Available online at www.sciencedirect.comProceedings

Proceedings of the Combustion Institute 32 (2009) 2489–2496

www.elsevier.com/locate/proci

of the

CombustionInstitute

Carbon emissions from smouldering peat in shallowand strong fronts

Guillermo Rein a,*, Simon Cohen a, Albert Simeoni b

a BRE Centre for Fire Safety Engineering, University of Edinburgh, King’s Buildings, AGB, Edinburgh EH9 3JL, UKb UMR CNRS 6134 – SPE, Universita di Corsica, France

Abstract

A series of experiments of shallow and strong smouldering fronts in boreal peat have been conductedunder laboratory conditions to study the CO and CO2 emissions. Peat samples of 100 mm by 100 mm incross section and 50 mm in depth were smouldered in the cone calorimeter apparatus. Two laboratory vari-ables, moisture content and the external heat flux, are varied over a wide range of values to establish dif-ferent burning rates and front thicknesses. This provides a novel framework to study smoulderingdynamics by varying the controlling mechanisms and providing burning conditions that otherwise cannotbe obtained. Measurements of the burning rate and gas flow, yields and ratio for CO and CO2 are reportedat steady state burning conditions. Average mass flow rates per area of smouldering front are reported herefor the first time to be 0.27 g/s/m2 for CO and 0.65 g/s/m2 for CO2. This CO2 mass flux is about 3000 timeslarger that the natural decomposition flux from peatlands. The CO yield in dry base is 17% g/g and the CO2

yield 42% g/g. The CO and CO2 total yield is of 59% g/g, and the CO to CO2 ratio was on average 0.43.The results indicate that peat with high moisture content smoulders producing larger CO2 yield but thesame CO yield compared to dryer peat. This suggests that smouldering of biomass at lower moisture con-tents develops wider pyrolysis fronts that release a larger fraction of other carbon-containing gas species.� 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Biomass burning; Carbon monoxide, Carbon dioxide; Emission factor; Smoldering

1. Introduction

Smouldering fires can consume more than halfof the biomass burned during temperate, borealand tropical wildfires [1,2]. Thus, it contributessignificantly to atmospheric emissions fromwildfires.

Smouldering is a flameless form of combus-tion, deriving its heat from heterogeneous reac-tions occurring on the surface of a solid fuel

1540-7489/$ - see front matter � 2009 The Combustion Institdoi:10.1016/j.proci.2008.07.008

* Corresponding author. Fax: +44 131 650 6554.E-mail address: G.Rein@ed.ac.uk (G. Rein).

when heated in the presence of oxygen [3]. Thefundamental difference between smouldering andflaming combustion is that in smouldering, oxida-tion of the reactant species occurs on the surfaceof the solid rather than in the gas phase. The char-acteristic temperature, spread rate and heatreleased during smouldering are low comparedto those in the flaming combustion of a solid.Smouldering fires in forest biomass propagateon average at around of 10–30 mm h�1 and thepeak temperature is around 550–650 �C [4].

Smouldering fires of peatlands and otherorganic-soil systems represent a large perturbationof the global atmospheric chemistry. When active,

ute. Published by Elsevier Inc. All rights reserved.

Fig. 1. Depth vs. time sketch of a downward smoulder-ing front showing the evolution of the front structure.Vertical lines marked the beginning and end of thesteady-state regime.

2490 G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496

the burning of ground and subsurface biomasslayers can last for long periods of time and emitlarge quantities of combustion products causingthe deterioration of the air quality [5,6]. Such firesare difficult to extinguish despite extensive rainsand firefighting attempts. The fire removes layersof soil and the prolonged heating kills roots, seedsand plant stems. By propagating in the subsur-face, it offers the means for flaming combustionto re-establish during wildfires in unexpected loca-tions (e.g., across a fire break) and at unexpectedtimes (e.g., long after burn out of the flame front).The ignition, depth, duration, and extent ofsmouldering fires are governed primarily by thediffusion of heat and oxygen through the porousfuel layers from/to the propagation front [7,3].The peat properties affecting these two mecha-nisms are in turn, the moisture and inert contents,botanical composition, bulk density, matrix per-meability, and the presence of cracks and deepchannels.

