You are on page 1of 7

Fuel xxx (2014) xxxxxx

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Towards a hybrid EulerianLagrangian CFD modeling of coal gasication


in a circulating uidized bed reactor
Adam Klimanek , Wojciech Adamczyk, Anna Katelbach-Wozniak, Gabriel Wecel, Andrzej Szlek
Institute of Thermal Technology, Silesian University of Technology, Konarskiego 22, 44-100 Gliwice, Poland

h i g h l i g h t s
 We developed a CFD model of coal gasication in a circulating uidized bed reactor.
 EulerianLagrangian approach was used to simulate the uidized bed hydrodynamics.
 Gasication with air and air/steam mixture was considered.
 Fourteen heterogeneous and homogeneous reactions were considered.
 Results of the simulations coincide well with the measured syngas composition.

a r t i c l e

i n f o

Article history:
Received 28 July 2014
Received in revised form 10 October 2014
Accepted 22 October 2014
Available online xxxx
Keywords:
Coal gasication
Circulating uidized bed
EulerLagrange
DDPM

a b s t r a c t
Numerical model of coal gasication in circulating uidized bed (CFB) using EulerianLagrangian
approach is presented in this paper. The Dense Discrete Phase Model (DDPM) model of ANSYS FLUENT
is used to simulate the ow of the particulate phase in the coal gasier. The coal particles, with a size
distribution, are tracked in the uid velocity eld including coupling between the phases. Kinetic Theory
of Granular Flow is utilized to model the particles interactions. The analyzed CFB comprises a small scale
experimental facility in which coal is gasied in air and air/steam mixture. The reactor is composed of a
barrel like bottom part with developed internal recirculation of the solid phase and 3.74 m high riser section. The homogenous gas phase reactions are modeled using the nite rate and eddy dissipation models.
The heterogeneous reactions on coal particles surface are modeled using the nite rate chemistry. A total
number of 14 reactions are considered. Results of the simulations were compared with experimental
data.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Circulating uidized beds (CFBs) are frequently used in industry
for variety of processes, for example combustion, gasication,
heterogeneously catalyzed reactions, etc. Modeling of the CFBs is
challenging due to complexity of the hydrodynamic behavior of
the particle laden ow, which is further augmented by the heat
and mass transfer as well as multiple species and reactions occurring in reactors and boilers. The uidized medium is composed of
particles usually of various sizes and frequently of various materials, e.g. coal and sand. Modeling of industrial CFBs by means of
computational uid dynamics tools faces further challenges due
to large geometrical scales of these facilities and inherent long
computational times of these transient ows. Several approaches
Corresponding author. Tel.: +48 32 2372974; fax: +48 32 2372872.
E-mail address: adam.klimanek@polsl.pl (A. Klimanek).

can be used to model gas-particle ows. The methods differ by


the covered spatial and temporal scales of the ow phenomena
[1]. The associated computational effort increases quickly with
increasing spatial and temporal resolution of the models and
therefore not all approaches can be applied to pilot scale and large
industrial facilities, which however depends on the available
computational resources. The mathematical models covering
scales larger than the particle size, namely the EulerianEulerian
(multi-uid) and hybrid EulerianLagrangian are nowadays most
frequently used for this purpose. In the EulerianEulerian approach
both the solid and the uid phase are treated as interpenetrating
continua. Use is made of the Kinetic Theory of Granular Flow
(KTGF) ([2,3]) which allows determination of the solid stresses
by many closure terms in the set of governing equations of the
model. The solid phase is represented by its density and a characteristic diameter. In reacting ows frequently the particle size distribution (PSD) is of importance. If the PSD needs to be taken into

http://dx.doi.org/10.1016/j.fuel.2014.10.058
0016-2361/ 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

