Upward flame spread

Upward flame spread is the advance of burning along a combustible surface in an upward direction, assisted by the rising flow generated by the fire. On a vertical wall, the flow carries flame and hot gases towards material above the burning region. Quintiere classifies this as wind-aided spread: the relevant flow can be generated by buoyancy, without an external wind.

The visible flame and the surface region supplying combustible gases have different boundaries. Flames can extend above the actively pyrolysing material and heat the unburned surface ahead. The surface spread rate describes the movement of the pyrolysis front, rather than simply the movement of the visible flame tip.

Upward spread can accelerate because the growing burning region increases heat release and extends the region heated by the flame. This feedback competes with burnout of material already involved. The outcome therefore depends on the material, exposure, geometry and flow. Upward orientation creates the potential for rapid growth, but does not establish one universal spread speed or growth law.

Heat feedback and the direction of flow

Surface flame spread requires heat to pass from the existing fire to material that has not yet become involved. For a solid, this heating produces combustible gases through thermal decomposition. Those gases burn near the surface and contribute further heat. The moving pyrolysis boundary is the material-side expression of that continuing process.

In upward spread, buoyancy aligns the hot flow with the direction of advance. The flame and combustion products can heat a substantial length of surface above the pyrolysis front. In opposed-flow flame spread, the flow is directed against the advance, and the region supplying forward heating is much shorter. Quintiere identifies this difference in heating extent as central to the different behaviour.

The flame can heat material above the advancing pyrolysis front. Based on Quintiere (2002), pp. 2-251–2-252.

Wind-aided spread is the broader category. Quintiere includes spread driven by external flow and spread beneath a ceiling within that category, as well as upward wall spread. The classification describes the relationship between the flow and the advancing front. It does not mean that all these geometries have the same flame attachment, heat transfer or spread correlation.

The relevant heat flux is distributed over that forward region. A high value at one location does not by itself determine the spread rate. The length exposed, duration of heating and material response also matter. Quintiere's simplified comparison explicitly includes both the flux and the distance over which it heats unburned material.

This explains why comparing only peak heat fluxes can be misleading. Quintiere reports an orientation study in which downward spread had a higher local maximum than upward spread. The upward configuration nevertheless exposes a longer region ahead of the front. A point measurement and an extended preheating process describe different aspects of the heat feedback.

Ignition position can therefore change the subsequent development even when the material and pilot flame are similar. Iqbal and Salley use a curtain as an example. Ignition near its lower end permits rapid upward spread; ignition near its upper end leads to slower downward spread. The example illustrates the role of orientation and available fuel above the ignition location.

Acceleration, flame extent and burnout

A growing pyrolysis region can supply more fuel vapour and increase heat release. A longer flame can then heat more material ahead of the front, allowing the burning region to grow further. Quintiere relates the flame extent to heat release per unit width and uses this connection in models of upward spread. The feedback links material burning behaviour with flame geometry.

Flame height therefore helps describe the region capable of heating the surface, but it must be distinguished from pyrolysis height. The flame may extend beyond the point at which the solid is producing combustible gases. Treating the flame tip as the boundary of already burning material would overstate the pyrolysing area and alter the inferred heat release.

Burnout introduces a second moving boundary behind the advancing front. Material that has exhausted its available fuel no longer contributes in the same way to the burning region. Quintiere represents the active region between the burnout position and the pyrolysis front. Its length can grow, approach a limit or decline, depending on the relative progress of these boundaries.

Growth depends on forward spread and the loss of active fuel through burnout. Based on Quintiere (2002), pp. 2-246 and 2-252–2-253.

The competition between spread time and burning duration is consequently important. If new material ignites rapidly while previously ignited material continues burning, the active region can expand. If the material burns out before sufficient new area becomes involved, growth can slow or stop. The upward direction alone does not remove this competition.

