Chemical Engineering Science 69 (2012) 669–678

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Science
journal homepage: www.elsevier.com/locate/ces

Dynamic modeling of combustion in a BioGrate furnace: The effect of
operation parameters on biomass firing
Alexandre Boriouchkine n, Alexey Zakharov, Sirkka-Liisa Jämsä-Jounela
Aalto University, School of Chemical Technology, Department of Biotechnology and Chemical Technology, Process Automation Research Group, P.O. Box 16100, 00076 Aalto, Finland

a r t i c l e i n f o

abstract

Article history:
Received 1 February 2011
Received in revised form
28 October 2011
Accepted 21 November 2011
Available online 28 November 2011

The development of efficient operation and control of a biomass boiler requires extensive knowledge of
the combustion process inside the boiler furnace However, it is not possible to obtain the required
knowledge through process measurements because the high temperatures and aggressive environment
inside the furnace prevent taking accurate sensor readings. Instead, the process can be studied with the
help of mathematical modeling. This paper describes dynamic modeling of bed combustion in a
BioGrate boiler furnace. The developed dynamic model is heterogeneous, including solid and gas phases
and corresponding reactions. The model is used for process phenomena investigation; the results are
presented and discussed.
& 2011 Elsevier Ltd. All rights reserved.

Keywords:
BioGrate boiler
Renewable energy system
Grate firing of biomass
Power generation
Dynamic modeling
Conservative finite difference method

1. Introduction
Legislation supporting sustainable development is forcing
industry to increase the share of electricity produced from renewable fuel. Consequently, the combustion of biomass is increasingly
drawing the attention of the industry. However, the utilization of
renewable energy sources poses several challenges. The challenges
are mainly the result of significant variations in the quality of the
fuels used. Biomass fuel quality depends mostly on the origin of
the material and its processing technique. Moreover, different fuel
batches can contain several different types of biomass. Consequently, each batch can consist of a wide range of fuels, whose
varying properties cause large fluctuations during combustion and
create challenges for process control and optimization. Extensive
understanding of biomass combustion is therefore required to
solve the challenges related to the process. Mathematical modeling can be used to acquire such knowledge (Jämsä-Jounela, 2007).
One of the most recently developed successful processes that use
wood waste as fuel is based on the BioGrate boiler technology
developed by MW Biopower. This technology was developed
especially for the combustion of fuel with high moisture content,
for example, wood chips. It is therefore important to study, among
other effects of fuel properties, the effect of moisture content on the
process behavior, and it is important to further optimize the process
to ensure the maximal energy production rate. Wood combustion is
n

Corresponding author. Tel.: þ358 9 470 23178; fax: þ 358 9 470 23854.
E-mail addresses: alexandre.boriouchkine@aalto.fi (A. Boriouchkine),
zakharov@cc.hut.fi (A. Zakharov), sirkka-l@tkk.fi (S.-L. Jämsä-Jounela).
0009-2509/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2011.11.032

however a very complex process, involving several tightly coupled
chemical reactions and phase changes. Furthermore, the operational
conditions of the furnace greatly affect the yields of chemicals
produced during the combustion process, for example, fractions of
tars, gases, and char. Moreover, not only the yields of chemicals
differ under different combustion conditions but also their reactivity
in succeeding reactions. As a result of such complexity, optimization
of boiler operation requires a detailed process model.
The most recent publications on the modeling of solid fuel
combustion on a grate concentrate on the combustion of either straw
or municipal waste, including biomass. Shin and Choi (2000) have
developed a one-dimensional model of waste incineration to better
understand the phenomena occurring inside a municipal solid waste
(MSW) incinerator. Van der Lans et al. (2000) have developed a twodimensional, homogeneous model of the straw combustion process
to optimize its design and operation parameters. Goh et al. (2001)
have constructed a model of grate combustion of municipal solid
waste in order to study the process, since efficient incinerator design
requires extensive knowledge of the combustion process. Later, Yang
et al. (2002) developed a 2-D model of an MSW incinerator, which
was then verified using experimental data obtained from a pot
reactor. Kær (2005) has developed a one-dimensional model to
describe the fixed-bed combustion of straw and using the walking
grate concept extended to cover the moving bed of a straw bed.
Similarly, Zhou et al. (2005) assume insignificant horizontal temperature gradients in order to simplify to one dimension the thermal
conversion model of a straw boiler furnace.
This paper describes the development of a mechanistic model
of a BioGrate furnace. The aim was to construct a dynamic model

670

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

providing an insight into the chemical and physical phenomena
occurring in the process. Moreover, in describing the fuel combustion, close attention was paid to the kinetics of reactions
because many previous studies neglected the fact that reaction
rate equations are only valid for the conditions at which they
were defined. Specifically, the validity of the reaction rate equations used was evaluated under the operation conditions of the
BioGrate boiler furnace. Furthermore, the newest available kinetic
knowledge on char combustion and pyrolytic gas composition
was utilized to describe wood combustion. This paper is organized as follows: Section 2 describes the power generation
process in a BioGrate power plant, Section 3 presents the model
and its aspects, Section 4 discusses the implementation details of
the model, Section 5 presents the simulation results, and Section
6 summarizes the results of simulations.