2. The numerical model


2.1. Geometry and mesh
The modeled reactor is installed at the Institute for Chemical
Processing of Coal in Poland [23] where measurements of the analyzed cases have been done. The reactor has been used in the past
for coal pyrolysis and gasication and recently it has been proposed to apply it to gasication with carbon dioxide [2426]. The
reactor encompasses a barrel like bottom part, where internal
recirculation of the solid phase develops, a riser section and an
expanding section at the top from which the gases are directed
to the solids separator. Geometry of the model with its main
dimensions is presented in Fig. 1 and more details of the test facility can be found in [26,27]. Coal is introduced at the side of the barrel part and the gasifying agent at the very bottom of the reactor
through a grid inlet. The unreacted char can be circulated to the
barrel like part of the reactor, however in the investigated cases
it was not. Two meshes were generated and examined: mesh 1
built of 53.7 thousand elements and mesh 2 composed of 90.8
thousand elements. Both quadrilateral and tetrahedronal elements
were used to create the mesh. Simulations were done for both
meshes to verify the sensitivity of the results to the mesh density.
2.2. Governing equations
The DDPM formulation is based on governing equations similar
to those of the multi-uid approach. Additionally, the particle
equation of motion is solved for each parcel and the parcel
properties are projected to the Eulerian grid. This provides mean
solids velocity and volume fraction in each grid cell. Therefore
the continuity and the momentum equations of the solid phase
are not solved in the Eulerian frame. The particleparticle interactions are determined by the KTGF on the Eulerian grid and are

0.65

syngas
outlet

3.74

0.22

0.14

5.46

account, a couple of solid phases of the same medium with distinct


representative diameters can be included. This however increases
substantially the computational effort with increasing number of
phases. An alternative approach is to apply one of the methods
of moments QMOM ([4]) or DQMOM ([5]). The EulerianEulerian
approach is frequently applied to gassolid ow and specically
to circulating uidized beds modeling of small and medium scale
facilities [612]. Due to tremendous computational resources
required to simulate large industrial facilities, there are only a
few reports of this modeling approach in the literature [13]. In
the hybrid EulerianLagrangian approach groups of particles
(parcels) are tracked in a Lagrangian frame of reference. The particleparticle interactions are not simulated directly, as in discrete
element method (DEM), but are accounted for by means of KTGF
in the Eulerian frame, where the particles are projected. An
important feature of the approach is that the particle size distribution is naturally incorporated in the model. The hybrid Eulerian
Lagrangian approach is used in this study to simulate gasication
of coal in a circulating uidized bed reactor. The coupling between
the complex hydrodynamics of the uidized bed and the chemical
reactions is realized within the commercial code ANSYS FLUENT
14.0. The EulerianLagrangian model in this code is called Dense
Discrete Phase Model (DDPM). The DDPM approach is still under
development and just a few studies applying this method can be
found in the literature [1418]. In this study we apply it for gasication modeling, where the mechanisms of heterogeneous and
homogeneous reactions are included. An alternative Lagrangian
approach is the multi-phase-particle-in-cell (MP-PIC) method
[1921], which has recently been also applied to modeling of coal
gasication in a large bubbling bed reactor [22].

0.97

0.38

coal
inlet

0.2
gasifying
agent inlet
Fig. 1. Geometry of the model (dimensions in meters).

introduced back to the particle equation of motion as a source.


The continuity, momentum, energy and species transport equations for the gaseous phase f are given by Eqs. (1)(4)




@ 
ef qf r  ef qf uf Smas
@t



@ 
ef qf uf r  ef qf uf uf ef rp r  sf ef qf g
@t


K DPM us  uf Smom




@
@p
ef qf hf r  ef qf uf hf ef sf : ruf  r  qf
@t
@t
"
#
m
X
 r  ef
hf ;k Jk Sf ;rad Sf ;rea Sen