Peat is an accumulation of partially decayedvegetation matter. Peatlands are the most wide-spread of all wetland types in the world. Theycover over 4 � 106 km2 (3% of the Earth’s landsurface). They are important ecosystems for awide range of wildlife habitats supporting biolog-ical diversity, hydrological integrity and carbonstorage. These ecosystems hold one third of theworld’s soil carbon and 10% of global freshwaterresources. Their total carbon pool exceeds that ofthe world’s forests and is comparable to that ofthe atmosphere [8].

Peatlands play an important role in the globalcarbon balance and recent environmentalchanges, such as climate change and human activ-ities including drainage, peat harvesting and airpollution have raised questions regarding the longterm stability of these carbon sinks. After the 2002study of the Borneo episodes [5], smouldering bio-mass fires have started to be seen as an emergingthreat, posing a global risk with social, economicand environmental consequences in both the shortand the long terms.

The largest peat fires registered to date tookplace in Indonesia during the El Nino dry seasonof 1997 (previously occurred in 1982, 1991, and1994, and later repeated in 1998, 2002, 2004, and2006) and lasted for several months and destroyedover 104 km2 of peat swamp with a loss of peat lay-ers between 0.2 and 1.5 m deep. According to Pageet al. [5] the peat fires accounted for 20% of the totalburn area but produced 94% of total emissions. Thesmoky haze covered large parts of South East Asiafor weeks, disrupting shipping and aviation andcausing very large economic losses, long term dam-age to the environment and health care problems. Ithas been estimated that the 1997 fires releasedbetween 0.8 to 2.6 Gton of carbon emissions intothe atmosphere, equivalent to 13–40% of the globalfossil fuel emissions of that year [5].

Research in smouldering fires is at presentmodest and little is known [4,5,9,10]. More exper-imental and theoretical studies are needed, espe-cially to explore the issues of ignition,propagation dynamics, emissions and extinctionmechanisms. The aim of this paper is to examinethe carbon emissions from smouldering peat.

2. Emissions and structure of the smouldering front

Biomass smouldering fires can burn in shallowor deep fronts [10]. Each has different dynamics.Shallow fronts, the objective of this study, burnnear the free surface and are open to the atmo-sphere, thus having large supplies of oxygen avail-able but being exposed to convective heat losses.Deep subsurface fires burn many meters belowthe ground, and thus have a limited supply of oxy-gen but are insulated from heat losses to theatmosphere.

Shallow smouldering fronts in the organic lay-ers of the ground propagate laterally and down-wards. The front structure is similar to that ofthe forward-smouldering configuration, wherethe drying and the pyrolysis fronts move aheadof the oxidization front where the heat is released.When the top surface of the fuel is exposed toambient air and an imposed heat flux, heat andoxygen have to diffuse to the lower layers whichremain colder and in oxygen-poor atmospheresbelow a depth of a few centimeters. The structureof a smouldering front is formed of four differentsub-fronts. The spread rates, widths and overlap-ping of these sub-fronts depend on the particularthermal and chemical conditions. This structureis illustrated in Fig. 1 for a front propagatingdownward through porous biomass. It is com-posed of the following sub-fronts:

G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496 2491

� Preheating of the undisturbed peat: Heat fromthe upper layer is conducted downwards pre-heating the peat up to temperatures wherewater evaporation takes place. This front doesnot emit gases in any significant quantity.� Evaporation: This endothermic reaction occurs

within a range of temperatures around 100 �C,emitting water vapour. In this front the mass lossdepends on the moisture content (e.g., 50% of thetotal weight for a moisture of 100% in dry base).� Pyrolysis: At temperatures of above 200 �C