2. Process description of a BioGrate power plant
A BioGrate furnace consists of the following functional main
parts: a grate and a water-filled reservoir below the grate to
collect the ash. The BioGrate furnace is surrounded by heatinsulating refractory brick walls (combustion chamber), which
reflect the heat radiation back to the grate.
The grate consists of several ring zones. Half of these rings
rotate, and the other half are fixed. Every second rotating ring
rotates clockwise, and the others rotate counterclockwise. This
helps spread the fuel evenly over the surface of the conical grate.
Fuel is fed from below into the center of the grate, where it
quickly dries as a result of the heat radiation that is emitted by
the combusting flue gas and reflected back to the grate by the
brick grate walls. The dry fuel then proceeds to the outer ring of
the grate, where pyrolysis, char gasification, and combustion
occur. The ash and carbon residues fall off the edge of the grate
into a water-filled ash pit. Fig. 1 is a sketch of a BioGrate furnace.
More details can be found in Anon (2011).
The air required in the gasification and combustion reactions is
fed into the grate through grate nozzles underneath the grate
(primary air) and through nozzles in the grate walls (secondary
air). In addition, in order to ensure clean combustion, additional air
can be supplied through the nozzles located at the top of the grate
(tertiary air). Burning produces heat, which is absorbed in several
steps. First, the boiler walls, acting as an evaporator, absorb part of
the energy retained in the flue gases. Next, part of the energy of the
flue gases is transferred to superheaters. In the third phase, heat is
transferred to a convective evaporator. Finally, economizers remove
the remaining heat energy from the flue gases.
The operation principle of a BioGrate power plant is based on
steam generation. As any other biopower plant, a BioGrate power
plant comprises several parts, including a boiler, a turbine generator, a feed-water tank, a water treatment plant, and a flue-gas

Grate rings
Primary air flow Fuel Flow
Fig. 1. BioGrate furnace.

Primary air flow

cleaning system. Solid fuel is fed into the furnace of the boiler,
where it is combusted, generating heat and flue gases. The heat is
used to produce steam. Flue gases contain fly ash, which comprises several harmful components. Before being emitted into the
atmosphere, the flue gases are therefore subjected to a multistage
cleaning procedure that removes fly ash.
The steam produced in the boiler is directed into a generator
turbine, which converts the mechanical energy of steam into
electricity. As steam performs mechanical work, its pressure
decreases, after which it is used to heat utility streams, for
example, of water (Kiameh, 2002). After steam has released
enough energy, it condenses. The condensed steam is called
condensate, which is fed into a feed-water tank along with
pretreated feed water. Inside the tank, the liquid is heated with
bled steam from the turbine. This procedure increases the
energetic efficiency of the process (Kiameh, 2002).

3. Dynamic modeling
The dynamic model of a BioGrate furnace presented in this
article aims to describe the combustion dynamics of biomass fuel
inside the furnace. To provide acceptable accuracy, the model
must consequently not only include the combustion kinetics, but
also the shape of the furnace. The grate is conical and fuel is fed
from the center, which means that the fuel spreads out on the
grate as it moves outward. This enhances combustion by exposing
more fuel to radiation reflected from the wall lining. As a result,
the ‘‘walking grate’’ concept used to simulate combustion on
sloping grates is modified to fit the BioGrate furnace.
The model of combustion kinetics is specifically selected to fit
the operational conditions of the BioGrate. Furthermore, an
experimental model is used to model the radiation distribution
inside the furnace.
Despite the differences between the BioGrate furnace and
those used in traditional combustion techniques, a biomass bed
reacts in all these furnaces in a series of four different chemical
reactions: drying, pyrolysis, char gasification, and char combustion (Peters and Bruch, 2001). Active drying starts when the
temperature of a particle reaches the boiling point of water. Then
high temperature initiates a pyrolysis reaction. The pyrolysis
reaction produces gases, char, and tar. Gases are mainly composed of CO, CO2, H2, and C1–C3 hydrocarbons. Tar contains many
organic components, such as levoglucosan, furfural, furan derivatives, and phenolic compounds (Di Blasi, 1996). Each reaction will
be discussed in detail below.
3.1. Continuity equation models, their parameters and assumptions
3.1.1. Assumptions
Several assumptions are made in order to simplify the modeling work. The assumptions are listed in descending order of
importance:
1. The system is assumed to be one dimensional, so the changes
in temperature and mass along the horizontal axis are
determined by the residence time. Visual observations of
the actual combustion process reinforce this assumption,
since the fuel appears to travel slowly along the grate in a
pile without losing its form. Moreover, heat transfer inside
the bed can be assumed to occur mostly in the vertical
direction, since heat spreads towards cold areas in the
direction of a normal to the bed surface. Since the height of
the bed is significantly smaller than the length, the temperature gradient forms an almost 901 angle with respect to the
horizontal axis.

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

671

 e  T 3
s
1e 100

ð7Þ

2. Plug-flow gas assumption (Zhou et al., 2005). The gas phase is
assumed to be ideal (Zhou et al., 2005; Kær, 2005)
3. The solid is assumed to be a porous material (Yang et al.,
2003).
4. Diffusion in the gas phase is neglected since the effect of
convection on the transportation of the gas is significantly
greater (Peters and Bruch, 2001).
5. Pressure dynamics are ignored because the release of gaseous
species is negligible in comparison with the primary air flow,
and pressure evolution can therefore be neglected (Zhou
et al., 2005).
6. Heat produced in char combustion is assumed to be retained
in the solid phase (Zhou et al., 2005).
7. No volume reduction (shrinkage) occurs during drying, pyrolysis, and combustion (Di Blasi, 2009).
8. The temperature of the gas released from solids is the same as
that of the solids (Zhou et al., 2005).
9. The temperature of solids in a discretized block is uniform
(Zhou et al., 2005).
10. The heat capacity of the wood is assumed to be constant (Kær,
2005).
11. No heat loss.

3.1.2. Solid phase continuity equation
The solid phase reactions consist of drying, pyrolysis, and char
combustion:
ð1Þ

3.1.3. Energy continuity equation of the solid phase
The energy equation for the solid phase correlates heat
conduction, heat exchange between phases, energy lost in drying
and pyrolysis reactions, and energy gained in char combustion:


@T s
@
@T s
kcond
C s rs ¼
þ kconv vp ðT g T s ÞRevap DHevap
@x
@t
@x
Rpyr DHpyr þRcomb,C DHcomb,C Rgasi,CO2 DHgasi,CO2
Rgasi,H2 O DHgasi,H2 O

ð2Þ

The radiation reflected from the grate walls to the fuel bed is
described through boundary conditions defined as follows:
At the surface of the fuel bed, x¼a

@T s 
kcond
¼ Iin esT 4s
ð3Þ
@x x ¼ a
To describe the energy flux Iin, an experimental model is used.
The model is based on experimental data from a BioGrate boiler
located in Trolhättan, Sweden.

