External flaming

External flaming is combustion outside a fire compartment by flames extending through an opening. It commonly occurs when a compartment produces more combustible gases than the available internal oxygen can burn. Those gases can mix with fresh air outside and release part of their energy beyond the enclosure. Window flames rise along the exterior and transfer heat to surrounding surfaces through radiation and convection.

The distinction between fuel production and internal combustion is central to predicting this behaviour. A lower calculated temperature inside the room does not necessarily indicate a smaller exterior exposure. Calculations must distinguish the potential energy of the pyrolysed fuel, the energy released inside, and the energy available for combustion outside. The opening geometry and the receiving surface then determine how that external combustion affects the building.

Physical development and external exposure

A growing room fire heats its enclosure and produces a buoyant layer of combustion products. Following the development of a post-flashover fire, intense heating can continue to generate fuel vapour throughout the compartment. However, the rate at which fresh oxygen enters through the available openings can restrict combustion inside. Excess combustible material then leaves with the outgoing gases. The location of energy release changes even though the fuel originates within the room.

External flaming is therefore closely associated with a ventilation-controlled fire. The relevant comparison concerns oxygen demand and oxygen supply, rather than fuel quantity alone. A compartment can contain considerable combustible material while releasing only part of its potential power internally. Fuel chemistry, the rate of pyrolysis and the thermal environment all affect the resulting balance. The exterior flame cannot be characterised from the internal temperature alone.

Oxygen-limited compartment burning can leave combustible gases to burn outside. Based on Walton and Thomas (2002), pp. 3-172–3-175, and Lattimer (2002), pp. 2-292–2-293.

The distinction also separates an exterior flame exposure from burning of the exterior construction itself. A facade fire may involve combustible surfaces exposed to a window flame, but the initiating window flame can exist without those surfaces burning. Lattimer describes experiments with target walls above openings specifically to quantify this external exposure. The tests varied window geometry and fire size while measuring the resulting heat transfer above the opening.

Heat can reach a receiving surface through flame contact, hot gases and radiation. The applicable mechanism depends on the relationship between the surface and the flame. Assessment of fire spread between buildings consequently requires a representation of the exterior source and its relationship to the exposed construction. A calculation of internal compartment temperature supplies only part of that information. The available window-flame measurements describe particular geometries and cannot define a universal separation distance.

Pyrolysis, oxygen supply and energy release

Pyrolysis rate measures the mass of solid fuel converted into combustible gases per unit time. Heat release rate measures the energy actually released by combustion per unit time. These quantities are related through an effective heat of combustion when the relevant fuel burns under the assumed conditions. They need separate treatment when some fuel leaves the compartment before releasing that energy. Treating all mass loss as internal combustion would assign external energy to the room.

The internal oxygen inventory initially contributes to combustion alongside oxygen entering through openings. A restriction based solely on the incoming oxygen is therefore inappropriate while that inventory remains available. The OZone model described by Cadorin and Franssen tracks an oxygen balance and applies its external-flaming limit after the initial inventory has been consumed. Its combustion chemistry uses a wood surrogate requiring 1.27 kg of oxygen per kilogram of fuel. This coefficient belongs to that specified chemistry.

Under the model's ventilation-controlled condition, internal heat release is calculated from the incoming oxygen. The pyrolysis rate remains at its prescribed value. The following expression reproduces that model assumption using heat release in kilowatts and effective heat of combustion in kilojoules per kilogram:

Q̇_inside = (ṁ_O₂,in / 1.27) ΔH_c,eff

Symbol	Meaning	Unit
Q̇inside	Heat release rate inside the compartment	kW
ṁO₂,in	Incoming oxygen mass flow	kg/s
1.27	Oxygen demand of the modelled wood surrogate	kg oxygen/kg fuel
ΔHc,eff	Effective heat of combustion	kJ/kg

This expression does not predict the complete external flame. It allocates energy within an oxygen-limited compartment calculation. Additional information is needed to establish whether and where the escaping combustible gases burn. The model assumes external combustion while retaining the prescribed mass loss, so its resulting internal temperature depends on that choice. When oxygen becomes available again during decay, the model returns to the fuel-controlled relation between mass loss and heat release.