[11] and in the absence of oxygen, pyrolysisof peat dominates the mass loss. Subsequentheating above this temperature increases thepyrolysis rate. Carbonaceous char is formedas a product. In this front the mass loss of peatranges from 5% to 10% of the dry mass [11]and emits volatile organic species (e.g., CH4,C3H8, CH3OH), polyaromatic hydrocarbons,trace levels of CO and CO2, and water vapour.� Oxidation: This front is at the top of the sam-

ple and the closest to the atmospheric air. Itinvolves the exothermic oxidation of the dryorganic content left by the pyrolysis front. Peatoxidation, as seen in thermogravimetric analy-sis at low heating rates [11], occurs at temper-atures over 350 �C but the actual temperaturerange would be lower for the higher heatingrates expected in biomass smouldering. Theoxidation reaction is the main source of COand CO2, and produces a fine layer of ashes(the mineral content). CO2 is formed wherethe oxygen supply is large (i.e., closer to thetop free surface) and CO where it is small(i.e., deeper into the sample layers). Most ofthe organic content in the peat will be lost inthis front, ranging between 50% and 70% indry base [12,11] depending on the composition.This front could overlap with the pyrolysisfront depending on the oxygen availability.

The combustion reaction in smouldering ischaracteristically incomplete. It emits partiallycombusted gases at a higher yield than flamingfires, e.g., volatile organic compounds, CO andpolyaromatic hydrocarbons. The field studies byBertschi et al. [1] of natural biomass fires in thetropical savanna show that the smouldering phasereleases on average 130% more CO and 670%more hydrocarbons, but 15% less CO2 (and noNOx) than the flaming phase.

Field studies can provide atmospheric gas con-centrations during real fires, but laboratory stud-ies are required to measure the emission factorsand capture their dependence on different burningconditions. These emission factors are essentialinput data to models of pollution transport inatmospheric chemistry simulations (e.g. [13]).

The available literature on emissions fromsmouldering peat is scarce. Muraleedharan et al.

[14] reports the emissions from the heating oftwo small samples of around 50 mg in sealedbeakers. Christian et al. [15] reports the averageemission from a single block of peat 25 cm in sizethat smouldered in open air. Both studies mea-sured that 95–96% of the total gas emissions areCO2 and CO species (when excluding watervapour). Other species present with yields above1% are CH4 and C3H8 in one study [14], andCH4 and NH3 in the other [15]. A review of theseresults is provided in Section 5 as a comparison tothe results reported in this paper.

3. Experiments

The transient evolution of CO and CO2 emis-sions during the smouldering combustion of peatsamples has been measured experimentally usingthe cone calorimeter apparatus [16]. The samplesize for the tests was 100 mm by 100 mm in crosssection and 50 mm in depth (typical dimensionsfor the cone calorimeter’s samples). The fuel sam-ple is radiated at the top exposed surface (config-uration similar to that in Fig. 1). The mass lossrate is measured and the exhaust gases are ana-lyzed for mass flow and composition.

The cone calorimeter is an apparatus com-monly used in fire engineering to study the burn-ing behaviour of condensed fuels [16]. It is usedhere to study smouldering combustion of bio-mass. It allows a shallow and horizontal smoul-dering front to propagate downwards through asample several centimeters thick. The test proce-dure is carried out in accordance of the ISO5660, which outlines the standard operating pro-cedures and practices, but without using thesparking ignition device since flaming combustionis not of interest here. The top surface of the sam-ple is exposed to a uniform radiant heat flux largeenough to ignite a smouldering reaction andestablish stable propagating-fronts. In this study,the external heat flux ranged from 30 to70 kW m�2. The mass loss of the sample is mea-sured with a balance. The released gases are chan-neled towards the exhaust tube where volumetricflow rate is measured by a Pitot tube. The COand CO2 concentrations are measured using a cal-ibrated infrared system.

Natural peat fires ignite and self-propagateonly at moisture contents below the critical value.The critical moisture for the boreal peat of theseexperiments is 125% in dry base [4]. Dependingon composition and thermal properties, the criti-cal moisture for different peat is in the range100% to 130% in dry base [9]. With the relativehigh external heat flux used in these experiments,a strong smouldering front is initiated on thetop of each sample and maintained during theexperiments, even for samples with high moisturecontents. The external heat flux enhances the

Fig. 2. Example of measured burning rate and CO andCO2 flows during a cone calorimeter test (samplemoisture content of 260% and heat flux of 50 kW m�2).Vertical lines marked the beginning and end of thesteady-state regime.