@T s 
kcond
¼ esT 4s
ð4Þ
@x x ¼ 0
The heat conduction coefficient from the study of Yagi and
Kunii (1957) is used to describe heat conduction in the bed:
0

ke
bð1eÞ
dp krv
þ eb
¼
kg
gðkg =ks Þ þðð1Þ=ðð1=fÞ þ dp krs =kg Þ
kg

ð5Þ

where the coefficient kr describing the void-to-void heat conduction coefficient is presented in Eq. (6). Eq. (7) describes the solid
surface-to-solid surface heat conduction coefficient, krv. The
normal values of g, j, and b for cylinders and spheres, as well
as for granules, are as follows (Yagi and Kunii, 1957):

g ¼ 1, b ¼ 1, and j ¼ 0:040:



0:1952
1e
Ts 3
krv ¼
1 þðe=2ð1eÞÞ e
100

Since wood is a porous material, inter-particular radiation
must also be included in the effective heat transfer coefficient.
According to Janssens and Douglas (2004), the radiative heat
conduction coefficient inside the particle is
ks,rad ¼

4eparticle sT s 3 dcavity
1eparticle

dcavity ¼ 3:5  105



ð6Þ

ð8Þ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffi

eparticle

ð9Þ

During the combustion process, the porosity of the wood
particle changes due to moisture evaporation, pyrolysis, and char
combustion. Increasing porosity consequentially improves heat
transfer inside the pore through heat radiation; therefore, porosity change is also included in the model. In Bryden and Hagge
(2003), particle porosity evolution is accounted for with Eq. (10).

eparticle ¼ 1ðrw þ rc Þ=1500rm =1000

ð10Þ

The effective heat conduction coefficient of the solid phase is
(Janssens and Douglas, 2004):
ks:ef f ¼ eparticle kmax þ ð1eparticle Þkmin þ ks,

Next, the simplified continuity equations are presented.

@rs
¼ Rf
@t

krs ¼ 0:1952

rad

ð11Þ

where kmin is given by Eq. (12) and kmax (W/(m K)) by Eqs. (12)
and (13).
kmin ¼

kg kf iber

ekf iber þ ð1eÞkg

kmax ¼ ekg þ ð1eÞkf iber

ð12Þ
ð13Þ

where kg is the heat conductivity of the gas inside the wood pores
which assumed to be air.
According to Yagi and Kunii (1957), flowing fluid improves the
heat conduction of the bed due to the effect of axial dispersion;
therefore, the effective heat conduction coefficient becomes
kcond ¼ kcond,0 þ abURePr

ð14Þ

The geometrical constant is reported to typically take values
between 0.1 and 0.13 (Yagi and Kunii, 1957). In the model, the
average of ab ¼ 0.115 is used.
3.1.4. Gas phase continuity equation
Reacted solid components of wood are transferred to the gas
phase; in addition, the gas phase continuity equation includes gas
flow. Gas phase continuity equation is defined as follows:
@
@
ðr e Y Þ ðv r e Y Þ ¼ Ri
@t f b i @x f f b i

ð15Þ

3.1.5. Energy continuity equation of the gas phase
Assuming no heat loss will occur, the energy continuity
equation can be denoted as follows:
@hf
@
r ¼  ðeb vf hf Þkconv vp ðT g T s Þ
@x
@t f
þRcomb,CO DHcomb,CO þ Rcomb,H2 DHcomb,H2

ð16Þ

3.2. Chemical reactions in the model
The thermal decomposition of wood comprises three main
chemical reactions: drying, pyrolysis, and char gasification with
char combustion. In general, the chemical reactions can be
described using experimental or semi-experimental models.

672

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

However, since Arrhenius dependence equations are simple to
use and accurate, they have been used in this work.
3.2.1. Moisture evaporation
Commonly, fuels used in combustion processes contain moisture. Depending on the type of fuel, a fuel particle can contain
various amounts of moisture. According to Thunman et al. (2004),
fuel particles can contain up to 60 wt% of moisture with the char
residue being as low as 10 wt% of the wet wood. Water can be
bound to the structure of a wood particle or can reside in its pores.
Di Blasi et al. (2003) presented a simple yet accurate model to
describe drying kinetics in updraft gasifiers, which, similarly to the
BioGrate furnace, utilize countercurrent combustion conditions:
8

Revap ¼ 5:6  10 expð88½kJ=mol=R=T s rWater

ð17Þ

pyrolytic gas, while the fraction of other combustible gases
remains under 10 wt%.
In the presence of water steam and oxygen, carbon monoxide
is known to follow the following rate equation (Babushok and
Dakdancha, 1993):
0:5
Rcomb,CO ¼ 1:3  1014 expð125:5½kJ=mol=R=T g ÞUC CO UC 0:5
O2 UC H2 O

ð20Þ
Hydrogen is reported to react according to the following rate
equation (Babushok and Dakdancha, 1993):
Rcomb,H2 O ¼ 2:96  1011 expð6900½K=T g ÞUC H2 UC 0:92
O2

ð21Þ

Reaction rate of ethylene (Babushok and Dakdancha, 1993):
1:65
Rcomb,C2 H4 ¼ 2  1012 expð126 ½kJ=mol=R=T g ÞUC 0:1
C2 H 4 C O 2

ð22Þ

Reaction rate of acetylene (Babushok and Dakdancha, 1993):
3.2.2. Pyrolysis
After a particle has dried, the next reaction that occurs is
pyrolysis. In pyrolysis, a dry wood particle is decomposed into tar,
volatile organic components, and char. However, the fractions of
tar, gas, and char found in the product yield heavily depend on the
reaction conditions of the combustion process.
Alves and Figueiredo (1989) have presented a mathematical
model of wet wood pyrolysis. The simulation results of this model
were experimentally validated in the temperature range of 298–
780 1C with a wet pine cylinder having the radius of 18.5 mm,
with the water content being 45–49 wt%. The experimental and
simulated results agreed. In the model, pyrolytic components are
assumed to be cellulose and hemicellulose, since other components form less than 9% of the total wood mass (Alves and
Figueiredo, 1989). However, these 9% were included in the mass
fraction of hemicellulose, since these components and hemicellulose have the same reaction enthalpy and reaction rates of
comparable magnitude. Cellulose constitutes 50 wt% of dry wood,
while hemicellulose and other pyrolytic components 28 wt%.
Furthermore, since devolatilization of cellulose is an endothermic
reaction, the overall pyrolysis is also endothermic. Table 1 presents the reaction enthalpies of each pyrolytic component.
Cellulose reacts according to the following reaction rate
equation (Alves and Figueiredo, 1989):
Rdevol,cel ¼ 2  10