The ventilation factor provides a useful description of opening-controlled air supply in simplified compartment calculations. It does not eliminate the distinction between incoming air, incoming oxygen and combustible mass leaving the room. Walton and Thomas also emphasise that different fuels can pyrolyse at different rates under similar thermal conditions. Greater production of volatiles can increase the exterior hazard while reducing the calculated temperature inside the compartment.

Combustion assumptions in compartment models

Cadorin and Franssen describe three combustion options with different purposes. Their predetermined option applies the specified heat release and pyrolysis curves without modifying them for oxygen availability. This can represent experiments where those quantities have been measured independently, or calculations in which fuel-controlled burning is assumed. It does not establish that the prescribed internal heat release is achievable under any ventilation condition.

Their external-flaming option preserves the prescribed pyrolysis curve while limiting internal combustion through the oxygen balance. Some potential energy consequently leaves the compartment with combustible gases. The modelled fuel mass can disappear at its original rate while the room receives less heat than the input fuel curve implies. This is the relevant option when the analysis intentionally represents a division between internal and external combustion.

The extended-duration option instead reduces pyrolysis during ventilation control and eventually releases the fuel energy inside over a longer period. The paper presents this as a structural design assumption intended to avoid uncertainty in the maximum pyrolysis rate. It cannot simultaneously represent the same unmodified mass-loss history as the external-flaming option. The two choices change the location or timing of energy release and therefore answer different questions.

Van Rossum makes the same distinction when explaining oxygen-limited zone calculations. The chosen combustion option should accompany the reported temperature history because that history cannot identify the assumption by itself. A lower peak with external flaming and a longer internal fire with extended duration are different outcomes of the energy allocation. Comparing them requires attention to the complete heating duration as well as the peak temperature.

Changes to the opening can alter these conditions during the fire. Glass breakage in fire, automatic opening and firefighting actions are examples identified in the OZone paper. The described implementation allows opening size to vary with time or temperature. This supports explicit scenarios for a changing opening, but it does not provide a universal glass-failure temperature or guarantee the time of an actual change.

The reported validation of the OZone compartment model used measured fuel mass-loss histories as inputs. Those comparisons included 36 wood-fuel compartment tests with different openings and fuel loads, rather than a general validation of facade fire spread. The authors selected external flaming so that ventilation could reduce internal heat release without changing the prescribed fire duration. This choice is consistent with using an independently measured pyrolysis history. Agreement with internal temperature measurements, however, does not by itself establish the detailed shape or heat flux of the exterior flame.

Window geometry and flame extent

Window flames usually rise along the building exterior because the hot combustion products are buoyant. Experiments reviewed by Lattimer measured the heat flux to walls above full-scale fire compartments. The arrangements included different opening sizes and aspect ratios, with the target wall extending above the fire room. These geometric changes affected the measured exposure even when the general combustion arrangement remained similar.

The chapter reports measured heat fluxes as high as 200 kW/m² in the reviewed investigations. This is an experimental observation, not a uniform flux for every window flame. The plotted propane fire strengths were ideal heat release rates obtained from supplied fuel flow and heat of combustion. They must not be substituted directly for the heat release occurring outside the compartment. That distinction matters when applying the separate exterior flame-length relation.

Lattimer presents the following relation for average flame extent using the exterior heat release rate. The effective opening diameter is defined by the second expression, including the factor of two inside the denominator:

L_f = 0.0321 (Q̇_outside / D)^2/3

D = 2 √[H_o W_o / (2π)]

Symbol	Meaning	Unit
Lf	Distance from opening bottom to average flame height	m
Q̇outside	Heat release rate outside the compartment	kW
D	Effective opening diameter used in this correlation	m
Ho, Wo	Opening height and width	m

The stated vertical datum is the bottom of the opening. A distance above the lintel therefore requires subtraction of the opening height. The result describes an average flame extent, rather than a permanent sharp boundary or an instantaneous maximum. It also supplies no direct surface temperature, ignition time or structural response. Those consequences require a heat-transfer assessment appropriate to the receiving surface and exposure.

The correlation measures average flame extent from the opening bottom. Based on Lattimer (2002), p. 2-293; values are calculated for the illustrative case.