Some simplified limiting cases give exponential growth, but that result depends on particular relationships and negligible burnout. Quintiere discusses other observed and modelled behaviours, including an active region approaching a steady limit. An exponential fit should therefore be tied to its assumptions and conditions. It is not an inherent law applying to every vertical combustible surface.

The feedback also depends on configuration. A relation between heat release and flame extent established for one arrangement is not automatically valid for another. Quintiere requires configuration-specific descriptions of flame heat flux and length when developing a spread solution. This requirement becomes important when moving from a flat wall to a corner, ceiling or constrained channel.

Different parts of the active surface can also be at different stages of their burning history. Quintiere combines the contributions of previously ignited regions when formulating the heat release that drives flame extent. A single peak heat release rate multiplied by the entire involved area does not preserve that timing. The history of ignition and subsequent burning is part of the feedback calculation.

A simplified relation and worked example

Quintiere expresses the spread speed through the distance between the pyrolysis front and the flame extent, divided by an associated ignition time. The relation connects the extent of forward heating with the time required for the heated material to become involved. It is a simplified representation, including an assumption of uniform flame heat flux over that region.

V = dx_p/dt = (x_f − x_p) / τ

Symbol	Meaning	Unit
V	Upward movement rate of the pyrolysis front	m/s
xp	Pyrolysis-front position measured from the chosen origin	m
xf	Flame extent measured from the same origin	m
t	Time	s
τ	Ignition time associated with the represented flame heat flux	s

For a vertical surface, the positions can be expressed as heights above a common origin. The quantity in the numerator then represents the forward heating length in the simplified model. The associated ignition time must correspond to the heating conditions being represented. An ignition time measured under an unrelated exposure cannot be substituted merely because it is available.

Consider hypothetical positions of 0.60 m for the pyrolysis front and 1.20 m for the flame extent. With an associated ignition time of 60 s, the heating length is 0.60 m. The calculated instantaneous spread speed is 0.010 m/s, or 10 mm/s. These values are selected to illustrate the relation, rather than taken from a particular material test.

The simplified relation separates the effects of heating length and associated ignition time. Based on Quintiere (2002), p. 2-252, with hypothetical inputs.

If the flame extent becomes 1.50 m while the other example inputs remain unchanged, the heating length becomes 0.90 m. The calculated speed increases to 15 mm/s. Alternatively, retaining the original heating length but reducing the associated ignition time to 30 s gives 20 mm/s. The examples separate the influence of heating extent from that of material ignition response.

Hypothetical comparison	Forward heating length (m)	Associated ignition time (s)	Calculated speed (mm/s)
Base illustration	0.60	60	10
Longer heated region	0.90	60	15
Shorter ignition time	0.60	30	20

These calculations are conditional snapshots. They do not predict how flame extent, pyrolysis height or ignition response will change during a real fire. A time-dependent model requires additional relationships for heat release, heat transfer and burnout. Holding the example inputs constant throughout an assumed growth period would add a further assumption that has not been demonstrated here.

The relation also presumes a flame extending ahead of the pyrolysis front and a positive associated ignition time. If those conditions no longer represent the fire, the simplified expression is outside its intended interpretation. A zero or negative numerator should not automatically be treated as a validated prediction of extinction or reverse spread.

Material response and the imposed exposure

Material properties enter through the heating needed to produce combustible gases and through the subsequent burning history. Quintiere distinguishes thermally thin and thermally thick behaviour in simplified spread analyses. A thin specimen can approach a uniform temperature through its thickness; a thick specimen develops a temperature gradient. The heat required to advance the front consequently depends on the relevant thermal response.

Thickness alone is not a universal classification independent of time and exposure. Quintiere discusses the effects of thermal penetration and the adjacent substrate, particularly for intermediate thicknesses. A surface layer tested on one backing can therefore have a different heating response from the same layer in another construction. The represented material arrangement needs to accompany the material name.