9

expð146 ½kJ=mol=R=T s Þrcel

ð18Þ

Hemicellulose is reported to react according to the following
reaction rate equation (Alves and Figueiredo, 1989):
Rdevol,Hemi ¼ 7  104 expð83 ½kJ=mol=R=T s ÞrHemi

ð19Þ

3.2.3. Combustion of pyrolytic gases
The yield of pyrolytic gases is around 85 wt% under the
operation conditions of a BioGrate furnace, for gasifying pyrolysis
is the dominant pyrolysis mode under these conditions. This
means that a significant amount of the energy used by the boiler
comes from the combustion of gases, which indicates that the
combustion of pyrolytic gases is the most important energy
source. However, the composition of the gaseous products of
pyrolysis reported in the study of Dupont et al. (2009) suggests
that carbon monoxide has the highest concentration in the
Table 1
Parameter values for pyrolysis reactions (Alves and
Figueiredo, 1989).
Component

Enthalpy of reaction (kJ/kg)

Cellulose
Hemicellulose

322
 233

1:25
Rcomb,C2 H2 ¼ 6:5  1012 expð126 ½kJ=mol=R=T g ÞUC 0:5
C2 H2 C O2

ð23Þ

Reaction rate of methane (Babushok and Dakdancha, 1993):
Rcomb,CH4 ¼ 5:6  1012 expð103800 ½kJ=mol=R=T g ÞUC CH4

ð24Þ

3.2.4. Char conversion reactions
Char combustion in the model is accounted for with the model
presented in Senneca (2007). This model is chosen here because it
is valid over the temperature range 440–800 1C, which corresponds to the temperature of char combustion in the BioGrate
furnace. In addition, the pyrolysis conditions of char particles that
were used to obtain the model parameters are similar to the
pyrolysis conditions inside the BioGrate. The reaction rate equation for gas combustion defined in Senneca (2007) is presented
below:
RCO ¼ 0:9X  1:4  105 expð100 ½kJ=mol=R=T s ÞP 0:9
O2 rC

ð25Þ

The gasification reaction of char with carbon dioxide (Senneca,
2007):
Rgasi,CO2 ¼ 1:1  109 expð260½kJ=mol=R=T s ÞP0:64
CO2 ð1XÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 12 lnð1XÞrC

ð26Þ

The gasification reaction of char with water steam can become
significant in the case of a BioGrate furnace because the fuel used
contains a large amount of moisture. As a result, the recirculation
gas used to control the temperature of the furnace can contain a
high amount of water steam, promoting char gasification reactions with gaseous water (Matsumoto et al., 2009):
4
Rgasi,H2 O ¼ P0:22
H2 O  9:99  10 expð136½kJ=mol=R=T s Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1XÞ 110 lnð1XÞrC

ð27Þ

4. Implementation of the models and description of the
testing environment
The model is implemented in the MATLAB environment, in
which a set of finite difference methods (FDMs) is used to solve
the continuity equations. The reason for utilizing FDM is that it is
well studied and all its drawbacks and stability issues are known;
as a result, its suitability for a particular task is easy to evaluate.
The Crank–Nicolson method is utilized to solve the energy
equation for the solid phase because it involves heat conduction,
which is a cubic function of the temperature. As a result, even
small differences in temperature between subsequent time steps
can cause severe oscillations; it is therefore necessary to use a
second-order method in time. In addition, the Crank-Nicolson

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

Mass Continuity
of Solids

Computation of
reaction rates

Mass Continuity of
Gases
Next Time Step
Energy Continuity
of Solid Phase
Energy Continuity
of Gas Phase

673

This paper therefore focuses on analyzing the operation of the
furnace based on flue gas composition and mass flow as well as
temperature. The simulation is conducted with pine wood chips
as a fuel with a dry bed density of 150 kg/m3. The bed porosity
was assumed to be 0.5 while the ash content was fixed to be
1 wt% of the dry bed density. The effect of ash on heat conduction
and convection is disregarded due to the small amount of ash
compared with the total bed weight. Furthermore, when the
density of a discretized volume has reached the density of ash,
it is removed from the simulation. To make the visualization of
the result presentation more convenient, the fuel layer height is
fixed at the value of 10 cm despite the actual fuel layer height of
the furnace being several times larger. This is due to fuel movement on the grate which lessens the reaction front depth for
burning matter spreading to the larger area of the outer grate
rings. The spatial step size used in simulations is 0.001 m, while
the time step size is 1 s.

Analytic
Solution
Upwind
Finite
Difference
Method
CrankNicolson
+SecondOrder
Boundary
Conditions

Fig. 2. Scheme of the solution algorithm.