Worked mass and energy example

Consider an illustrative compartment calculation after its initial oxygen inventory has been exhausted. Assume a prescribed pyrolysis rate of 0.25 kg/s and an effective heat of combustion of 16,000 kJ/kg. Assume that 0.127 kg/s of oxygen enters and is fully consumed internally. These are selected calculation inputs for the wood-surrogate model, not measurements from a reported fire. The purpose is to demonstrate the distinction between mass loss and internal combustion.

The prescribed fuel stream has a potential heat release of 4,000 kW. Dividing the oxygen flow by the model's oxygen demand gives an internal fuel consumption of 0.100 kg/s. Multiplication by the effective heat of combustion gives 1,600 kW released inside. The remaining 0.150 kg/s represents combustible fuel leaving under the assumed external-flaming allocation. Its corresponding potential heat release is 2,400 kW.

A hypothetical oxygen balance allocates 1600 kW inside and 2400 kW to potential external combustion. Method: Cadorin and Franssen (2003), pp. 419–421. Based on Cadorin and Franssen (2003), pp. 419–421, equation 48; hypothetical inputs.

Suppose that this remaining fuel burns outside and that the opening is 2.60 m wide and 1.37 m high. These dimensions match a geometry discussed by Lattimer, but the combined numerical case remains illustrative. The effective diameter is 1.506 m. Substituting the assumed exterior heat release gives an average flame extent of 4.38 m from the opening bottom, or 3.01 m above the lintel.

The energy ledger is the main result: 4,000 kW of fuel potential divides into 1,600 kW internally and an assumed 2,400 kW externally. Applying all 4,000 kW inside would represent a different model. Using all 4,000 kW in the exterior flame relation would also represent a different source. Neither substitution follows from the specified oxygen-limited allocation.

The chosen example also separates an oxygen flow from an air flow. Its 0.127 kg/s input refers to oxygen, rather than the total incoming gas mixture. Using the same number as total air would supply less oxygen and change the internal allocation. The OZone formulation assigns fresh air an oxygen mass fraction of 23%, which belongs in that conversion when total air flow is the available input.

Interpretation and limitations

An external-flaming assessment needs a consistent connection between the compartment fire, the opening and the exterior exposure. The fuel history determines the available combustible stream. The oxygen balance determines how much combustion the internal model permits. The exterior geometry then governs flame development and heat transfer to a target. Each stage requires its own inputs; a single reported heat release rate can be ambiguous without its location and definition.

The OZone formulation is a specific engineering representation of oxygen-limited burning. Its wood-surrogate chemistry and uniform compartment oxygen accounting should remain attached to the quoted expression. They do not constitute a detailed chemical prediction of flame extinction or mixing outside a window. Similarly, the window correlation summarises a defined class of exterior flame behaviour rather than every opening, facade shape or external flow condition.

Energy leaving an opening can take more than one form. Hot gases carry sensible energy already released within the room, radiation can pass through the opening, and combustible gases can carry chemical energy. External flaming concerns combustion of the latter stream. Counting all outgoing heat as exterior combustion would combine different mechanisms and could assign the same energy twice. The compartment energy balance and exterior source definition should keep these contributions distinct.

Surface heat-flux measurements also require careful interpretation. Lattimer explains that water-cooled gauges generally remain cooler than the construction being assessed. Their measured total flux includes radiation and convection, while a heated surface has a different net heat balance. A measured gauge exposure cannot be treated as absorbed heat in every material without considering surface temperature and thermal properties. These distinctions determine what an external-flaming result can support in a subsequent assessment.

See also

• Large-scale facade fire test

References

• Cadorin, J.-F., & Franssen, J.-M. (2003). A tool to design steel elements submitted to compartment fires: OZone V2. Part 1: Pre- and post-flashover compartment fire model. Fire Safety Journal, 38, 395–427.
• Lattimer, B. Y. (2002). Heat fluxes from fires to surfaces. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 2-269–2-296). National Fire Protection Association.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering.
• Walton, W. D., & Thomas, P. H. (2002). Estimating temperatures in compartment fires. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 3-171–3-188). National Fire Protection Association.