Wildfire Combustion Chemistry & Smoke
Wildfire Combustion Chemistry & Smoke
Wildfire Combustion Chemistry & Smoke
& Smoke
TABLE OF CONTENTS
1. INTRODUCTION _________________________________________________________1
REFERENCES:_____________________________________________________________38
1
1. INTRODUCTION
Forest fires are rather impressive by the power of released elements and by the other hand
the inadequacy of the mankind in coping with it. Knowledge of fires, which is gradually
accumulated over a very long period has until recently been purely empirical. The main
reasons for that fact are: complexity of the process (the geometry of the objects included, its
chemical structure, the dynamics of the ambient atmosphere, etc.). Only until recently (recent
twenty year form the previous century) scientific methods of fire analysis have been
developed. In recent years several research centres made considerable advances in this area of
human knowledge - both experimental and experimental [1].
The free fires are wide known around the world, but less investigated due to:
• uncontrollable environment in which they occur ;
• lack of information for the conditions at which this phenomena happens –atmosphere
conditions ;
• unknown properties (in some cases roughly known) of the fuel ;
• scarce detailed information for the properties of the combustion process ;
Forest fires, especially large-scale ones result in great amount of money in financial losses
annually around the Globe [2]. These fires can locally disrupt complete ecosystems and even
produce changes in the meteorological processes. Thus if there are computer models developed
to simulate forest fire spread under variety of topographical and vegetation conditions and
much more, they should be applicable for example for making fire-fighting strategies [3].
Forest-fires are extremely complicated natural phenomena – it is a collection of many different
phenomena, each is itself quite complex. The fire is example of combustion process –
unsteady, no homogeneous environment with mass, heat and momentum transport and the
details could not be depicted. Thus there are needed great efforts in developing accurate
models for the chemical kinetics and related burning rates, heat release rates and flow
temperature field.
The free fire processes as well as forest fires are widely spread around the World. Some of
them are threats, endangering the environment and the human habitat [2, 4]. Forest fires are
known to change the landscape of wide regions of the globe, especially Australia and Central
Latin America. These fires are presumably controllable and the expected results are easily
achieved. The actual results are far from expectation of the actuators – great amounts of
pollutants are released in the atmosphere, the vegetation as well as animal environments are
threatened from extinction. Some of the released combustion gases (NOx) are polluting the
Earths atmosphere by causing depletion of the protective ozone layer, thus increasing the risk
from dermatological cancer of the Earth’s animal habitants and influencing the vegetation. The
free fire burning process is accompanied with release of number of pollutant gases, like CO,
NOx, hydrocarbons, SOx as well as solid matter – primarily soot [forest fires, principles of fire
behaviour].
With the continuing increase in the computer power as well as Internet technology it is
becoming possible to make numerical modelling applications in forest-fires prediction, rather
then phenomenological models developed so far. The detailed numerical models reveal
practical workability in great variety of applications, such as air flow modelling, flow around
buildings, automotive applications, and combustion modelling in its wide range of engineering
tasks. Parallel supercomputers are being used for solving such problems as their productivity
approaches Tflops (1012 flops – floating point operations per second). [5, 6, 7].
2
Nowadays even the personal computers (as their power approaches Gflops and the
available allocated memory is in the range of gigabytes) are used for implementation of
sophisticated and realistic models [5, 8], such as forest fires.
The models of discussion include equations of fluid motion, heat and mass transfer as well
as chemical reactions, occurred in a fire process. However the complexity of the entire process
does not give opportunity to go in deep details for each item, so reasonable approximations are
needed in order to achieve useful processing speeds [9, 10, 11]. Thus the fire flow field
modeling implementation is based on well known Navier-Stokes equations with additional
models in order to represent turbulence phenomena, species chemical reactions and heat
(conduction, convection, radiation) transfer involved, depicted in figure 1.
COMPLEX
BOUNDARY MATHEMATICAL GEOMETRY
CONDITIONS MODEL
The process of the propagation of forest fires is extremely complex and difficult to depict.
Wide ranges of unknown variables are used to describe forest fires: fuel moisture content, type
of the biomass and its thermo-chemical properties, wind flow profiles over complex terrains,
atmospheric conditions, etc [12]. Various factors affect the fire spread rate, the most important
is the local wind velocity and is the most difficult to forecast. The other factors can be
evaluated or at least their values vary over a limited range depending on the seasons of the
year, the geographic position of the place and the local topography [13].
The fire itself is complex process of fuel oxidation, gas products, soot and ash and heat
energy is released as well. Thus the process includes chemistry, heat, mass and momentum
transfer and combustion process. That’s why the depiction of process needs knowledge of all
these phenomena in order to try to understand the “jungle” of free fires as well as forest ones.
Actually the description of the process relies both of the physics of the process as well as
accumulated data for real observed fires in order to build a model that tends to depict this
extremely complex phenomena.
Forest fires are difficult to describe, because many aspects are hard to depict [3], such as:
• forest fuel material is no homogeneous and irregularly distributed;
• chemical reactions are complex and radiation properties of the combustion process are
basically unknown;
• turbulence of the flow field influences on the entire process;
• the combustion process is frequently two-phase flow, i.e. volatiles oxidation and
particle’s char combustion.
The fire process involves hundreds of different materials, but the most important is wood,
i.e. biomass. Consequently knowledge of the combusting material is required in order to depict
such a complex process. The work presented does not appeal to resolve all the items pointed,
but to reveal an approach for investigation of such a complex process with the problem
described in details in the following sections.