2492 G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496

spread rate and widens the burning conditions,assisting the propagation of the drying, pyrolysisand oxidation fronts. Burning of very wet peatlayers assisted with strong external heat fluxes isnot observed in natural fires but provides a novelframework to understand the combustion dynam-ics. The combination of different moisture con-tents and heat fluxes results in variations of thecontrolling mechanisms and allows the establish-ment of stable smouldering fronts in a wide rangeof burning rates and front thicknesses that cannotbe obtained otherwise in small-scale samples.Moreover, small-scale experiments of smoulder-ing propagation with no external heat flux are sig-nificantly influenced by the edge effects. The effectof the size of the samples is always a concern insmall-scale experiments because smouldering iscontrolled by heat losses (together with supplyof oxygen [3]) and thus the presence of the wallsaffects the propagation. A strong external heatflux provides uniform heating and overcomes theeffect of heat losses.

The moisture content of the peat samples isdetermined by taking a small sub-sample and cal-culating the content in dry base uw using Eq. (1).

uw ¼m0 � md

md

ð1Þ

where the mass of the sub-sample, m0, is recordedand then placed in an oven at 80 �C, then removedafter 48 h and weighed again, providing the drymass, md. All moisture values in this paper are ex-pressed as % of the dry mass, which results inmost values greater than 100%.

The boreal peat used, collected from Edin-burgh, Scotland, was old, moderately decom-posed, and of herbaceous composition. Themeasured bulk density of the dry mass was430 kg m�3 and the mineral content was 8 ± 2%in dry base [4]. After being dug from layers 0.1to 0.5 m deep, the blocks were stored indoorsand left to dry naturally during the course of thestudy. They were turned regularly to induce mois-ture homogenization. Experiments were per-formed over a period of several months. Duringthis time, the samples dried considerably givingthe desired wide range of moisture contents, from600% initially to 80% at the end. Because the sizeof the samples is fixed, the wide range of moisturecontent tested resulted in samples with differentinitial mass between 250 and 500 g and bulk den-sities between 500 and 1000 kg m�3. When cuttingsamples from the blocks, care was taken to ensurethat they were of homogenous moisture and struc-ture. The sides and bottom of the cut sample arewrapped in foil so that during the experiment onlythe top surface is exposed to ambient oxygen.

Efforts were made in the preparation protocolto keep the sample size, packing conditions andhomogeneity invariable for all tests. Howeverany peat sample has an inherent inhomogeneity

in composition, cracks and gas permeability thatinduces scatter in the measurements. The identi-fied and quantified sources of experimental uncer-tainties are: the radiant heater heat flux ±5%(calibrated); load cell ±2% (calibrated); moisturecontent ±5% (measured); sample surface areaand cracks ±10% (estimated); bulk density ±3%(estimated); and steady-state mass flow values±5–15% (measured). These errors are combinedfor each measurement, resulting in a total uncer-tainty of ±18–31% for CO2 mass flow; ± 17–30% for CO mass flow; and ±17.6% for the massburning rate.

Figure 2 shows the evolution of the burningrate and the CO and CO2 mass flows during a par-ticular experiment. This figure can be used to illus-trate a generalized time history of the processshowing three different regimes. All experimentsshow similar evolution but different time scales.The external heat flux is turned on at time zeroand very rapidly (less than 1 min) CO and CO2

are being produced at significant quantities, thusindicating that a smouldering front is ignited onthe top surface of the sample. The smoulderingfront, assisted by the radiant heat flux, propagatesdownwards through the peat. The gas mass flowsshow an initial transient response (first regime) of7 min for the CO and of 21 min for the CO2, afterwhich the front is fully developed and steady-stateconditions are reached (second regime). Duringthe pseudo steady-state regime, lasting 55 min inthe experiment in Fig. 2, approximate constantvalues are maintained with some degree of fluctu-ations. The third phase is reached when the ther-mal front approaches the end of the sample75 min after ignition and is characterized by theinfluence of the lower boundary. The smoulderinggoes through a decay of the burning rate, theCO yield decreases and the CO2 yield increases.

Fig. 3. Measured CO and CO2 mass flow rates at steady-state; (a) vs. moisture content; and (b) vs. heat flux for moisturecontent in the range 350% to 550%.