k1




@ 
ef qf Y f ;k r  ef qf uf Y f ;k r  ef Jk ef Rf ;k Ssp
@t

where q is the density, e is the volume fraction, u is the velocity vector, p is the pressure, sf is the shear stress tensor, g is the gravitational acceleration vector, s is the solid phase index and K DPM is
the interphase exchange coefcient due to drag calculated for solids
volume fraction in the Eulerian frame. h is the specic enthalpy, k is
the species index, Y f ;k stands for the mass fraction of species k; m is
the number of species in the gaseous phase, Jk is the diffusion ux of
species k and q is the heat ux. The source term Sf ;rad accounts for
the contribution to the energy equation due to radiation and Sf ;rea
denes the energy released in homogeneous reactions. The Rf ;k
source represents the net rate of production of homogeneous species k. The Smas ; Smom ; Sen and Ssp are sources due to exchange of
mass, momentum, energy and species between the continuous
phases and the discrete phase, respectively. The energy source Sen
includes the enthalpy transfer due to convection, chemical reactions
and radiation between the phases. The momentum source term
Smom determines the exchange of momentum between the phases.
The set of Eqs. (1)(4) is supplemented with the particle equation
of motion given by Eq. (5). The particle equation of motion which

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

equates the particle inertia with the forces acting on a particle,


reads

gqp  qf rp r  rs
dup
F D uf  up


aother
dt
qp
qp
qp

where subscript p denotes the particle properties in the Lagrangian


frame, qp is the particle density, rs is the solids stress tensor due to
particleparticle interactions predicted by the KTGF and F D uf  us
is the particle acceleration due to drag. Both the solid stress tensor
as well as the drag coefcient F D are calculated in the Eulerian
frame. The term rp=qp denes the particle acceleration due to
pressure gradients and aother represent acceleration due to additional forces that can be included, e.g. virtual mass force, lift force,
etc. It should be stressed here that the acceleration due to pressure
gradients and due to additional forces contribute to the source term
Smom in the momentum Eq. (2). Furthermore, the drag coefcient F D
and the interphase exchange coefcient K DPM are determined from
the same drag law in the Eulerian grid.
Application of the KTGF for the particleparticle interactions
requires many closure models which are implemented in the
ANSYS FLUENT code. These closures are presented in detail in the
software documentation ([28]) and have been widely described
in the literature ([10,7,12]), therefore they are not presented here.
Instead the submodels used within this study are presented in
Table 1 and for each model references to the literature are given.
The Standard k model with mixture option ([28]) was used as
the turbulence model and the Discrete Ordinates ([28]) model with
weighted sum of gray gases model for the gas absorption coefcient was used to take into account the radiative heat transfer.
2.3. Gasication modeling
A step by step procedure is applied in the coal utilization processes. The injected coal particles are rst heated up and dried
until all moisture is removed. Then devolatilization starts which
is described by a constant rate model. This is followed by multiple
surface reactions on the remaining char. The composition of volatiles is known from coal analysis and it is assumed that it does not
change during the devolatilization. A VOL pseudo-species is introduced which is released in the devolatilization process and decomposes quickly afterwards in a fast volumetric reaction to form the
nal mixture of species forming the known from measurements
volatiles, as shown in Eq. (6)

VOL ! aCH4 bCO2 cCO dH2 eH2 O f TAR

The product species can take part in homogeneous and heterogeneous reactions. Additionally to the VOL, a pseudospecies called
TAR is introduced to the gas phase mixture. TAR is composed of

Table 1
Closure models and parameters used in the simulations.
Granular viscosity
Granular bulk viscosity
Frictional viscosity
Angle of internal friction
Frictional pressure
Frictional modulus
Friction packing limit
Granular temperature
Solids pressure
Radial distribution
Elasticity modulus
Packing limit
Drag model
Restitution coefcient
Normal discrete phase reection coefcient
Normal discrete phase reection coefcient

Gidaspow et al. [29]


Lun et al. [2]
Schaeffer [30]
30
Based on KTGF, [28]
Derived, [28]
0.61
Algebraic equation, [28]
Lun et al. [2]
Lun et al. [2]
Derived, [28]
0.63
Gidaspow et al. [29]
0.9
0.8
0.5