3
Biomass is compared with other fuels in the so-called coalification diagram [14]. In the
following figure the approximate boundaries between different classes of fuels are shown. The
diagram can be used in order to infer the chemical structure and some inorganic fuel aspects.
Figure 2. Coalification diagram showing compositional differences among coals and biomass.
The release of atomically dispersed inorganic material from fuel particle is influenced both
by its inherent volatility and the reactions of the organic portions of the fuel. Materials that are
inherently volatile at combustion temperatures include derivatives of some alkali and alkaline
earth metals, most notably potassium and sodium. Other, non-volatile material can be released
by convective transport during rapid pyrolysis. Alkali material plays a central role in both
pyrolysis and char oxidation processes. Potassium is the dominant source of alkali in most
biomass fuels. Although the total amount of potassium in wood is typically much lower than in
straws, the fraction in the ash may be higher [14]. Calcium reacts with sulfur to form sulfates
in ways somewhat analogous to potassium, but the mobility of calcium and the properties of
the deposits it forms are both more favourable than the ashes formed from straw and grasses.
Chlorine is a major factor in ash formation. Chlorine facilitates the mobility of many
inorganic compounds, in particular potassium. Potassium chloride is among the most stable
high-temperature, gas-phase, alkali-containing species. Chlorine concentration often dictates
the amount of alkali vaporized during combustion as strongly as does the alkali concentration.
In most cases, the chlorine appears to play a shuttle, role, facilitating the transport of alkali
from the fuel to surfaces, where the alkali often forms sulphates. In the absence of chlorine,
alkali hydroxides are the major stable gas-phase species in moist, oxidizing environments, i.e.
combustion gases [14]. The composition of biomass are complex, involving six major
elements (C, H, N, S, Cl, O) in the organic phase and at least 10 other elements (Si, Al, Ti, Fe,
Ca, Mg, Na, K, S, P), not including heavy metals, in the inorganic phase important to ash
characterization [14].
The inorganic content of the fuel defines the ash behaviour and especially ash melting
temperature [14]. Principally the detailed chemistry of ash formation is not fully developed,
the removal of alkali and other elements is known to increase the fusion temperature of the
ash. For example the experiments in leaching of alkali metals plus chlorine by simple water
washing reveals dramatic improvements in fusion temperature for straw ash – the result is
about 80% of alkali and about 90% of the chlorine reduction and thus reduction of the acid gas
emissions as well as chlorine facilitation in ash deposition. Principally biomass washing
increases the melting temperature of the ash. The potential of decreasing the fusion
temperature also exists. There are other impacts of leaching on the combustion behaviour.
5
The amount of fuel lost during the pyrolysis stage of combustion increases with increasing
hydrogen to carbon ration and, to lesser extent, with increasing oxygen to carbon ration.
Typically, the volatiles loss during early pyrolysis of biomass is about 75%. Biomass material
consists of polymeric organic compounds, generally described by chemical formula CxHyOz.
The actual form of the chemical formula depends on many factors: type of the plant, material
conditions – dry or wet.
H C O H
H C O
H
C
O H H C
O
C C H
H O H
Figure 3. Cellulose chemical structure, a polymer built from the glucosan monomer [1].
R3
R1 Dehydrocellulose Char + H2 + H2O + CO
+ H2O + CO2 + NH3 + ....
Cellulose
R2 Tar
The resulting dehydrocellulose is further decomposed to form char and gaseous products in
an exothermic reaction with a rate R3. The tar produced is mostly composed from levoglucosan
[1] – figure 5:
6
H2C O
H
C O
H
C
OH H C
HO
C C H
H OH
The released levoglucosan is volatile and in the temperature range in question it vaporizes
to form light gases that support gas-phase flame. The gases evolved in the dehydration path are
partly non-combustible and are produced in small quantities. Therefore this reaction can
support only the surface oxidation or “glowing” combustion of char. The yield of levoglucosan
in the second reaction indicates that tar evolution results from a radical “unzipping” process of
the basic polymer. Physically, the process greatly resembles the coal devolatilization process
[18] (well, coal is millions of year’s old transformed biomass). The devolatilization rate and tar
production are represented by an Arrhenius type equation:
⎛ E1 ⎞
⎜− ⎟
R1 = A1 exp ⎝ RT ⎠
- devolatilization rate , [1/s] (1)
and respectively
⎛ E2 ⎞
⎜− ⎟
R2 = A2 exp ⎝ RT ⎠
tar release rate , [1/s] (2)
where :
A1, A2 – pre-exponential factors for the corresponding reactions , [1/s] ;
E1, E2 – activation energy for the corresponding reactions , [J/(mol.K)] ;
These parameters (A1, A2 , E1, E2) are strongly fuel dependent as well as the conditions at
which the reactions are realized, primarily heating rates. There are quite reliable data for a
wide range of fuels of interest, but mostly at conditions far from the wild land fire. Thus these
data are appropriate to be utilized with having in mind that additional investigation for the
influence of these parameters on the model developed.
The other two major components of biomass – hemicelluloses and lignin have less regular
structures that cellulose and consequently will show more complex behavior at the heating
process. Therefore the overall kinetics of biomass pyrolysis process varies from one species to
another but, typically is similar to cellulose.
The demonstration of that fact is that thermal analysis of different biomass show common
general characteristics – the observed endothermic region at around 100oC where hygroscopic
water is evaporated (the drying process), next as the material is heated up at temperature
region of about 200 – 280oC is identified as shallow endothermic (and intensive mass loss due
to volatiles release), plateau or slightly exothermic, next the deep endothermic region at about
280-340oC (the ignition and combustion of the volatiles) which changes the strong exothermic
behavior at about 350oC [15, 19].
