Skip to content
Upcoming courses

Fire modelling

Tools
From Fire Safety Academy Wiki
Revision as of 18:14, 2 November 2026 by Lisa (talk | contribs) (Add Intermediate SVG figures with Fire Safety Academy watermark at planned locations; preserve article text)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Contents
  1. 1What a fire model is
  2. 2Types of fire models
  3. 3Inputs
  4. 4Use in design and regulation
  5. 5Verification, validation and sensitivity
  6. 6Limits
  7. 7See also
  8. 8Notes
  9. 9References

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.[1] 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.[2][3] 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.[4][5][6]

The three main types are hand calculations, zone models and computational fluid dynamics (CFD), and they differ greatly in detail and computing time.[7] 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.[8][9] 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.[1] For that reason, guidance in several countries asks that models are verified, validated and used within their field of application.[10][11]

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.[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.[1] However polished a simulation looks, it does not take over the thinking of the fire safety engineer.[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.[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.[12] 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.[13]

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.[2] Models for large fire compartments often compare a large compartment with one of, for example, 1,000 m².[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.[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.[14] 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.[7]

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.[7] 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.[15] 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.[14]

A zone model divides each room into two layers: a hot upper layer of smoke and a cooler lower layer.[16] The fire acts as a source of energy and mass, and its plume carries air from the lower layer into the upper layer.[16] Conservation equations for mass and energy are solved for each layer, and the properties within each layer are taken as uniform.[8] Before flashover, a zone model uses these two zones. When conditions approach flashover, the room may be treated as one well-mixed zone instead.[17]

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.[16] In 1985, ASET-B became the first zone model written for personal computers.[4] The most used zone models in the Netherlands include the TNO filling-time model, CFAST, OZone and the NEN 6093 method.[18]

Computational fluid dynamics (CFD) divides the space into a grid of cells and solves the equations of fluid flow numerically in every cell.[9] Walton calls these "field models" and notes that they need substantial computer resources and are relatively complex to use.[14] Unlike zone models, CFD was developed outside the fire community and imported into it.[19]

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.[19] 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.[20] 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.[21]

Inputs

According to Introduction to Fire Safety Engineering, the most important inputs are the heat release rate and the energy of the fire.[22] CFAST does not predict fire growth on burning objects; the user specifies the heat release rate.[23] 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.[24] 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.[25] 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).[25]

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.[22]

The building inputs include the dimensions and shape of rooms, material properties, the number, size and position of openings, and any mechanical ventilation.[22] 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.[26]

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.[27] 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.[28]

Use in design and regulation

In performance-based design, fire models often supply the time to untenable conditions.[29] 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.[30] 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.[6] The method is intended for professional fire engineers who are proficient in fire engineering modelling methods.[31]

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.[32] The choice between hand calculations, zone models and CFD for each enclosure is agreed as part of the performance-based design brief.[29] The handbook notes that zone and CFD models treat the separate phenomena "in a more holistic manner" and are therefore in common use.[15]

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.[12] 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.[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.[33] In the United States, the standard NFPA 805 for nuclear power plants requires that fire models are verified and validated.[10]

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.[10] The predictions sometimes agreed with the data, but they also showed both under- and over-prediction in some circumstances.[34] 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.[35] In Australia, the FSVM handbook asks for a review that checks whether a method is used within its field of application with suitable inputs.[11]

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.[4]

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.[4] 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.[36]

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.[37] Its simple point-source model of fire radiation limits its accuracy within a few fire diameters.[23]

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.[38] In a zone model, the plume reaches the ceiling instantly, so transport times are not accounted for.[39] 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.[40]

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.[9] Capturing the chemical reaction zone of a fire would need a mesh finer than 1 mm.[19] 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.[3] 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.[24]

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.[24] It also advises a sound background in physics and much practical experience before CFD is used.[24] Notarianni adds that uncertainties "hard-wired" into scientific tools often go unrecognised.[36]

See also

Notes

  1. ↑ 1.0 1.1 1.2 1.3 1.4 1.5 Van Rossum (2027), §8.1.
  2. ↑ 2.0 2.1 2.2 2.3 2.4 Van Rossum (2027), §8.2.
  3. ↑ 3.0 3.1 McGrattan et al. (2026), p. 3.
  4. ↑ 4.0 4.1 4.2 4.3 Walton (2002), p. 3-190.
  5. ↑ Van Rossum (2027), §6.1.
  6. ↑ 6.0 6.1 Ministry of Business, Innovation and Employment (2025b), p. 14.
  7. ↑ 7.0 7.1 7.2 Najafi et al. (2007), p. 2-13.
  8. ↑ 8.0 8.1 Peacock et al. (2026), p. 1.
  9. ↑ 9.0 9.1 9.2 Van Rossum (2027), §11.1.
  10. ↑ 10.0 10.1 10.2 Najafi et al. (2007), p. xv.
  11. ↑ 11.0 11.1 Australian Building Codes Board (2020b), p. 108.
  12. ↑ 12.0 12.1 Phillips (2002), pp. 5-112–5-113.
  13. ↑ Phillips (2002), p. 5-113.
  14. ↑ 14.0 14.1 14.2 Walton (2002), p. 3-189.
  15. ↑ 15.0 15.1 Australian Building Codes Board (2020b), p. 109.
  16. ↑ 16.0 16.1 16.2 Quintiere (2002a), p. 3-162.
  17. ↑ Van Rossum (2027), §9.1.
  18. ↑ Van Rossum (2027), §9.5.
  19. ↑ 19.0 19.1 19.2 Cox and Kumar (2002), p. 3-194.
  20. ↑ McGrattan et al. (2026), p. 2.
  21. ↑ Van Rossum (2027), §11.5.
  22. ↑ 22.0 22.1 22.2 Van Rossum (2027), §8.4.
  23. ↑ 23.0 23.1 Peacock et al. (2026), p. 3.
  24. ↑ 24.0 24.1 24.2 24.3 Van Rossum (2027), §11.4.
  25. ↑ 25.0 25.1 Van Rossum (2027), §8.6.
  26. ↑ Van Rossum (2027), §8.5.
  27. ↑ Ministry of Business, Innovation and Employment (2025b), p. 17.
  28. ↑ Ministry of Business, Innovation and Employment (2025b), p. 16.
  29. ↑ 29.0 29.1 Australian Building Codes Board (2020b), p. 21.
  30. ↑ National Research Council of Canada et al. (2005), p. 2.5-2.
  31. ↑ Ministry of Business, Innovation and Employment (2025b), p. 7.
  32. ↑ Australian Building Codes Board (2020b), pp. 107–108.
  33. ↑ Najafi et al. (2007), p. 1-2.
  34. ↑ Najafi et al. (2007), p. xvi.
  35. ↑ Najafi et al. (2007), p. 2-33.
  36. ↑ 36.0 36.1 Notarianni (2002), p. 5-44.
  37. ↑ Peacock et al. (2026), p. 2.
  38. ↑ Van Rossum (2027), §9.2.
  39. ↑ Quintiere (2002a), p. 3-163.
  40. ↑ McGrattan et al. (2026), pp. 1–2.

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.
Retrieved from "https://fire-safety.academy/w/index.php?title=Fire_modelling&oldid=2616"