Table 2
Measured and included in the simulation volatiles species.
Measured

Mole fraction

Included

Mole fraction

CH4
CO2
CO
H2
H2 O
H2 S
NH3
C2 H6
C6 H6
C10 H8

0.176
0.036
0.077
0.385
0.172
0.012
0.047
0.043
0.040
0.012

CH4
CO2
CO
H2
H2 O
TAR

0.176
0.036
0.077
0.385
0.172
0.154

devolatilization products known from coal analysis, whose amount


is relatively small and do not contribute to the main species
concentrations substantially. Such an approach has been utilized
to reduce the number of transported species in the complex
unsteady multiphase reacting ow in the reactor. It should be
stressed that the contribution of the minor species can be easily
included in further analyzes. The measured and included in the simulations devolatilization product species are presented in Table 2.
Further species taken into account in the simulation are O2 and
N2 . The volumetric reactions are modeled by means of nite rate/
eddy dissipation model ([28]) with a general rate expression for
the nite rate model of the form
n
Y
RV A expT a =T ci gi

i1

where A is the preexponential factor, T a is the activation temperature, n is the number of species affecting the reaction, ci  is the
3
molar concentration of species i expressed in mol=m and gi is
the reaction order with respect to i. The constants for the volumetric reactions are given in Table 3, where the volumetric reaction rate
RV is given in kmol/m3s.
The reaction rates are taken from ANSYS FLUENT database [28]
besides the watergas shift reaction which is taken from [31]. The
char surface reaction rates are calculated using the user dened
function (UDF) mechanism from general rate expression given by

RS AT b expT a =T  a

n
Y

ci gi mp X ch

i1

where mp is mass of coal contained in a grid cell volume, X ch is the


char mass fraction. The constants for the surface reactions are given
in Table 4, where the reaction rate RS is given in kg/s. For reaction 1
in Table 4 the coal mass mp in Eq. (9) shall be replaced with the ratio
mp =qp , where qp is the material density of the solids (coal). The rate
expressions for the heterogeneous reaction 1 is taken from [32] and
reactions 2 through 7 are taken from [33]. Since the reactions and
their rates in the selected reaction set come from various literature
sources, they do not comprise a consistent reaction mechanism. An
appropriate approach would be to use a reduced mechanism based
on global reactions with rate expressions tted by means of measured data. Such a reduced mechanism would be appropriate for a
given system and conditions, therefore its applicability would be
limited. On the other hand the importance of reaction kinetics is
increasing in states far from equilibrium, usually fast changing
and low residence times. It is expected that the outlet gas composition is close to equilibrium and the inconsistent reaction mechanism will have a minor contribution. It should be stressed here
that much more detailed and complex coal devolatilization, char
conversion and homogeneous reactions approaches exist and have
already been applied to coal gasication modeling. An example of
such analysis with application to entrained ow gasication modeling is given in [34,35]. Such an approach at the current stage of

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

Table 3
Rate constants of the considered homogeneous reactions.

1
2
3
4
5

Reaction

Ta, K

g1

g2

g3

CO H2 O ! CO2 H2
CO2 H2 ! CO H2 O
CO 1=2O2 H2 O ! CO2 H2 O
H2 1=2O2 ! H2 O
CH4 2O2 ! CO2 2H2 O

2.75e+6
1.04e+8
2.24e+12
9.87e+8
2.12e+11

10,055
14,010
20,086
3728
24,380

0.5
1
1
1
0.2

1
0.5
0.25
1
1.3

0.5

Table 4
Rate constants of the considered heterogeneous reactions.

1
2
3
4
5
6
7

Reaction

A, units vary

Ta, K

g1

g2

C 1=2O2 ! CO
C CO2 ! 2CO
2CO ! C CO2
C H2 O ! CO H2
CO H2 ! C H2 O
C 2H2 ! CH4
CH4 ! C 2H2

1.762e+6
76.31
6.262e3
76.31
6.262e3
8.206e2
9.058

13,587
22,645
2363
22,645
6319
8078
13,578

0
0
20.92
0
17.29
7.087
0.372

1
1
2
1
2
1
0.5

1
1
2
1
1
1
0.5

Table 5
Input data used as boundary conditions.