5.1. Fuel moisture content

Qn

xn-1

xn-1/2∆h

xn

xn+1/2∆h

xn+1

Qn-1
Fig. 3. Conservative modification of the Crank–Nicolson method.

method is coupled with second order boundary conditions to
smooth out the steep temperature gradients at boundaries. The
solution algorithm is presented in Fig. 2.
Furthermore, the Crank-Nicolson method is made conservative
by considering the heat conduction term in the middle of a
discretization interval instead of considering it at the discretization points themselves. The reason for this is that FDMs, in
general, evaluate the heat flux leaving from point xn  1 towards
point xn, given in Eq.(28), and separately evaluate the heat flux
arriving at point xn from xn  1, defined in Eq. (29). This property of
FDMs can result in energy imbalance inside the system if the heat
conduction coefficient itself depends on the temperature. Eqs.
(28) and (29) show that these two fluxes are equal only when
kn  1(t) and kn(t) are equal. Therefore, evaluating the heat conductivity coefficient as a function of the average temperatures of
two subsequent discretization points ensures that kn  1 and kn are
equal and that energy is conserved (Ferziger and Peric, 2000).
Fig. 3 illustrates the modification made to the numerical scheme.
Q n1 ¼ kn1
Q n ¼ kn

ðun þ un1 Þ
h

ðun1 un Þ
h

ð28Þ

ð29Þ

5. Simulation results
This section presents the simulation results obtained with
different fuel parameter values.
The operation of a boiler is based on the energy released from
the combustion of fuel. Fuel combustion produces energy in the
form of thermal radiation and the release of flue gases. However,
the refractory grate wall directs radiation back to the fuel bed; in
contrast, flue gases proceed to an evaporator, which produces
steam. This makes flue gases, in addition to heat radiation, the
most important energy source for the operation of the furnace.

Fuel moisture content is the most important quality property
of fuel because excessive moisture in fuel induces large fluctuations in the power production of the boiler.
To investigate the fuel moisture effect on combustion, a
simulation with four different moisture contents was conducted:
70, 60, 50, and 40 wt% of the total fuel weight. Primary air flow
was fixed at 1 m3/s, the particle size was assumed to be 35 mm,
and the bed porosity was assumed to be 0.5. Fig. 4 depicts gas
production rates for this simulation.
The simulation shows that fuels with 60 and 70 wt% moisture
content exhibit a rapid increase in gas production after being
subjected to the furnace temperature. Moreover, these fuels show
a noticeably larger initial gas production rate than fuel with a
lower moisture content. This is because the more moisture the
fuel has, the quicker it will lose excess moisture and reach a stable
drying rate. Nevertheless, after the top layer of the fuel bed has
reached equilibrium in the drying rate, the gas production starts
increasing, suggesting that fuel has ignited. Particularly, drier fuel
ignites much more rapidly than wet fuel. Furthermore, the fuel
with the highest moisture content, 70 wt%, seems not to burn
properly, since it produces a rather slow increase in the gas
production rate. After ignition, all fuels, however, achieve a
relatively constant combustion front propagation rate: 0.03 kg/
m2s for the driest fuel and 0.02 kg/m2s for the wettest one, with
the values for other fuels in between these.
Later, the amount of released gas starts to decrease. The
decrease is the result of the diminished contents of moisture
and organic matter, for reaction rates are directly proportional to
the concentration of reactants. The release of the flue gases
continues to decrease until the fuel is dry. The dry fuel then
undergoes a rapid devolatilization stage because energy is no
longer consumed by drying. This causes a large spike in the gas
production rate, which reaches 0.3 kg/m2s for fuels with moisture
contents of 40 and 50 wt% and reaches as high as 0.40 kg/m2s in
the case of the wettest fuel. The complete devolatilization of
wood leaves only char. Char then participates in gasification and
combustion reactions with carbon dioxide and oxygen. The
devolatilization takes place throughout the entire grate, so there
is time for char combustion on the grate before the char falls off
the rotating grate rings. This allows a gasification reaction
between char and water steam which produces hydrogen.
In addition to the combustion rate propagation, the moisture
content also affects the how much gas is produced during
combustion. The simulation shows that fuel with high moisture
content tends to release less combustible gases because high
moisture content prevents the temperature from rising, thereby

674

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

Fig. 4. (a) The overview of gas production rates of fuels with different moisture content and (b) the gas production rates, a close up.

delaying pyrolysis and limiting char combustion and gasification.
Moreover, reactions that require high temperatures cannot start
at all because the bed temperature does not rise above 500 1C.
There is specifically no gasification at all of char with steam.
The results show that fuels with large moisture content burn
cleaner because they have more oxygen available for the reactions; more specifically, they produce less carbon monoxide than
carbon dioxide. Further, water steam forms almost 80% of the
weight of the gas phase when the fuel moisture content is 70 wt%.
Conversely, in the case of drier fuel, the weight fraction of carbon
monoxide is almost twice as large as that of carbon dioxide.
Fig. 5(a) and (b) presents the compositions of gases emitted from
fuel beds with 40 and 70 wt% moisture content, respectively. In
addition, dry fuel, while combusting, releases higher amounts of
hydrogen, which is produced in the char gasification reaction
with steam as well as in pyrolysis. In addition, the ignition of dry
fuel is more apparent since a sharp rise in CO2 can be observed.
However, the CO2 production from moist fuel increases gradually
until the maximum is reached, after which it remains constant at
a near-maximum level.
The composition of the flue gases consequently affects their
temperature, for a large amount of energy is produced by the gas
phase reactions. The temperature profiles of the gas phase clearly
show that dry fuel produces considerably hotter gas than the fuel
with the greatest moisture content. Specifically, the temperature
of the flue gas leaving the fuel bed in the case of the fuel with a
moisture content of 70 wt% has a temperature of 1000 1C,
whereas, the corresponding temperature for fuel with a moisture
content of 40 wt% is over 1500 1C.
The same effect on the fuel layer temperature profile is
observed, for the drier fuel exhibits higher temperatures during
combustion. More specifically, the temperature of the moist fuel
does not rise above 500 1C, whereas the temperature of fuel with
a moisture content of 40 wt% remains above 600 1C. Figs. 6 and 7
show the temperature profiles of fuels in the gas and solid phases

