<ref name="CK-3-214">Cox and Kumar (2002), p. 3-214.</ref> The doors that then opened to admit replacement air deflected the plume, and smoke spread across the atrium floor for about 8 minutes.<ref name="CK-3-214" />

== Limits and uncertainty ==
The calculation of beam response is simple, but its inputs are uncertain. Smoke measurements depend on the wavelength of the light used.<ref name="SMC-4-22" /> Optical density data are only useful for predicting a beam detector if they were measured at the wavelength of the detector's own light source.<ref name="SMC-4-22" /> Many detectors also use response algorithms instead of simple thresholds, and manufacturers generally do not publish them.<ref name="SMC-4-22" />

Mixing assumptions are a second source of error. The uniform-path method assumes that the smoke is spread evenly along the beam.<ref name="SMC-4-23" /> In reality a rising plume may cross only a short part of a long beam.<ref name="MBIE25-28" /> Mass optical densities from small-scale tests also introduce error when they are applied to large spaces, and the correlation breaks down for complex fires.<ref name="SMC-4-26" /> Air movement from ventilation systems can dilute the smoke or carry it away, and there are no quantitative methods to estimate these effects.<ref name="SMC-4-26" />

For these reasons, the SFPE Handbook treats all smoke detector estimates as rough.<ref name="SMC-4-23" /> Because their errors are not generally known, it advises against comparing the estimates with egress time calculations or the time to untenable conditions.<ref name="SMC-4-23" /> It recommends using them to compare how the response of one detector changes when its spacing or location changes.<ref name="SMC-4-23" /> C/VM2 takes a regulatory route instead: it fixes the inputs so that all designs are assessed on the same basis.<ref name="MBIE25-27" />

Stratification calculations have their own limits. The SFPE method for the height of smoke rise ignores heating and air-conditioning systems and warm or cold walls.<ref name="SMC-4-27" /> A beam placed below the expected stratification height can intercept the plume, but the plume widens only gradually as it rises.<ref name="Milke-4-305" /> Its width is estimated at 25% to 50% of the height above the top of the fuel.<ref name="Milke-4-305" /> The placement of the beam is therefore part of the analysis, not only its sensitivity.

== See also ==
* [[Aspirating smoke detection]]
* [[Detector activation time]]
* [[Heat detector]]
* [[Flame detector]]
* [[Multi-sensor detector]]

== Notes ==
<references />

== References ==
* Cox, G., & Kumar, S. (2002). Modeling enclosure fires using CFD. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 3-194–3-218). National Fire Protection Association; Society of Fire Protection Engineers.
* 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, Office of Nuclear Reactor Regulation.
* McGrattan, K., Hostikka, S., Floyd, J., McDermott, R., Vanella, M., Mueller, E., & Paul, C. (2026). ''Fire Dynamics Simulator technical reference guide: Vol. 1. Mathematical model'' (6th ed., NIST Special Publication 1018-1). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.1018
* Milke, J. A. (2002b). Smoke management in covered malls and atria. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-292–4-310). National Fire Protection Association; Society of Fire Protection Engineers.
* Ministry of Business, Innovation and Employment. (2025b). ''Verification Method C/VM2: Framework for fire safety design'' (2nd ed.).
* National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). ''International fire engineering guidelines'' (2005 ed.). Australian Building Codes Board.
* Schifiliti, R. P., Meacham, B. J., & Custer, R. L. P. (2002). Design of detection systems. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-1–4-43). National Fire Protection Association; Society of Fire Protection Engineers.
* Stroup, D., Lindeman, A., McGrattan, K., Peacock, R., Overholt, K., Joglar, F., LeStrange, S., & Montanez, S. (2016). ''Verification and validation of selected fire models for nuclear power plant applications: Supplement 1'' (NUREG-1824, Supplement 1; EPRI 3002002182). U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research; Electric Power Research Institute.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.

[[Category:Detection & Alarm]]