Gasifying agent
Coal owrate, kg/h
Coal inlet temperature, K
Air owrate, kg/h
Steam owrate, kg/h
Gasifying agent inlet temperature, K
Coal (ar)
HHV, MJ/kg
Char, %
Volatiles, %
Ash, %
Moisture, %

Case 1

Case 2

Air
171
288
193

520

Air/steam
181
288
233
18.3
510

28.1
48.0
34.1
12.6
5.3

Fig. 3. Instantaneous (left) and mean (right) solids volume fraction.

28.1
55.1
27.7
10.6
6.6

m/s

m/s

Fig. 4. Instantaneous (left) and mean (right) gas phase velocity magnitude.
Fig. 2. Measured and RosinRammler t of the PSD.

model development could not be used here due to associated long


computational times.

about a stable mean, data are collected for computing ow parameter averages.
3. Analyzed cases

2.4. Solution procedure


The governing equations are solved as unsteady using phase
coupled simple algorithm. The solution procedure is started with
a gas ow only and the coal particles are later subsequently
injected. As the outlet ow variables and mass of coal in the reactor
achieve a pseudo-steady state in which the ow parameters vary

Two case studies have been analyzed in which coal was gasied
in air (case 1) and air/steam mixture (case 2). As mentioned earlier
two meshes were used in the simulations. The mesh independence
has been done for both analyzed cases. In Table 5 the input data for
coal and gasifying agent used in the simulations are presented. The
input data, as well as other experimental results compared later

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

4. Results and discussion

K
Fig. 5. Instantaneous (left) and mean (right) gas phase temperature.

with the simulations, have been obtained from the Institute for
Chemical Processing of Coal in Poland [23]. The details regarding
the experimental facility and measurement procedure can be
found in [26].
Coal particles size distribution has been measured and was
tted by means of the RosinRammler distribution

 n
Y d expd=d

 is the size constant, n is the


where d is the particle diameter, d
distribution parameter and Y d is the mass fraction of particles of
diameter greater than d. The size constant and distribution parameter were determined by tting the measured PSD. The constants
 1:197 mm and n 0:895. With this information the PSD
are: d
of particles can be specied at the boundary for injected coal particles. In Fig. 2 the measured and tted particle size distribution is
presented.

CO

As mentioned earlier the simulations were run unsteady until a


pseudo-steady state was obtained. After this point collection of
data for obtaining averages of ow parameters started. The averaging times in the analyzed cases were as long as 90 s of real time
ow, however it was noticed that averaging times as short as
2030 s were required to obtain meaningful means. In Figs. 37
sample results of the obtained ow eld variables are presented.
As can be seen in Fig. 3 the highest concentration of solids is at
the bottom in the barrel like part of the reactor. Only a small
amount of solids can be found in the riser tube and some of the solids are trapped in the expanding section at the very top. It was
observed that most of coal particles remain at the bottom where
internal recirculation of solids was visible. As the particles are
sucked to the riser section they will leave the reactor due to high
velocity in this part (cf. Fig. 4 or a small portion of them will be
trapped in the expansion. It was observed that some of the particles remained inside the reactor for very long time and their residence times were of the order of simulation time. In Fig. 5 the
temperature of the gas phase is presented. As can be seen the predicted gas phase temperature is relatively low at the bottom part of
the reactor and increases to approximately 1150 K at the top,
nonetheless all oxygen is consumed immediately after entering
the reactor, cf. Fig. 7. In Figs. 6 and 7 the gas phase species mass
fractions have been presented. As can be seen carbon monoxide
is formed in the dense bottom part of the bed, as could be
expected, and then is partially converted to carbon dioxide through
the combustion reaction, due to still available oxygen at the bottom, and water gas shift reaction, due to available steam and
increasing temperature. The highest concentrations of methane,
hydrogen and steam occur in the regions of highest devolatilization intensity, specically in the center of the reactor where the
injected coal particles stream passes through.
In Figs. 8 and 9 results of measurements and simulations for air
and air/steam gasication have been compared, respectively. The
measurements are given with error bars for O2 ; H2 ; CO;CO2 and
CH4 . The measurement error is not available for tar and since N2
was used to close the mass balance, errors for these two species
are not presented. As can be seen in general good agreement
between the data has been obtained. It should be however stressed
that our rst simulations led to results presented in 8 and named
simulation for which CO and CO2 mole fractions depart considerably

CO2

CH4

Fig. 6. Mass fraction distribution of gaseous species: CO, CO2 and CH4 .

