Smouldering

Smouldering is slow, relatively low-temperature and flameless Combustion sustained by heat released when oxygen reacts directly with a fuel surface. The reaction occurs on the material itself, unlike a flame in which gaseous fuel reacts. It commonly occurs in porous, char-forming materials and can spread within a fuel bed with little external sign. This can significantly delay recognition.

The process presents two principal fire hazards. It can convert a relatively large proportion of fuel into toxic compounds, although more slowly than many flaming fires. It can also provide a pathway to flaming from an initial heat source too weak to ignite a flame directly. These hazards can develop over long periods and in concealed locations.

Smouldering should not be confused with smoke produced by forced thermal decomposition. Pyrolysis can gasify an organic material with little or no oxidation, whereas smouldering requires surface oxidation. The presence of smoke alone therefore does not establish which process is occurring. Assessment has to address hidden reaction zones, accumulated exposure, fuel-bed geometry and possible later flaming, rather than relying on visible flame or one universal spread rate for every material and airflow condition.

Conditions that sustain smouldering

Finely divided fuel provides a large surface area per unit mass, which facilitates oxygen attack. A permeable aggregate allows oxygen to reach the reaction zone by diffusion and convection. If the aggregate also resists heat loss, the limited energy released by surface oxidation can maintain the reaction.

Char formation is central for many cellulosic fuels. Char is typically richer in carbon and has greater surface area per unit mass than the original material. Oxygen can react with this porous residue to produce mainly carbon monoxide and carbon dioxide, supplying much of the heat for sustained smoulder propagation.

A burning cigarette combines several favourable features: finely divided fuel, a permeable structure and thermal insulation. It can also initiate smouldering in upholstery and bedding. These composite products involve both fabric and filling, so their combined response can differ from either material tested alone.

The conditions needed to initiate smouldering are not necessarily sufficient for it to spread away from the initiation region. Continued propagation depends on heat generation, oxygen transport and heat transfer to adjacent fuel. The factors controlling a later transition to flaming are different again.

Self-heating can contribute to the initial temperature rise when oxidation within a material releases energy faster than it is lost. Fuel size and shape influence this balance by changing heat retention. Self-heating describes energy accumulation, while smouldering describes the sustained surface-combustion process that can follow under suitable conditions.

Propagation and oxygen supply

After initiation, the smouldering zone often consumes oxygen as quickly as it reaches the reaction region. Propagation is then strongly controlled by oxygen transport. Thick fuel layers slow oxygen delivery but also retain heat, allowing very slow reaction zones to persist.

In reverse smouldering, oxygen moves opposite to the direction of front propagation. It can reach the reaction zone by diffusion or by forced airflow through unburnt fuel. In forward smouldering, oxygen and the front move in the same general direction. These relative directions change preheating, heat retention and product composition.

Forward and reverse smouldering differ in relative oxygen-flow direction. Based on Ohlemiller (2002), pp. 2-201–2-203.

Ohlemiller describes sawdust experiments in which smouldering took about two weeks to move upwards through a layer 1 m deep. In another set of reverse-smoulder experiments, velocities were of the order of 0.0001 m/s across several fuels under the stated airflow conditions. These results illustrate possible timescales but do not provide a universal smouldering speed.

The diffusion-controlled sawdust case has a useful scaling relation. As the front advances through unburnt fuel towards the surface, the distance through which oxygen must travel changes. Ohlemiller reports that the penetration time is approximately proportional to the square of the layer depth. The proportionality constant must be obtained from a measurement on the fuel configuration being considered:

$$t_L=A{L}^2$$

Symbol	Description	Unit
$t_L$	Time for the front to penetrate the fuel layer	s
$L$	Layer depth	m
$A$	Empirical coefficient fitted to the same fuel and oxygen-supply conditions	s/m²

For an illustrative scaling calculation, represent the reported two weeks at 1 m as 14 days. Keeping the same coefficient, a layer half as deep gives one quarter of that penetration time: 3.5 days. The result follows from the source's square-law model, rather than from a separate test of that shallower layer:

$$t_{0.5\mathrm{m}}=14{\left(\frac{0.5}{1.0}\right)}^2=3.5\mathrm{d}\mathrm{a}\mathrm{y}\mathrm{s}$$

This estimate applies to approximately one-dimensional, diffusion-dominated propagation under comparable conditions. A non-uniform deposit can contain easier airflow paths, and buoyancy can introduce convection through the fuel. The fitted constant and square-law assumption cannot then be transferred automatically. The calculation illustrates the importance of oxygen transport and depth; it does not predict an incident's duration or extinction time.