In the burning region, the rate of mass loss proceeds so rapidly that it reaches to its
maximal value. The rapid mass-loss is immediately slowed down at temperatures 350 – 450oC.
After then, burning rate apparently decreases and consequently some small losses in the mass
7
of the fuel sample continuously goes on as long as particle temperature is increased to about
1000oC – this is the region of slow carbonized residue oxidation. As char combustion process
is finished no mass loss is observed and the solid residue consists of mineral matter - ash. The
most important characteristic temperatures of a burning profile are ignition temperature and
peak temperature [15]. The ignition temperature is fuel temperature at which the burning
profile underwent a sudden rise due to volatiles oxidation. The ignition temperatures of
samples are determined from the burning profiles, obtained in the laboratory experiments. The
burning profile peak temperature is usually taken as a measurement of the reactivity of the
sample.
The fuel of concern is biomass, especially forest residues. Biomass is considered to be
treated as low calorific fuel and only recently its properties are under investigation, primarily
because its CO2 pollution neutral behavior. Furthermore the Kyoto protocol [20] states that up
to year 2015 at least 15% of the human kind consumed energy should be replaced with
biomass matter, thus reducing the intensity of fossil fuel consumption as well as oil and even
nuclear fuel (unfortunately there are countries that still have not signed it). That’s why many
laboratories are occupied in achieving properties of the biomass matter. The properties of
concern are the proximate and ultimate analyses, principally most of the biomass matter of
interest are already investigated. Nevertheless some specific (rare) biomass fuels need to be
investigated.
However the achieved laboratory results for the fuel contents, i.e. proximate and ultimate
analyses are not enough for characterization of the fuel. Further needed data are drying,
devolatilization and char oxidation rates. These data are mostly achieved by TGA (thermo-
gravitometric analysis) at specified laboratory conditions. Unfortunately the laboratory rates
data are principally quantitative, because of the methodology, primarily because of the low
heating rates at which experiment is done. The nature rates of drying, devolatilization and char
combustion are far higher that the laboratory achieved. Nevertheless the values are post-
adjusted in order to become applicable for modeling.
Dead plant matter accumulated on the forest floor is rich in nutrients and is transformed by
microorganisms into humus in which individual plants are no longer identifiable. This process
results in increase of the sulfur and nitrogen concentrations due to microorganisms, which
utilize these elements in their growth and reproduction [15]. In addition, tree leaves and
needles generally have higher composition of N and S than stems and limbs. Nitrogen is one of
the most dominant of the macronutrients and thus is of primary interest, because of the large
number of nitrogen-based compounds produced at biomass combustion.
Compared to the fossil fuels there is a relative higher content of oxygen in the biomass,
which predetermines specific approaches for describing the chemistry of the combustion
process.
zone where the oxygen and fuel are mixed at stochiometric level and intensive chemical
reactions are proceeded which result in the visible emission of light called flame [12]. The heat
energy from the chemical reactions also causes feeds back energy source to the interior of the
flame envelope and ahead of the flame envelope. This causes further devolatilization of fuels
with low vapor pressures and pyrolysis of solid fuels.
Initial heating of unburned fuel releases the ore volatile components by distillation, which
then leads to pyrolysis and fragmentation of polymers and the release of oxygenated organic
compounds. Flaming is initiated when there is a source of ignition and the fuel-to-oxygen
mixture reaches flammable proportions. Flaming and smoldering combustion are reasonable
distinct regimes of the combustion processes that involve different chemical reactions as well
as appearance.
Experimental investigation [21] of the reactivities of carbonaceous materials of biomass
primarily consists of isothermal and non-isothermal thermo gravimetric techniques. In the
following table the proximate analysis for some biomass samples are given:
A plot of rate of weight loss against temperature while burning a sample under an oxidizing
atmosphere is referred to as the “burning profile”. The burning profiles of the biomass samples
are shown in the following figures. The first peak observed on the burning profiles of the
biomass samples corresponds to their moisture release. After releasing the moisture, some
small losses in the mass of the sample occurred due to desorption of the adsorbed gases. A
sudden loss in the mass of the samples started at the temperatures between 450 and 500 K,
representing the release of some volatiles and their ignition. In the rapid burning region, the
rate of mass loss proceeded so rapidly that it reached its maximum value. The rapid loss of
mass immediately slowed at temperatures between 600 and 700 K. After this point, the
burning rate apparently decreased, and consequently, some small losses in the mass of the
sample continued as long as the temperature was increased.
9
Figure 8 Burning profile of pine cone Figure 9 Burning profile of sunflower shell
Figure 10 Burning profile of cotton refuse Figure 11 Burning profile of colza seed
10
The observation of the applied figures show that although the proximate analysis differ
considerably (Table 1), the ignition temperatures of the biomass samples changed in narrow
interval (Table 2). It has been observed that an increase in the volatile matter contents of the
biomass samples cause, as general tendency, an increase in the peak temperature. The rate of
weight loss at the burning profile peak temperature is called the “maximum combustion rate”.
Different biomass samples have close values of the maximum combustion rates, the
differences can be attributed to the differences in their chemical and physical properties.
Table 2. Combustion sample properties.
Sample Ignition temperature Maximum combustion Peak temperature ,
[K] rate [mg/min] [K]
Sunflower shell 475 5.50 573
Colza seed 423 2.80 535
Pine cone 475 5.20 565
Cotton refuse 423 3.70 598
Olive refuse 473 3.40 537
The presented data for different biomass mater reveals that the burning profiles have
qualitative common behaviour and the flaming temperatures are in close range. It can be
concluded that common approach could be applied in order to describe the biomass
combustion behaviour, considering the ultimate analysis data for the specific biomass material.