and with different moisture contents. The combustion of drier
fuel produces higher temperatures in both the gas and the solid
phases, so it also affects the rates of chemical reactions involved
in the biomass combustion. As a result, the oxygen content inside
the fuel bed is also affected by the moisture.
The oxygen content profile reveals that the fuel with a
moisture content of 70 wt% consumes almost no oxygen at all,
whereas combustion of the drier fuels suffers from severe oxygen
deficiency. This suggests that fuels with a low moisture content
can be combusted with air flows greater than 1 m3/s, whereas
very moist fuel requires a much smaller air flow since excess air
will only decrease the temperature of the bed.
In order to study the effect of a volumetric air flow, additional
simulation was conducted using varying air flows while the fuel
moisture content was fixed.
5.2. Effect of air flow variation on fuel combustion
The effect of air flow variation on fuel combustion was studied
using the four different air flows of 0.5, 1, 2, and 4 m3/s while the
fuel moisture content was fixed at 50 wt%.
The simulation results show that the reaction front width
decreases as the combustion air flow is increased. The decreased
width is a consequence of increased oxygen flow since all
available char can react. As a result, the mass of the fuel bed
decreases more rapidly. Moreover, the increased air flow also
cools the gas released from the bed because the air flow
temperature is significantly lower than that of the flue gas.
Increased air flow results in flue gas temperatures below
1000 1C with an air feed of 4 m3/s, compared to above 1500 1C
with an air flow of 1 m3/s. Nevertheless, the difference in the
mass degradation rates caused by the air flows of 1 and 4 m3/s is
insignificant, for the fuel reaches a point 1.8 m from the grate
center with the air flow of 4 m3/s, whereas the fuel is fully
degraded at a point 1.6 m from the center with the air flow of

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

675

Fig. 5. The composition of flue gas originating from fuel with a moisture content of (a) 40 wt% and (b) 70 wt%.

Fig. 6. Gas phase temperature profiles for fuels with (a) 70 wt% and (b) 40 wt%
moisture contents with primary air flow of 1 m3/s.

Fig. 7. Solid phase temperature profiles for fuels with (a) 70 wt% and (b) 40 wt%
moisture content with primary air flow of 1 m3/s.

1 m3/s. Fig. 8 presents a comparison of the mean gas production
rates for different primary air flows. The comparison shows that
the mean combustion front propagation rate reaches its maximum when the air flow is 2 m3/s. Higher air flows begin to cool

the bed, resulting in a decrease in the propagation rates; especially carbon dioxide is produced less when the temperature has
decreased below the critical point.
Moreover, air flows of 2 and 4 m3/s cause oscillations in the
gas production because they accelerate the combustion of the

676

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

in the case of particles measuring 15 mm. Chemical reactions heat
the fuel layer, but as its temperature rises, the temperature
gradient between the solid and gas phases become larger. As a
result, the rate of heat transfer between these phases increases,
resulting in oscillations. In contrast, large particles resist cooling
more efficiently because they provide less surface for heat
transfer than smaller particles do. Moreover, larger particles
improve heat transfer inside the bed, enabling heat to spread
more evenly in the bed, which causes much smaller temperature
gradients in the bed. Consequently, improved heat transfer
enhances the drying of the bed, but since the local temperatures
of large particles are smaller than those of small particles,
pyrolysis and char combustion take much longer. Table 2 shows
the mean gas production rates for different particle sizes. Results
show that as particle size is decreased by 10 mm the average
combustion front propagation is increased by 2.5%, the largest
propagation rate for particle diameter 15 mm reaches up to
0.0225 kg/(m2s), while largest particle size, 50 mm, stays well
below 0.02 kg/(m2s).
Fig. 8. Mean gas production rates for different primary air flows.

5.4. Effect of gas recirculation on biomass fuel combustion
available char, as a result of which the char content is depleted
and nothing happens until more char is formed in the pyrolysis
reaction. In contrast, the flow of 1 m3/s enables a steady char
combustion rate without oscillations. In addition, the increased
amount of oxygen supplied by larger air flows accelerates the
combustion of the char. However, large airflows decrease the rate
of the other reactions. These include the gasification reactions and
the reactions taking place in the gas phase because they have high
activation energies and are; therefore, limited by the decreased
temperatures of the gas and the solid phases. In addition,
simulations show that the larger the air flow, the more carbon
monoxide is produced, whereas the release rate of carbon dioxide
shows the opposite behavior: it decreases with increasing air
flow. The reason for such behavior is that additional oxygen is
consumed by char oxidation and no oxygen is left to oxidize CO to
CO2.
To conclude, the simulation results suggest that the char
combustion reaction is diffusion limited, whereas the gasification
and gas phase reactions are kinetically limited. Optimization of
fuel combustion thus requires a compromise between char
combustion and the other reactions by choosing an appropriate
volumetric air flow. Particle size also affects the convective heat
exchange because a bed with smaller particles provides a larger
area for heat transfer. In addition, heat conduction inside the bed
also depends on particle size because a bed with smaller particles
has a larger area for absorbing the radiated energy. The next
investigation is; therefore, on the effect of particle size on the
combustion process.

Gas recirculation is an important way of controlling the
combustion efficiency and temperature of the furnace. It is,
therefore, important to study how the recirculation specifically
affects the combustion of fuel. In order to investigate the effect of
recirculation on the burning behavior of biomass, the initial
conditions of gas enthalpy and composition were changed. The
Table 2
Mean combustion front propagation rates for
different particle sizes.
Particle
diameter (mm)

Mean front propagation rate
(kg/m2 s)

15
25
30
40
50

0.0224
0.0218
0.0209
0.0202
0.0197

5.3. Effect of particle size on biomass combustion
The simulation with different particle sizes was conducted
using five different particle sizes: 50, 40, 30, 25, and 15 mm.
Moisture content and airflow were fixed at 50 wt % and 1 m3/s,
respectively.
The simulation results show that the bed with the smallest
particle size ignites most rapidly, for smaller particles absorb heat
radiation energy more efficiently. The ignition delays are 23, 29,
37, 45, and 53 s for the particle diameters 15, 25, 30, 40, and
50 mm, respectively. Ignition time increases with particle size,
and the largest particles take twice as long to ignite as those with
the smallest diameter. However, smaller particles are more prone
to cooling, and this effect is observed in the simulation results as
temperature oscillations. The oscillations are the most significant

Fig. 9. Comparison of temperature profiles of fuel bed (a) with and (b) without
recirculation.