G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496 2493

The decay lasts for about 45 min for the presentedexperiment.

It is deemed that the steady-state regimereached in these experiments is more representa-tive of free propagation (not influenced by theboundaries and scale effects). Thus, the analysisof the experimental results focuses on the valuesreached during the steady-state. Note that mostvalues reported in the literature are for averageemission over the entire test, and thus take intoaccount the initial and decay transient responseswhich are influenced by the boundaries anddepend on the size of the sample.

4. Results and discussion

Changes in the burning dynamics imposed bythe different moisture contents and the externalheat fluxes have an impact on the gas emissions.The different emissions resulting from thesedynamics are investigated in these experiments.

Figure 3a shows the steady-state mass flowrates of CO and CO2 vs. moisture content for allthe heat fluxes tested. The trend has a negativegradient showing that the higher the moisturecontent, the lower the mass flow of carbon emis-sions. Figure 3b shows the flow rates vs. heat fluxfor the samples with moisture between 350% and550%.1 The trend has positive gradient with theheat flux. Based on all the experiments, the aver-age and standard deviation of all measurementsfor the emission rate of CO is _m00g;CO ¼ 0:27�0:09 g s�1 per m2 of smouldering front, and forCO2 is _m00g;CO2

¼ 0:65� 0:24 g s�1 m�2.The burning rate _m00b (i.e., mass loss rate) of

peat is an important variable needed to character-

1 Figure 3a includes all experiments but Fig. 3b showsonly the results from a selected range of moistures onlyto show the weaker dependence on heat flux.

ize the combustion regime and explain the emis-sions rate. Figure 4a shows the steady-state peatmass loss vs. moisture content for all the heatfluxes tested. The trend has a positive gradientshowing that the higher the moisture content,the higher the mass loss. This can be explainedby the larger mass of water that is lost by evapo-ration as the front progresses. Figure 4b shows themass loss vs. heat flux for the samples with mois-ture contents between 350% and 550%. The trendhas a strong correlation with positive gradientshowing that the higher the heat flux, the higherthe mass loss rate. This seems obvious since thelarger the heat flux on the sample the larger thewater evaporation, pyrolysis and oxidation rates.Based on all the experiments, the average andstandard deviation for the burning rate is_m00b ¼ 6:1� 1:4 g s�1 per m2 of smouldering front.

Peat can embed very large quantities of waterresulting in moisture contents in excess of 600%in dry base (85% of the total weight is water).When referring to the combustion yield of speciesi (ci), it is important to define it as a function ofthe dry mass containing the organic fraction andnot the total weight that includes the water. Here,the yield is expressed in mass released per drymass loss of peat (i.e., % g g�1), also called emis-sion factor in atmospheric sciences [13]. It is calcu-lated using Eq. (2).

ci ¼_m00g;i_m00

b

uwþ1

� � ð2Þ

where _m00g;i is the measurement of the species massflow rate, _m00b the peat burning rate and uw themoisture content in dry base.

Factoring in all the experiments, the averageand standard deviations for cCO is 17 ± 3.3%g g�1, and for cCO2

is 42 ± 13.4% g g�1. Figure5a shows the steady-state CO and CO2 yields vs.moisture contents for all the heat fluxes tested.The trend for cCO shows that it is independent

Fig. 4. Steady-state burning rate of peak measured as the mass loss rate; (a) vs. moisture content; and (b) vs. heat flux formoisture content in the range 350 to 550%.

Fig. 5. Measured CO and CO2 steady-state yields in dry base; (a) vs. moisture content; and (b) vs. heat flux for moisturecontents in the range 350% to 550%.

2 After correcting their values for the mineral contentand mass conservation.

2494 G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496

of the moisture content. For cCO2, the trend has a

small but positive gradient indicating that forhigher moistures, the CO2 yield is larger. Theobservation that cCO2

increases slightly with mois-ture whereas cCO remains constant indicates thatas water content increases, the burning conditionare such that more carbon content of the peat isbeing oxidized into CO2. A possible explanationfor this is presented at the end of the section. Fig-ure 5b shows the yields vs. heat flux for the sam-ples with moisture content between 350% and550%. The near zero gradient of the trendlineimplies that the yields of CO and CO2 are inde-pendent of the heat flux within a narrow rangeof moisture contents. This observation is also sup-ported by the data in Fig. 5a, where samples withsimilar moisture contents but different heat fluxesare adjacent to each other. Summing the twoyields, the average and standard deviation forthe yield of total CO and CO2 emissions is59 ± 15.8% g g�1.