IIb. Abundant evaporation of oleoresins occurs with partial oxidation due to the intense
thermal environment and extreme chemical bond rupturing resulting from the high-intensity
radiant energy income;
III. Penetration of evaporated oleoresins into a high-temperature oxygen-deficient
environment with considerable oxidation occurring as molecules diffuse through the flame
envelope. Those not undergoing complete oxidation may be fragmented into ethylene units
and/or free radicals;
IV. Very high-temperature environment with oxygen depletion occurring as the carbon
fragments reach the tip of the flame envelope. The amount of compound oxidation is
dependent on the depth of the flame envelope and the amount of ventilation (turbulence). As
the flames become taller, the heat feedback to the solid fuel becomes less, and the radiant
energy loss becomes greater. The result is that particle formation/polymerization reactions may
be increased because of the loss of heat energy within the fuel-rich zone;
V. Recombination takes place with the formation of compounds not found during the
evaporation phase or inside the flame envelope due to pyrolysis. Aromatic hydrocarbon
molecules are synthesized during this phase of transport;
VI. Products of pyrolysis and glowing combustion are transported across the flame surface;
VII. Transport of products of pyrolysis and glowing combustion completely miss the flame
envelope and enter with no additional oxidation;
The content of particulate matter varies between flaming and smoldering combustion.
Particle’s mass may consist of trace elements of potassium, chlorine, sulphur, phosphorus and
sodium – principally between 1 and 10%. On a percentage basis, the mass of trace elements
contained in particles from smoldering combustion is 10-20% of that from the flaming phase.
Emissions of graphitic and organic carbon are especially important because of the
increased absorption of light energy by smoke particles that are high in graphitic carbon
content. Absorption of light is due to the black carbon content of the particles. Though
graphitic or black carbon is produced proportional to the intensity of fire, it is generally true
that emission factors for PM2.5 are inversely proportional to higher intensity burns. The
organic fraction of the particles smaller than 2.5µm is as much as 50-70% of the mass of
particulate matter for the smoldering phase, but can be lower for the flaming phase emissions.
Further studies of the released smoke release that “young” smoke (less than 2 hours old) in
comparison to “aged” smoke (from 2 to 4 days) from biomass fires in the very humid Brazilian
Amazon region, the hydrocarbons of fewer than 11 carbon atoms were depleted over time and
converted to CO2, CO, and reactive molecular species and likely removed through dry
deposition and/or by conversion to particulate matter. Although somewhat contradictory, the
argument concludes that individual particle mass increased over time, and it was estimated that
20-45% of the mass concentration of the particles was due to the condensation of organic
compounds. The investigations show that most of the active “new” fires are lit between 1:00
and 3:00 pm local time and active flaming is nearly complete in 3 - 4 hours with smoldering
combustion is the dominating mode and continues for several days. The organic content of
particles produced during smoldering combustion of biomass is approximately 20% higher that
the organic content of particles produced during the flaming combustion phase. There are still
many issues to be resolved before it can be conclusively demonstrated that condensation of
hydrocarbons occurs at higher rates than volatilization in contributing to the mass increase of
particulate matter once the smoke is more than a few minutes old.
Emissions of NOx and SOx arise predominately from N and S in the fuel. In most cases of
industrial biomass combustion utilization the operating temperatures are low enough that
thermal NOx contributes only to small fraction of the total. NOx in combination with HC
photochemically leads to the formation of ozone, which is a lung and eye irritant and major
problem in urban environments. Ozone is also damaging to plants. SOx are respiratory
irritants, and their effects are enhanced in the presence of PM due to transport deep within the
lung. Both NOx and SOx participate in the reactions leading to acid rain. Uncontrolled NOx
emissions also depend partly on stoichiometry. For HC fuels, NO formation from fuel N
occurs on time scales comparable to the HC oxidation, and is known to be sensitive to
equivalence ratio, with fuel lean conditions producing high yields and fuel rich conditions
producing low yields. Under fuel rich conditions, the relatively fast conversion of fuel C to CO
competes for oxygen, leading to a reduced availability of O2 for NOx production. The
fractional conversion of fuel N to NOx has been shown to decrease with increasing fuel N
concentration for HC fuels, coals and biomass. The declining N to NO conversion is postulated
to be due to the formation of a N containing species important to both the production and
destruction of NOx.
The following table presents the pollutant emission factors (% dry fuel) for open field
burning, commercial biomass-fuelled FBC, and experiments using an entrained flow multi-fuel
combustor (MFC) laboratory combustor without pollution control [14].
Table 3 The pollutant emission factors (% dry fuel) for open field burning biomass.
reveal that NH4 is observed as well. All these data show that at biomass volatilization and
combustion the yield of N2 and NOx can be significant, measured per total burned area.
2.3.2.3.5. Methane
Methane and nonmethane hydrocarbons – methane is produced in much larger quantities
during the smoldering combustion phase than in the flaming phase. Emission factors are about
two to three times greater for the smoldering phase than for the flaming phase. For example, at
an experiment they ranged from 5.7 to 19.4 g/kg of fuel consumed for smoldering phase and
the flaming phase emission factor ranged from 1 to 4.2 g/kg. The volatile hydrocarbons are
well correlated with methane. Experimental data shows linear relationship between CH4
emission factors and modified combustion efficiency for different fires in diversity of regions
16
(savanna and forest fires for example). In addition to the different slopes of linear regression
models, typical modified combustion efficiency values were found to be 0.84 for slash fires in
primary forests of Brazil, 0.90 for second-growth deforestation incineration in the same area,
and 0.94 for cerrado savanna ecosystems.