Deep reaction zones can remain hidden until they approach a free surface. Oxygen access increases near that surface, while buoyancy and local airflow change heat and mass transfer. A fire can consequently show little sign internally before becoming more obvious at the exterior.

Real fuel beds often have two- or three-dimensional temperature and composition gradients. The upper part of a horizontal layer usually receives oxygen more readily than deeper material, producing an inclined reaction zone. Simple one-dimensional descriptions therefore apply only to selected configurations.

Heat loss, airflow and fuel-bed size

A self-sustaining reaction requires heat generation to keep pace with heat loss. Thin layers have a large surface relative to their volume and can lose energy too rapidly. Below a minimum thickness, smouldering therefore stops. That thickness depends on fuel type, particle size, bulk density and oxygen supply.

Fuel-bed thickness affects the heat balance for sustained smouldering. Based on Ohlemiller (2002), pp. 2-201–2-205.

Airflow can either strengthen or weaken the process. Increased oxygen supply can raise heat release and accelerate spread, but sufficiently strong flow can add cooling and disrupt the reaction zone. The response depends on flow direction, fuel-bed geometry and whether propagation is forward or reverse.

Fuel particle size affects more than chemical surface area. It changes permeability, insulation and the minimum layer depth needed for sustained spread. In the sawdust and cork experiments described by Ohlemiller, particle size had less influence on velocity than flow configuration, but it strongly affected limiting thickness.

Scale can also allow smouldering to pass beneath an intended barrier. In laboratory studies of deep fuel beds, heated and partly decomposed material below an initial reaction zone supported a second wave moving in the opposite direction. Ohlemiller warns that inert covering layers may slow rather than stop such fires.

Smoke, carbon monoxide and detection

Smouldering can produce substantial Carbon monoxide relative to the fuel converted. The yield and concentration vary with fuel and airflow, so a single value cannot represent all smouldering fires. The slow release rate does not remove the hazard because exposure can accumulate over a long period in an enclosure.

In one flexible-polyurethane-foam experiment, carbon monoxide comprised 6% to 7% of evolved gases on a molar basis at an imposed air velocity of 0.0015 m/s. Cellulosic insulation in another test varied from about 10% to 22% as airflow changed. These are configuration-specific measurements, not safe-exposure criteria or universal material yields.

The aerosol from smouldering differs from sooty flaming smoke. For the cellulosic-insulation source studied by Ohlemiller, the mass mean particle size was 2 to 3 micrometres, larger than cigarette smoke and much larger than particles from the flaming source used for comparison. The fuel bed also filtered part of the aerosol before it escaped.

This behaviour affects Smoke detector response. Ohlemiller links the large aerosol particles with relatively poor sensitivity of ionisation detectors to the realistic smouldering source examined. Deep smouldering may also remain undetectable until it nears an exposed surface because the fuel bed traps products.

The weak heat source presents a further detection problem. Reverse-smoulder sources discussed in the handbook produced only a few hundredths of a kilowatt in the stated configurations. Their buoyant plumes might not reach the ceiling of a room, limiting the transport of combustion products to ceiling-mounted devices.

Transition to flaming

Smouldering-to-flaming transition requires a flammable mixture of pyrolysis gases and air and sufficient heat to ignite it at the same place and time. Increased oxygen supply, larger scale and reduced heat losses can promote both conditions. The transition is therefore possible but not inevitable.

Forward smouldering can be particularly favourable in some fuel beds because heat transfer intensifies the reaction near its leading edge. Cavities formed there can collect fuel gases and stabilise initial flames. Different materials and configurations produce other mechanisms, so there is no single transition temperature or airflow for all smouldering fires.

Experiments with upholstered furniture demonstrate the wide time range. In a survey cited by Ohlemiller, about two thirds of tested chairs and beds made the transition, with reported times from 22 to 306 minutes after smouldering began. These results describe the tested objects and should not be used as a countdown for an actual incident.