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

following components were added to the initial conditions:
0.5 kg/m3 of water steam, 0.5 kg/m3 of CO and CO2 each. The
added components were assumed to have a temperature of
200 1C. Particle size and porosity were fixed at 35 mm and 0.5,
respectively. To avoid oxygen deficiency, primary air flow was
fixed at 2 m3/s.
The obtained simulation results showed that the maximum
temperatures of the fuel layer and the released gas indeed
decreased to 900 and 2000 1C when recirculation was introduced.
Prior to the introduction of gas recirculation, these were 1050 and
2700 1C, respectively. Despite the decreased maximal temperatures, the gas recirculation, nevertheless, accelerated the thermal
conversion of fuel because the increased temperature of the gas
flow enhanced the drying of the fuel, allowing the consequent
reactions to start earlier. Fig. 9 compares the fuel temperature
profiles of simulations with and without recirculation.

6. Conclusion
A one-dimensional model of a BioGrate furnace was developed
to investigate the process phenomena in the furnace. The model
was especially designed to study the effects of different operation
conditions of a BioGrate furnace on biomass firing. The newest
available reaction rate equations were utilized to describe char
combustion and gasification with CO2 and H2O. The simulation
results showed the significant dependence of the combustion
process on moisture content: the moisture content significantly
affected both the temperature of the gas and the solids. Lower
moisture content permitted higher temperatures of the flue gas
and fuel layer, allowing rapid combustion of fuel, whereas high
moisture content delayed pyrolysis and char reactions, decreasing
the overall conversion rate of fuel. The study of the effect of
primary air flow suggested that low air flows induced oxygen
deficiency in the combustion process, whereas larger flows cooled
the fuel bed and decreased the overall combustion rate of fuel.
Simulations with varying particle sizes revealed that small particles ignite quicker than large ones due to the ability of the small
particles to absorb heat radiation more efficiently because of their
large heat exchange area. In contrast, large particles absorbed
radiation less efficiently, but they were less affected by convective
heat transfer in the gas phase. This finding is supported by the
investigation of Horttanainen et al. (2002) who noticed that a
mixture of wood chips and saw dust burned better than when
these fuels burned separately. Horttanainen et al. (2002) proposed that sawdust helped ignite the larger particles. The current
simulation suggests that burning larger particles are less affected
by gas flow, enabling the use of larger air flows, which accelerate
fuel combustion. The study of flue gas recirculation proved the
ability of recirculated flue gas to stabilize the combustion temperature, which prevents rapid temperature spikes in both the
solid and gas phases. Recirculation also enhanced fuel combustion
by heating the solid fuel. Further studies will study the validity of
the developed model with data obtained from a BioGrate boiler
furnace since pot furnace experiments are not fully able to
describe the operation conditions of a boiler furnace. This is
supported by a finding of Thunman and Leckner (2001), who
noticed that fuel that burned in a boiler furnace with no problems
could not be properly fired in a pot furnace.

Nomenclature

DHcomb,C reaction enthalpy of char combustion (J/kg)
DHevap reaction enthalpy of drying (J/kg)

677

DHgasi,CO2 reaction enthalpy of char gasification with carbon
dioxide (J/kg)

DHgasi,H2 O reaction enthalpy of char gasification with water
steam (J/kg)
reaction enthalpy of pyrolysis (J/kg)
CC2 H2
concentration of acetylene (mol/cm3)
CCO
concentration of carbon monoxide (mol/cm3)
CC2 H4
concentration of ethylene (mol/cm3)
CH2
concentration of hydrogen (mol/cm3)
C CH4
concentration of methane (mol/cm3)
CO2
concentration of oxygen (mol/cm3)
C H2 O
concentration of steam (mol/cm3)
Cs
the heat capacity of the solid phase (J/(kg K))
dp
particle diameter (m)
esT4s
energy flux out of the system (W/m2)
h
step size (m)
hf
enthalpy of the gas phase (J/kg)
Iin
energy flux into the system (W/m2)
R
gas constant (J/(mol K))
kcond
heat conduction coefficient of the solid phase (W/(m K))
kconv
heat convection coefficient between the gas and solid
phases (W/(m2K))
kfiber
heat conductivity of wood fiber (W/(m K))
kg
heat conductivity of the gas inside the wood pores
(W/(m K))
0
ke
effective heat conduction coefficient of the packed bed
with no fluid flow (W/(m K))
kg
heat conduction coefficient of the fluid (W/(m K))
ks
heat conduction coefficient of the solid matter
(W/(m K))
krs
void-to-void heat conduction coefficient (W/(m K))
krv
solid surface-to-solid surface heat conduction coefficient
(W/(m K))
kmin
minimal heat conduction coefficient (W/(m K))
kmax
maximum heat transfer coefficient (W/(m K))
kn  1
value of the heat conductivity coefficient at xn  1
(W/(m K))
kn
value of the heat conductivity coefficient at xn (W/(m K))
PCO2
pressure of carbon dioxide (Pa)
P H2 O
pressure of water steam (Pa)
P O2
pressure of oxygen (Pa)
Qn  1
energy flux leaving from point xn  1 towards xn (W/m2)
Qn
energy flux arriving at point xn from xn  1 (W/m2)
Rcomb,C reaction rate of char combustion (kg/(m3 s))
Revap
reaction rate of drying (kg/(m3 s))
Rf
overall reaction rate of the solids (kg/(m3 s))
Rgasi,CO2 reaction rate of char gasification with carbon dioxide
(kg/(m3 s))
Rgasi,H2 O reaction rates of char gasification with water steam
(kg/(m3 s))
Rcomb,H2 oxidation rate of hydrogen (mol/(cm3 s))
Rcomb,CH4 oxidation rate of methane (mol/(cm3 s))
Rcomb,C2 H2 oxidation rate of acetylene (mol/(cm3 s))
Rcomb,C2 H4 oxidation rate of ethylene (mol/(cm3 s))
Rcomb,CO oxidation rate of carbon monoxide (mol/(cm3 s))
Ri
rate of formation of gaseous component i (kg/(m3 s))
Rpyr
reaction rate of pyrolysis (kg/(m3 s))
Tg
temperature of the gas phase (K)
Ts
temperature of the solid phase (K)
u
discretized value of the temperature function (K)
vf
gas flow velocity (m/s)
vp
density number (m  1)
X
degree of conversion of char
x
vertical coordinate (m)
Yi
mass fraction of the gaseous component i
eb
bed porosity