2.3.3.1. Drying
The drying is energy consuming (exothermic) process and it is considered to be finished
when the moisture contents of the fuel are evaporated. The moisture contents, achieved by
proximate analysis are definitive for the description of the entire process. The water content is
actually the mass of water, which mechanically conserved in the fuel sample and strictly
depends on the environment conditions. At this stage the temperature of the fuel sample rises
as heat energy is supplied up to the saturation temperature of water at the ambient pressure.
The heat-up of the fuel is evaluated as the rate of the change of its temperature, i.e. [K/s]. At
low drying rates (usually 10 – 103 K/s) the water contents are diffusing to the fuel surface and
evaporate, as well as released through the pores of the sample. It worth saying that the
structure (ash lattice) of the fuel sample is definitive for this process. At these levels of heat
exchange the temperature of the sample could be considered constant and homogeneous. At
high heating rates (up to 105 – 106 K/s) the moisture contents are released rather intensive and
even could evaporate inside the pores of the fuel particle (no homogeneous temperature
distribution in the fuel particle volume is observed), thus increasing internal pressure and even
burst the solid matter. However this rarely happens, so one could assume more convenient
(especially for modeling purposes) process, which occurs at low heating rates and resulting
homogeneous temperature distribution. At drying regime the temperature of the fuel sample is
relatively low (up to 100oC at atmospheric pressure) and no intensive thermo-destruction of the
hydrocarbons proceeded. The energy, consumed for water evaporation in case of forest fires is
energy (predominantly supplied by radiation) released during combustion of the biomass fuel
17
and rules as extinction factor for the burning process. This frequently occurs in case of
combustion of raw (and thus wet) biomass.
2.3.3.2. Devolatilization
When the drying process has finished the biomass fuel consists only of combustible matter
and ash (mineral matter). The combustible matter consists of volatiles, tar and solid carbon
residue. The volatile gases are products of thermal degradation of the combustible matter. Due
to external heat supplied to the fuel the chemical chains of the complex hydrocarbons are
broken and light organic matter – volatiles, is produced as well as tar. There is deep
commitment between the heat-up rate of the fuel sample and yield of volatiles – as the heating
rate is increased the released volatiles quantity is higher and respectively the tar yield is
relatively lower. The thermal degradation is basically exothermic process and three staged of
this process could be observed [12]. The energy, consumed for light gases devolatilization is
called latent energy of devolatilization. Some authors (10, 11) ignore this energy, when come
to build model for the devolatilization process, but principally this heat quantity is “hidden”
.into the latent heat for water content evaporation.
⎛ Evolatiles ⎞
⎜− ⎟
Rvolatiles = Avolatiles exp ⎝ RT ⎠
- devolatilization rate , [1/s] (3)
18
∂ ∂ ∂ ∂ ∂Y ∂ ∂Y ∂ ∂Y ~ (4)
( ρu Φ ) + ( ρvΦ ) + ( ρwΦ ) − ( ΓΦ ) − ( ΓΦ ) − ( ΓΦ ) = RΦ
∂x ∂y ∂z ∂x ∂x ∂y ∂y ∂z ∂z
where :
r r r r
u,v,w - velocity components of the flow velocity V = ui + vj + wk ;
x, y, z – cartezian coordinates;
ρ - density;
Φ – common variable (i.e. u, v, w, T, Yi) ;
ΓΦ – diffusion coefficient for Φ; YΦ - can be the mass fractions of fuel, oxidizer, products;
19
~ ~
RΦ - mean reaction rate (in case Φ is YΦ , then RΦ is mean chemical reaction rate.
The determination of the chemical reaction rate is the basic problem in combustion
processes modelling. The difficulties arise from the non-linear character of RΦ . The reaction
rate is a function of the temperature, pressure and mass fraction of the chemical substances, but
the mean value of RΦ is not function only of its averaged values. Therefore, in order to
determine the reaction rate RΦ , a model describing the chemical reactions should be
introduced. There is a large variety of models developed for turbulent combustion modelling.
There are more then 100 models known and published in the literature. Their description is a
huge task, but the models most widely used for practical predictions can be grouped in several
approaches and classified according their development, application and improvement. Some of
these models will only be mentioned in the following section and the emphasis will be on
applied combustion model – eddy dissipation model (EDM).
These assumptions lead to the following mathematical expression for the reaction rate
REBU:
ε
REBU = C EBU ρ Y ' 2fuel , [kg/(m3s)] (6)
κ
where:
CEBU - model constant,
νFuel - stoichiometric coefficient (amount of oxygen ,required to oxidize 1kg of fuel at
stoichiometric condition );
κ - turbulent kinetic energy , [m2/s2]
ε - dissipation rate of the turbulent kinetic energy , [m2/s3] ;
ρ – gas density, [kg/m3].
This equation expresses ideally the fact that the local chemical reaction rate depends on the
gas mixture fluctuating components Y`. However, in practice this leads to a number of
complications related to defining almost unpredictable variables such as fluctuations. The
application of the EBU model is also mostly limited to modelling only pre-mixed combustion
[23].