Reported smouldering-to-flaming times in a furniture-test survey. Based on Ohlemiller (2002), p. 2-208.

Neighbouring materials can change the outcome. Retardants that prevented transition in an isolated cellulosic insulation layer became less effective when that layer touched unretarded wood. Heat from the smouldering zone could ignite the adjacent material even when the original fuel did not flame.

Detection, extinction and reignition

Effective extinction requires more than suppressing visible glow. Oxygen exclusion can stop reaction temporarily, but oxygen readmission can cause reignition while the fuel remains hot. Ohlemiller states that oxygen removal may have to continue until cooling is sufficient, and that obtaining this cooling can be exceptionally difficult in a large porous bed.

Extinguishing agents can also miss concealed reaction zones. In grain silos, higher-permeability channels can allow an agent to bypass significant smouldering regions. Emptying before complete extinction introduces another hazard because combustible dust can become dispersed. The source treats this as a practical configuration problem rather than a simple choice of agent.

Assessment after an intervention therefore includes the temperature and location of remaining fuel, not only the atmosphere above it. A deep bed can retain reactive material for long periods. Restored airflow can then change both oxygen supply and heat loss, with outcomes that depend on the new conditions.

Buildings and outdoor fuel beds

Upholstery and bedding combine fabrics, fillings and crevices, producing behaviour that a single-material specimen may not reproduce. A cellulosic fabric can supply heat to polyurethane foam while both compete for oxygen. Ohlemiller reports that this combination can smoulder over a wider range of conditions than the foam alone.

Closed-room mattress experiments show why slow development still requires attention. In two of five cigarette-initiated cotton-mattress tests described in the handbook, flaming followed smouldering after 65 to 80 minutes. In other bedding and upholstery tests, flaming usually required two to three hours of prior smouldering. These observations belong to the tested products and rooms, and are not general predictions.

Solid wood presents a different configuration from loose fibres. Its lower permeability limits oxygen access to the decomposition region, and a single layer may lose too much heat to sustain smouldering without additional exposure. Ohlemiller reports a requirement of about 10 kW/m² of external heat input in the cited solid-wood configuration. Other wood forms and arrangements can behave differently.

Peat and organic material on a forest floor can retain smouldering after a flaming front has passed. The process is spatially irregular and can later support renewed flaming. This outdoor example involves a porous fuel bed with oxygen entering largely from above, rather than a manufactured uniform specimen.

Engineering use and uncertainty

Design-fire descriptions sometimes include a smouldering phase. The 2005 International Fire Engineering Guidelines distinguish smouldering fires that remain flameless from those that later become flaming. The transition remains difficult to predict from fuel identity alone because airflow, geometry and heat losses affect its onset.

The same older guidelines use a notional maximum heat release rate below 5 kW for an illustrated smouldering design fire. This is a modelling choice in that guidance, not a universal physical maximum. Experimental source strengths and spread rates need to be tied to the fuel bed and airflow from which they were obtained.

Ohlemiller concludes that quantitative guidance for realistic smoulder propagation, detection, toxic production and transition remains limited. Experimental results can support closely analogous situations, but dissimilar materials and geometries require caution. Fuel chemistry alone is not enough to predict spread when oxygen transport and heat retention dominate.

Toxic assessment needs the material and combustion mode as well as carbon monoxide. The 2005 guidelines identify carbon monoxide as the principal toxic species for many fires, but caution that atypical materials require broader analysis. They specifically note that smouldering polyurethane foam can yield significant hydrogen cyanide.

Source terms should therefore be based on appropriate experimental data for the combustible and the smouldering condition. A flaming-fire yield cannot be substituted automatically, and a slow source can continue for much longer than the test used to characterise it. Both yield and duration influence the exposure calculated for an enclosure.

See also

• Fire development stages
• Firebrands
• Hot work
• Mass timber
• Timber in fire

References

• Australian Building Codes Board, National Research Council of Canada, International Code Council, & Department of Building and Housing New Zealand. (2005). International fire engineering guidelines (2005 ed.). Australian Building Codes Board.
• Ohlemiller, T. J. (2002). Smoldering combustion. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 2-200–2-210). National Fire Protection Association.