These values can be compared with the onlydata available in the literature. Muraleedharanet al. [14] reported cCO between 3% and 7.5%

g g�1, and cCO2between 30–36% g g�1, providing

a total CO and CO2 emission in the range 33–43%g g�1. The experiments of Muraleedharan et al.did not reproduce smouldering conditions butstudy general thermooxidative emissions fromheated peat in sealed beakers. This explains theconsistently lower yields reported in [14]. Chris-tian et al.2 [15] reported a cCO of 9.5% g g�1,and a cCO2

of 77% g g�1 and a total CO andCO2 of 87% g g�1.

Figure 6a shows the CO to CO2 ratio withvarying moisture contents. The ratio is between0.3 and 0.7, with average 0.43 and standard devi-ation 0.12. The trend is a decrease of the ratiowith higher moisture content. Figure 6b showsthe CO to CO2 ratio dependence with heat flux,the trend line has a near zero gradient showingthat there is weak dependence. Comparing withthe values for tropical peat, the ratio is reportedbetween 0.1 and 0.2 [14], or 0.12 [15].

Fig. 6. Measured CO/CO2 ratio; (a) vs. moisture content; and (b) vs. heat flux for moisture contents in the range 350%to 550 %.

G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496 2495

The trend of the CO/CO2 ratio, together withthe increasing CO2 yield with moisture contentobserved in Fig. 5, could be explained in termsof the smouldering dynamics. Biomass with lowermoisture contents results in higher rates of heattransfer from the burning front to deeper layersof the soil. But oxygen diffusion depth from thefree surface into the soil is independent of mois-ture. Therefore, as the moisture content decreases,the thickness of the thermal front increases but theoxidation front thickness which is controlled byoxygen diffusion stays constant. Thus, lower mois-tures content could lead to wider pyrolysis fronts,allowing a larger fraction of the organic content atdeeper layers to be released without being oxi-dized. This mechanism explains the observedresults and would imply that deeper smoulderingfronts produce more gas species typical of pyroly-sis decomposition, e.g., volatile organic com-pounds and polyaromatic hydrocarbons.

5. Conclusion and final remarks

Measurements of the carbon emissions fromshallow and strong smouldering fronts in borealpeat have been conducted. Flow rates from thetwo most important carbon gases, CO and CO2,have been measured under a wide range of smoul-dering burning rates and front thicknesses result-ing from different moisture contents and externalheat fluxes. Burning at high moisture contentswith strong external heat fluxes is not observedin natural fires but provides a novel frameworkto understand the smouldering dynamics by vary-ing the controlling mechanisms.

Smouldering peat was found to release largeflows of carbon gases, with average mass flowrates per m2 of smouldering front of 0.27 ± 0.09for CO and 0.65 ± 0.24 g s�1 for CO2. This CO2

mass flux is equivalent to approximately 3000times the normal flux from peatlands due to natu-

ral decomposition at ambient conditions(0.31 g h�1 m�2 for sphagnum peat [17]). On aver-age, the measured CO yield in dry base is17 ± 3.3% g g�1, and the CO2 yield 42 ± 13.4%g g�1. The total CO and CO2 carbon yield is of59 ± 15.8% g g�1, and the CO to CO2 ratio wasmeasured in the range between 0.3 and 0.6, withaverage and standard deviation of 0.43 ± 0.12.

The experiments have been designed to mini-mize the scale effect by the use of the external heatflux and reporting measurements in the steady-state regime. The measurements under a widerange of burning conditions can be extrapolatedto shallow natural fires. Extrapolation of theburning rates to the condition of zero externalheat-flux (unaided smouldering) gives a peat massloss rate of 4.5 g s�1 per m2 of smouldering front.Extrapolation of emissions to the range of mois-tures were natural peat fire take place (50% to130% in dry base) give flow rates of 0.4 g s�1 m�2

of CO, 0.9 g s�1 m�2 of CO2; CO yield of 14%g g1, CO2 yield of 32% g g�1, and a CO/CO2 ratioof 0.45. This is the first time that emission massflows per unit area of smouldering front arereported.