3.1.3.2. Eddy Dissipation Combustion Model (EDM)
This model is proposed by Magnussen [Magnussen,Hejertager]. The model develops the
conception of eddy dissipation and overcomes successfully some problems related to EBU. In
this model the chemical reaction rate is presented by the mean mass fraction values of the
reacting species instead of their fluctuations – actually this is the main advantage of EDM
compared to EBU. The main factor influencing the rate of infinitely fast chemical reactions is
the diffusion of species. The chemical reaction itself takes place when the reagents have mixed
up to the molecular level at high temperature. The chemical reaction rate is determined by the
mixing rate, i.e. by the rate of dissipation of eddies. The fuel and oxidant appear as fluctuating
intermittent quantities so that a correlation is possible to exist between the fluctuations of the
chemical components and their mean values ( Y ≈ Y ' 2 ). Thus the reaction rate REDM can be
expressed by the eddy decay rate and average mass fraction of the reacting elements:
~ ~
ε ⎛~ Y Y ⎞
R EDM = A EDM ρ min ⎜ Y fuel , oxidizer , B products ⎟ , [kg/(m3s)] (7)
κ ⎜ ν fu 1 + ν fuel ⎟
⎝ ⎠
where :
AEDM = 4.0 - model constants;
BEMD = 0.5 - model constants;
~
Y fuel - mean value of the mass fraction of fuel compound, [kg/kg];
~
Yoxidizer - mean value of the mass fraction of oxidizer, [kg/kg];
~
Yproducts - mean value of the mass fraction of products, [kg/kg].
Because of its simplicity and testified applicability it is one of the most popular concepts
for modelling local reaction rates in practical turbulent combustion systems. This model has
21
the ability to handle premixed, partially premixed and non-premixed combustion with
sufficient accuracy.
This model will be used in the development of the complex model for free fire behaviour of
biomass.
where :
ν i' - stoichiometric coefficient of reactant i ;
ν i" - stoichiometric coefficient of product i ;
N – number of reacting species ;
The rate of production (consumption) of a specie i is defined as a sum of the forward, (qi)
and backward (pi) reactions, where i takes place:
R
R chem.reaction i = ω i ρ = ρ ∑ rij (q j − p j ) , [kg/(m3s)] (9)
j =1
where :
R, number of elementary chemical reactions ;
rij'
N ⎛ Yj ρ ⎞
q j = k jf ∏ ⎜ ⎟ - rate of the forward reaction (10)
⎜
i =1 ⎝ W j ⎠
⎟
rij''
N ⎛ Yj ρ ⎞
p j = k jb ∏ ⎜ ⎟ - rate of backward reaction (11)
⎜
i =1 ⎝ W j ⎠
⎟
β ⎛ Ej ⎞
where k jf = A jT j exp⎜⎜ − ⎟⎟ (12)
⎝ RT ⎠
kin
RCO
CO +1/2O2 → CO2 (14b)
where the global rates are presented as Arrhenius low as follows :
−T / T ~ − a ~1+ a
kin
RCH 4
= ACH 4 e 1 g ρ YCH Y
4 O2
(15a)
−T2 / Tg ~ ~ ~
kin
RCO = ACO e ρ (1+3b )YCOYOb2YH22bO (15b)
where:
ACH 4 =1.15x109 , [s-1] – pre-exponential factor for the CH4 chemical reaction ;
Actually the chemical kinetics plays a main role in the process of generation and
consumption of the chemical species involved in the combustion process. Therefore, including
the kinetics into the calculations and accounting the interaction between the turbulence and the
kinetics is appropriate way to represent more adequately the combustion phenomena. The
validity of the predictions can be improved significantly if the turbulent combustion model
includes the complex physical and chemical phenomena like local extinction. The appropriate
approach to achieve this is the implementation of the full mechanism of CH4 oxidation.
However, this approach is related with some difficulties concerning the necessary
computational resources. Therefore the efforts of the researchers are directed mainly to
implementation of the reduced mechanisms, which (although are simplified) provide the
necessary information for correct and in reasonable time calculation of the CH4 turbulent
combustion.
where
Reffectivee – effective rate of the chemical reaction in turbulent reacting flow.
1 1 1
= + , [kg/(m3.s)] (18)
Reffective Rturbulence Rchem.kinetics
This approach is testified and gives more realistic results and is widely applied in
applications involving combustion modeling. That’s why series approach will be used in
development the model for forest fires.
25
CX1HX2OX3NX4SX5ClX6SiX7KX8CaX9MgX10NaX11PX12FeX13AlX14TiX15 + n1H2O +
n2(1+e)(O2+3.76N2) →n3CO2 + n4H2O + n5O2 + n6N2 + n7CO + n8CH4 + n9NO + n10NO2 +
n11SO2 + n12HCl + n13KCl + n14K2SO4 + n15C + ....
The inclusion of 15 elements in the empirical formula for the fuel is incomplete. There are
many more, some of which are important to the issue of biomass combustion. Heavy metals,
for example, have a strong influence on ash disposal, but are not included in the elemental
structure above.
The detailed chemistry describing the simple global reaction above is far from understood.
Making generalization and engineering recommendation concerning the description of biomass
combustion systems is made difficult by the variable composition of biomass , as indicated by
the element coefficients for different types of biomass fuels. Although there are many
similarities, there are also many differences.
The heating value of the gas fraction stabilizes above 700 oC, at a value around 15-16
MJ/Nm3. The composition of the gas fraction versus temperature between 500 and 900 oC does
not vary a lot with the kind of the following tested materials – wood, straw and cocnut shells
[L.Fagbemi]. The concentrations of CH4 and C2Hx reach a maximum value at about 750 oC. A
regular decrease in CO2 concentration with temperature occurs with a simultaneous increase in
CO and H2 concentration. High temperatures are known to favour the production of H2 to the
detriment of higher hydrocarbons which are dehydrogenated by thermal cracking. The
evolution and CO2 concentrations are consistent particularly when considering the
heterogeneous gas-solid reaction at the thermodynamic equilibrium (C + CO2 = 2CO), an
increase in the temperature results in al larger in a larger concentration in CO. The
composition of the yields of the gas is presented in the following table.