DHpyr

678

rc
rcel
rCO
rf
rH2
rHemi
rO2
rWater
s

A. Boriouchkine et al. / Chemical Engineering Science 69 (2012) 669–678

density of the char (kg/m3)
density of cellulose (kg/m3)
density of carbon monoxide (kg/m3)
density of the gas phase (kg/m3)
density of hydrogen (kg/m3)
density of hemicellulose (kg/m3)
density of oxygen (kg/m3)
density of water (kg/m3)
Stefan–Boltzman constant (W/(m2 K4)

Acknowledgments
This work was partially supported by the Dynergia consortium
project of the Finnish Funding Agency for Technology and
Innovation (TEKES). The first author gratefully acknowledges the
financial support by Finnish Graduate School in Chemical Engineering. The authors would like to recognize the comments and
discussions provided by Dr. Juha Huotari and Dr. Jaakko Saastamoinen, whose contributions have been invaluable.
References
Alves, S.S., Figueiredo, J.L., 1989. A model for pyrolysis of wet wood. Chem. Eng. Sci.
44, 2861–2869.
Babushok, V.I., Dakdancha, A.N., 1993. Global kinetic parameters for high-temperature gas-phase reactions. Combust. Explosion Shock Waves 29, 464–489.
Bryden, K.M., Hagge, M.J., 2003. Modeling the combined impact of moisture and
char shrinkage on the pyrolysis of a biomass particle. Fuel 82, 1633–1644.
Di Blasi, C., 1996. Heat, Momentum and Mass Transport Through a Shrinking
Biomass Particle Exposed to Thermal Radiation. Chem. Eng. Sci. 5, 1121–1132.
Di Blasi, C., Branca, C., Sparano, S., La Mantia, B., 2003. Drying characteristics of
wood cylinders for conditions pertinent to fixed-bed countercurrent gasification. Biomass Bioenergy 25, 45–58.
Di Blasi, C., 2009. Combustion and gasification rates of lignocellulosic chars. Prog.
Energy Combust. Sci. 35, 121–140.
Dupont, C., Chen, L., Cances, J., Commandre, J.-M., Cuoci, A., Pierucci, S., Ranzi, E.,
2009. Biomass pyrolysis: kinetic modeling and experimental validation under
high temperature and flash heating rate conditions. J. Anal. Appl. Pyrol. 85,
260–267.

Ferziger, J.H., Peric, M., 2000. Computational Method for Fluid Dynamics, third ed.
Springer-Verlag, Berlin, Heidelberg, New York.
Goh, Y.R., Yang, Y.B., Zakaria, R., Siddall, R.G., Nasserzadeh, V., Swithenbank, J.,
2001. Development of an incinerator bed model for municipal solid waste
incineration. Combust. Sci. Technol. 162, 37–58.
Horttanainen, M., Saastamoinen, J., Sarkomaa, P., 2002. Operational limits of
ignition front propagation against airflow in packed beds of different wood
fuels. Energy Fuels 16, 676–686.
Janssens, M., Douglas, B., 2004. Wood and Wood Products. In: Harper, C.A. (Ed.),
Handbook of Building Materials for Fire Protection, McGraw-Hill, 542 pp.
Jämsä-Jounela, S.-L., 2007. Future trends in process automation. Annu. Rev. Control
31, 211–220.
Kær, S.K., 2005. Straw combustion on slow moving grates-a comparison of model
predictions with experimental data. Biomass Bioenergy 28, 307–320.
Kiameh, P., 2002. Power Generation Handbook. McGraw-Hill.
Matsumoto, K., Takeno, K., Ichinose, T., Ogi, T., Nakanishi, M., 2009. Gasification
reaction kinetics on biomass char obtained as a by-product of gasification in
an entrained-flow gasifier with steam and oxygen at 900–1000 1C. Fuel 88,
519–527.
Peters, B., Bruch, C., 2001. A flexible and stable numerical method for simulating
the thermal decomposition of wood particles. Chemosphere 42, 481–490.
Senneca, O., 2007. Kinetics of pyrolysis, combustion and gasification of three
biomass fuels. Fuel Process. Technol. 88, 87–97.
Shin, D., Choi, S., 2000. The combustion of simulated waste particles in a fixed bed.
Combust. Flame 121, 167–180.
Thunman, H., Leckner, B., 2001. Ignition and propagation of a reaction front in
cross-current bed combustion of wet biofuels. Fuel 80, 473–481.
Thunman, H., Davidsson, K., Leckner, B., 2004. Separation of drying and devolatilization during conversion of solid fuels. Combust. Flame 137, 242–250.
Van der Lans, R.P., Pedersen, L.T., Jensen, A., Glarborg, P., Dam-Johansen, K., 2000.
Modelling and experiments of straw combustion in a grate furnace. Biomass
Bioenergy 19, 199–208.
W.W.W., Anon, MW Power Brochure, /http://www.mwpower.fi/mwpower/ru/
mwpower_pages_ru.nsf/WebWID/WTB-090524-22575-77914/$File/Biopo
wer_5_screen.pdfS, 28.10.2011.
Yagi, S., Kunii, D., 1957. Studies on effective thermal conductivities in packed beds.
A.I.Ch.E. J. 3, 373–381.
Yang, Y., Goh, B., Zakaria, Y.R., Nasserzadeh, R., Swithenbank, J., V., 2002.
Mathematical modelling of MSW incineration on a traveling bed. Waste
Manage. 22, 369–380.
Yang, Y.B., Yamauchi, H., Nasserzade, V., Swithenbank, J., 2003. Effects of fuel
devolatilization on the combustion of wood chips and incineration of simulation municipal solid wastes in a packed bed. Fuel 82, 2205–2221.
Zhou, H., Jensen, A.D., Glaborg, P., Jensen, P.A., Kavaliauska, A., 2005. Numerical
modeling of straw combustion in a fixed bed. Fuel 84, 389–403.