The observed trends of the mass flow rates areexplained well by the burning rate and the dynam-ics of the smouldering front. Whereas the externalheat flux does not affect the emission yields, themoisture content affects the yield and the structureof the front. The results indicate that peat at lowmoisture smoulders with wider pyrolysis front,releasing a larger fraction of non-fully oxidisedvolatile carbon-containing species. This allowsconcluding that narrow burning fronts favorlower CO/CO2 ratios and wider fronts favor lar-ger ratios.

Based on these results, it is expected that indeep smouldering fires, where the lower oxygenavailability and heat losses result in wider thermalfronts, the ratio CO/CO2 will be larger than forshallow fires burning close to the free surface.

2496 G. Rein et al. / Proceedings of the Combustion Institute 32 (2009) 2489–2496

However, the findings of this study pertain toshallow burning fronts, and behaviour at otherconditions and configurations typical of smoul-dering fires, especially fire fronts propagatingmany meters deep into the subsurface, need tobe investigated further.

Acknowledgments

This research has been partially funded by theMet Office and Natural England and partlyfunded internally. The support and resources allo-cated by Jose Torero have been essential to con-duct this study. The authors thank the closecollaboration of David Fleming, Matt Daviesand Colin Legg. The help of Clare Ashton, Juande Dios Rivera, Alan Gray, Pedro Reszka, RoryHadden, Jose Garcia, and Ricky Carvel isappreciated.

References

[1] I. Bertschi, R.J. Yokelson, D.E. Ward, et al.,Geophys. Res. 108 (2003) 8472.

[2] J.B. Kauffman, D.L. Cummings, D.E. Ward, Oec-ologia 113 (1998) 415–427.

[3] T.J. Ohlemiller, Prog. Energy Combust. Sci. 11(1985) 277–310.

[4] G. Rein, N. Cleaver, C. Ashton, P. Pironi, J.L.Torero, Catena 74 (2008) 304–309.

[5] S.E. Page, F. Siegert, J.O. Rieley, H.D.V. Boehm,A. Jaya, S. Limin, Nature 420 (2002) 61–65.

[6] G.B. Stracher, T.P. Taylor, Int. J. Coal Geol. 59(2004) 7–17.

[7] K.N. Palmer, Combust. Flame 1 (1957) 129–154.[8] H. Joosten, D. Clark, Wise Use of Mires and

Peatlands. Background and Principles including aFramework for decision making, International MireConservation Group and International Peat Soci-ety. ISBN 951-97744-8-3, 2002.

[9] W.H. Frandsen, Can. J. Forest Res. 27 (1997) 1471–1477.

[10] H. Svensen, D.K. Dysthe, E.H. Bandlien, S. Sacko,H. Coulibaly, S. Planke, Geology 31 (2003) 581–584.

[11] A. Usup, Y. Hashimoto, H. Takahashi, H. Haya-saka, Tropics 14 (2004).

[12] S. Shimada, H. Takahashi, A. Haraguchi, M.Kaneko, Biogeochemistry 53 (2001) 249–267.

[13] A. Heil, B. Langmann, E. Aldrian, Mitig Adapt.Strat. Glob. Change 12 (2007) 113–133.

[14] T.R. Muraleedharan, M. Radojevic, A. Waugh, A.Caruana, Atmos. Environ. 34 (2000) 3033–3035.

[15] T.J. Christian, B. Kleiss, R.J. Yokelson, et al., J.Geophys. Res. 108 (2003) D23, 4719.

[16] A. Tewarson, Generation of Heat and ChemicalCompounds in Fires, The SFPE Handbook of FireProtection Engineering, 3rd ed., Society of FireProtection Engineers, pp. 3–82, 3–160.

[17] M.H. Nykanen, J. Alm, J. Silvola, K. Tolonen, P.J.Martikaine, Global Biogeochem. Cycles 12 (1) (1998)53–69.