28
Figure 13 [14] Relative values of the standard heating value and the heating value (heating value
divided by the mass of products, or equivalently the mass of air and fuel). The values shown are
relative to a bituminous coal (bituminous coal 1).
The versatility of the heating values (as well as the proximate and ultimate analysis of the
fuels) reveals that the combustion process is complex and no common combustion model
could be applied.
Pyrolysis if lingo-cellulosic biomass is very complex process of interdependent reactions;
nevertheless it can be reduced to the reaction illustrated in the following figure, universally
known as the Broido-Shafizadeh mechanism.
30
gas
Light gas
Biomass Tar
char
Char
Secondary reactions are related to the thermal degradation of volatile tars. Strong
interactions occur during secondary reactions. Thus, it was well established, that more solid
char is formed if the volatile compounds constituting tar are confined within the solid matrix,
by increasing the pressure or by slowing down the heating rate. The distinction is commonly
made between fast and slow pyrolysis. Fast (flash) pyrolysis is performed on a coarse material
or at low heating-rates yielding a solid char of up to 35%. Concerning the thermal degradation
of biomass materials, most of the investigations focused on their major component: the
cellulose. The decomposition of cellulose at low temperature leads to two groups of basic
compounds :
• monomeric volatile sugars, such as levoglucosan, resulting from a total depolymerisation
of cellulose ;
• partially-depolymerized compounds, called anhydrocellulose, which are precursors of
the residual solid char;
The subsequent degradation of the primary basic compounds at higher temperatures leads
to products distributed among various fractions or physical phases: solid char, light gas, water
fraction and tar fraction. A knowledge of the kinetics of pyrolysis reactions is useful for the
biomass combustion modelling. For pure cellulose, the kinetics of the primary reactions have
been extensively studied. By comparison, the secondary reactions of tar decomposition are
less well understood.
Fundamental to the combustion rate are the rates of fuel pyrolysis and char oxidation. The
standard method of measuring these rates is via dynamic thermogravimetric analysis (TGA),
whereby a small sample of the fuel (5-15 mg typically) is heated at a controlled rate in a
controlled atmosphere while simultaneously recording weight, time, and temperature. Other
techniques are also employed. The resulting thermogram has a characteristic shape for
biomass. Starting from room temperature, the sample is observed to dry (if the sample contains
moisture which is normally the case, biomass being hygroscopic, and extreme care is needed
in handling to achieve a bone dry sample in the TGA) with a small weight loss up to about 150
o
C. Details of the process are shown in section 2.2. (Pyrolysis and its products) of the report.
The shape of the thermogram (the conversion curve) is dependent on a number of factors,
including the type of fuel, the atmosphere (oxidative, reducing, inert), and the heating rate (for
most apparatus this rate is rather low at 2 K/s or less, whereas full scale combustion processes
in combustors as well as free fires, may heat fuel particles at 100 – 1000 K/s or more). From
the thermogram, kinetic parameters can be determined by which the overall rate of reaction
can be predicted. The following figure shows dimensionless thermogram for conversion of rise
straw
Figure 15 Conversion of rice straw in nitrogen at 1.7 [K/s] by dynamic TGA – experimental data and
three component model predictions [B.M. Jenkins].
A multi-component kinetic model for prediction of the actual conversion consists of three
simultaneous component reaction models employing first order Arrhenius kinetics and based
on the mass fractions of hemicellulose, cellulose, and lignin, are superimposed to yield the
total conversion[17]. Such multi-component models have proved useful in predicting the
overall kinetics behaviour of biomass fuel in oxidative, reducing, and inert atmospheres,
although the fitting technique is not entirely straightforward.
Figure 16 Rate of weight loss under dynamic TGA for rice straw in air at 1.7 [K/s]. The leached
straw was soaked in distilled water for 24 hours.
The figure shows the weight rate loss as s function of temperature. There exists a define
kinetic shift for the leached material relative to the fresh, untreated straw. Although the
activation energy for the leached material is found to be lower than for the untreated materials
in the main stage of pyrolysis, the frequency factor is also lower, resulting in a slower overall
rate of reaction. The result is consistent with what is known about the effects of alkali-
chlorides on the pyrolysis rates of biomass. However, under isothermal heating, the emission
rate for volatiles has been observed to terminate earlier with the leached material than for the
untreated material. The ignition requirements are lower for leached straw compared to
untreated straw (which can be observed quite readily in case of rice straw by suspending straw
vertically and igniting from below). Chlorine is known to retard flame propagation in
polymeric materials (e.g., polyvinylchloride) by terminating free radicals chain reactions.
Chlorine is leached from biomass by water, and the burning phenomena observed may be
related to the role of chlorine in flame suspension. Much remains to be understood about the
chemistry of biomass combustion.
The presented data reveals similarities of the properties of biomass and fossil fuel.
However, relative to the coal, the straw devolatilises more rapidly, produces a higher yield of
volatiles (mostly CO and H2) and consequently has a shorter ignition delay. The set of data
34
were applied in CFD model [23] and the results are compared with experimental data. This
information will provide useful insights into the methodology of biomass combustion
modelling.