Preheating changes the energy needed to reach the ignition condition. Van Rossum explains that upward orientation transfers heat to material above the burning region and can accelerate subsequent involvement. Additional heating in an enclosure can also raise the temperature of fuel ahead of the front. The initial surface condition is therefore part of the spread problem.

There is no single flame heat flux appropriate to every upward fire. Quintiere reports approximately 25 kW/m² ahead of the pyrolysis zone for moderate turbulent wall fires with flame extents below about 1.4 m. The same discussion shows that larger and laminar fires can exceed this value. The numerical approximation belongs to the described exposure regime.

Babrauskas describes a broader cluster of wall-flame measurements between approximately 25 and 45 kW/m², suggesting 35 kW/m² as representative in that context. He also notes the additional heating possible within a hot gas layer. These accounts illustrate why exposure selection needs its experimental basis and scenario conditions, rather than an unexplained universal flux assumption.

The heat release history used in a spread calculation must correspond to that exposure. Quintiere identifies cone calorimeter data as a possible source of material burning information, while warning that the relevant flame-spread flux need not equal the selected test irradiance. The transfer from a material test to a spread model therefore requires a reasoned exposure match.

Geometry, confinement and external walls

A wall affects the surrounding flow as well as providing a possible combustible surface. Iqbal and Salley explain that a fuel package near a wall has less surrounding cool-air entrainment than one burning freely. This can produce higher flames and temperatures. Surface orientation and proximity to boundaries therefore influence the exposure that drives further spread.

Corners and channels can introduce additional geometric effects. Quintiere describes how escalator sidewalls in the King's Cross fire caused flames to remain close to the steps. This is the trench effect context cited in his discussion of orientation. It demonstrates that inclined, constrained spread can differ from the behaviour represented by an isolated vertical flat-wall model.

In a facade fire, vertical travel of flames does not necessarily mean that the cladding itself is sustaining upward surface spread. The New Zealand commentary distinguishes spread over facade materials from fire plumes projecting through openings and exposing openings above. Identifying which mechanism is active is necessary before selecting a surface-spread model.

The external-wall assembly can also behave differently from a small isolated material specimen. The commentary identifies connections and jointing details as reasons why small-scale testing may be unsuitable. A large-scale facade fire test can address the behaviour of an arrangement whose relevant details are not represented by the material specimen alone. The cited commentary supports this distinction without defining one universal facade test procedure.

Interpreting tests and model results

A useful assessment identifies the measured or predicted front. Visible flame extent, pyrolysis height, burnout position and heat release describe different features of the event. Quintiere's model separates these quantities because their relationships drive the growth. Reporting only a maximum flame height can leave the actual surface involvement and its rate of advance unresolved.

A test rating is likewise not automatically a spread velocity. Quintiere describes the historical Steiner tunnel test as a complex wind-aided configuration and notes that it does not measure the pyrolysis or burnout positions. Its output cannot simply replace those variables in a physical spread equation. The test arrangement and reported quantity need to match the intended use of the data.

The material, orientation, dimensions, substrate, ignition source and external heating should therefore remain explicit when interpreting a result. Quintiere concludes that spread depends on the fuel's configuration and ambient conditions as well as its properties. Applying an experimental or modelled result beyond those conditions requires evidence that the controlling heat-transfer and burning processes remain represented.

References

• Babrauskas, V. (2002). Heat release rates. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., Chapter 3-1). National Fire Protection Association.
• Iqbal, N., & Salley, M. H. (2004). Fire dynamics tools (FDTs): Quantitative fire hazard analysis methods for the U.S. Nuclear Regulatory Commission fire protection inspection program (NUREG-1805). U.S. Nuclear Regulatory Commission.
• Ministry of Business, Innovation and Employment. (2013). Commentary for Building Code Clauses C1–C6 and Verification Method C/VM2.
• Quintiere, J. G. (2002). Surface flame spread. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., Chapter 2-12, pp. 2-246–2-257). National Fire Protection Association.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering. Fire Safety Academy.