YHg2O - mass fraction concentration of the moisture in the gas flow, [-];
35
The rate of formation of volatiles from biomass devolatilization is taken from an Arrhenius
type of the following expression:
Rvol=kvol.mvol – devolatilization rate , [kg/s];
where
⎛ 16600 ⎞
⎜− ⎟
⎜ Ts ⎟⎠
kvol – constant of the rate of devolatilization , k vol = 1.56.10 exp
10 ⎝
, [1/s];
mvol – mass of the volatiles remaining in the biomass particle , [kg] ;
γi – mass fraction of different species formed during biomass pyrolysis , [-] ;
Ts – temperature of the solid fuel, [K] ;
The yield and evolution patterns of the volatile products are a function of temperature, to
miner extent-heating rate and particle size, etc. The volatiles composition mass fraction differ
for biomass fuels, the data are obtained by laboratory experiments and is obtained from the
literature. For simplification, hydrocarbons such as acetaldehyde (CH3CHO), acetic acid (
CH3COOH ) and other very minor compounds are treated as tar. The yields of different
pyrolysis gas are presented in the following table:
The kinetic rate for tar oxidation is obtained by the following expression:
36
⎛ 9650 ⎞
⎜− ⎟
⎜ T ⎟
Rtar = 2.9.10 .Te . exp
5 ⎝ e ⎠ 0.5
YCH Y - tar kinetic reaction rate, [mol/(m3.s)]
1.84 O0.96 O2
3 RCH 4 = k CH 4 YCH
0.7
Y 0.8 ⎛ 24157 ⎞
⎜− ⎟
CH 4 + O2 → CO + 2 H 2 O 4 O2 ⎜ T f ⎟⎠
2 k CH 4 = 1.6.10 exp10 ⎝
1 ⎛ 1⎞ ⎛2 ⎞
C+ O2 → 2⎜1 − ⎟CO + ⎜ − 1⎟CO2 where
Θ ⎝ Θ⎠ ⎝Θ ⎠
Θ - stoichiometric ratio for char combustion,
1
1+
rc
Θ= , where
1 1
+
2 rc
rc – the ration of CO – CO2 formation rate, which can be estimated by :
37
⎛ 3300 ⎞
⎜− ⎟
CO ⎜ TS ⎟⎠
rc = = 12 exp ⎝
CO2
The char combustion rate is controlled by mixed gas film diffusion and chemical reaction.
The film diffusion is taken into account with a correlation for gas flow through combustion
volume. The combustion volume is modelled as packed bed with appropriate porosity.
⎛ 0.765 0.365 ⎞
ε b J = k⎜ 0.82
+ 0.368 ⎟ where
⎝ Re Re ⎠
ε b - the bed porosity ;
ρ fVf dP
Re = - Reynolds number, [-] ;
νf
dP – biomass fuel particles diameter , [m] ;
Vf – gas velocity , [m/s] ;
J – factor, defined as follows:
Sh
J= where:
⎛ 13 ⎞
⎜ Sc Re ⎟
⎜ ⎟
⎝ ⎠
kd d p
Sh = – Sherwood number, [-] ;
DO2
DO2 - molecular diffusion coefficient of O2 in the air ;
Vf
Sc =
(ρ f DO2 ) – Schmidt number, [-] ;
38
REFERENCES:
4. Johnson E. A., Miyanishi K. Forest fires, behavior and ecological effects, 2001.
Academic Press.
5. B.M. Jenkins, L.L. Baxter, T.R. Miles Jr., T.R. Miles, “Combustion properties of
biomass”, Fuel Processing Technology, 54 (1998) 17-46.
9. B.M. Jenkins, L.L. Baxter, T.R. Miles Jr., T.R. Miles, “Combustion properties of
biomass”, Fuel Processing Technology, 54 (1998) 17-46.
13. Weber RO. Modelling fire spread through fuel beds. Prog Energy. Combust Sci
1990;17:67–82.
14. Grishin AM. Mathematical modelling of forest fires. Num Meth. Cont Mech
1978;9:30–56.
15. Grishin AM. Mathematical modeling of forest fires and new methods of fighting them.
Tomsk State University: Publishing House of the Tomsk State University; 1997.
39
16. Larini M, Giroux F, Porterie B, Loraud JC. A multiphase formulation for fire
propagation in heterogeneous combustible media. Int J Heat Mass Trans 1998;41:881–
97.
17. Bellemare LO, Porterie B, Loraud JC. On the prediction of firebreak efficiency.
Combust Sci Technol 2001;163:131–76.
18. Sero-Guillaume O, Margerit J. Modelling Forest fire: Part 1 and Part 2. Int J Heat Mass
Trans 2002;45:1705–37.
19. Porterie B, Morvan D, Loraud JC, Larini M. Firespread through fuel beds: Modeling of
wind-aided fires and induced hydrodynamics. Phys Fluids 2000;12:1762–81.
20. Marcelli T, Santoni PA, Simeoni A, Leoni E, Porterie B. Fire spread across pine needle
fuel beds: characterization of temperature and velocity distributions within the fire
plume. Int J Wildland Fire 2004;13:37–48.
21. Morvan D, Dupuy JL. Modeling the propagation of a wildfire through a Mediterranean
shrub using a multiphase formulation. Combust Flame 2004;138:199–210.
22. Cheney NP, Gould JS, Catchpole WR. The influence of fuel, weather and fire shape
variables on fire spread in grasslands. Int J Wildland Fire 1993;3(1):31–44.
23. Cheney NP, Gould JS. Fire growth in grassland fuels. Int J Wildland. Fire
1995;5(4):237–47.
Wildfire Combustion Chemistry & Smoke