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

H2

O2

H 2O

Fig. 7. Mass fraction distribution of gaseous species: H2 ; H2 O and O2 .

from the measured data. This is attributed to the application of the


nite rate/eddy dissipation model for which the reaction rates are
predicted by both, kinetic rate and turbulent rate. The nal rate is
the smaller of the two. This leads to departure from expected close
to equilibrium conditions at the outlet of the reactor. To obtain
appropriate CO=CO2 ratio at the outlet the water gas shift reaction
rate has been articially increased by a factor of 102 , which lead
to the results presented in Figs. 8 and 9 named simulation modied
wgs. As the model will be further developed it is planned to use a
different approach to properly predict the conditions at the gasier
outlet. The results presented in Figs. 37 correspond to the air
gasication with the modied water gas shift reaction rate. As
can be seen the predicted TAR pseudo-species mass fraction is
higher than the measured values in both analyzed cases. The reason
for that may be the fact that the released TAR could react with oxygen only. Since in the upper part of the reactor there is no oxygen
available and TAR cracking reaction has not been implemented to
the model, almost no conversion of the TAR occurs. Furthermore
it can be also observed that the mole fractions of nitrogen predicted
in case simulation and simulation modied wgs of Fig. 8 are different although these are the same cases for which, with the same carbon conversion, nitrogen mole fractions should be the same. The

carbon conversions are however different. It was observed during


the simulations that the relatively large time step used in the simulations (Dt 0:02 s) is too large for the fastest heterogeneous
reaction of partial oxidation of carbon (reaction 1 in Table 4). In
the solution procedure the rate of heterogeneous reactions is
evaluated and maintained throughout the time step. This can lead
to situations in which all substrates are consumed within the time
step and more products are formed than possible. In case of partial
oxidation of carbon it is possible that more CO is formed than available O2 would allow. This leads to increased concentration of CO,
decreased concentration of N2 and higher carbon conversion. For
the case named simulation the carbon conversion, dened as the
ratio of carbon mass in the gas to carbon mass supplied in fuel
was c 0:53. To overcome this problem a smaller time step could
be used. This would however lead to a substantial increase of the
overall simulation time. Another possibility is to control the
reaction rate based on the amount of available oxygen in the cell
volume within a time step. If the calculated reaction rate, within
a time step, would lead to consumption of more oxygen than
available, the reaction rate is reduced to a level corresponding to
available oxygen in a single computational cell. Such a control procedure has been implemented to the CFD code by means of the user

0.6

0.6
experiment
simulation
simulationmodifiedwgs

0.5

0.5
0.4

0.4
0.3

0.3

0.2

0.2

0.1

0.1

experiment
simulationmodifiedwgs

N2

O2

H2

CO

CO2

CH4

tar

Fig. 8. Comparison of measured and calculated syngas composition for air


gasication.

N2

O2

H2

CO

CO2

CH4

tar

Fig. 9. Comparison of measured and calculated syngas composition for air/steam


gasication.

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

A. Klimanek et al. / Fuel xxx (2014) xxxxxx

dened function mechanism and applied in cases named simulation


modied wgs of Figs. 8 and 9. The carbon conversions for air gasication (case simulation modied wgs) was c 0:35 and for steam
gasication was c 0:41. It should be stressed that in some regions
of the simulation domain the reaction rate for carbon partial oxidation is smaller than that resulting from the chemical kinetics, which
can lead to slight extension of the oxidation zone at the bottom of
the reactor visible in Fig. 7. Application of this procedure led to
prediction of nitrogen mole fraction close to the measured value
for air gasication. For air/steam gasication the predicted N2 mole
fraction is slightly higher than measured which suggests smaller
carbon conversion than in the experiment. This issue will be examined in our future work.
5. Conclusions
CFD model of coal gasication in a uidized bed reactor has
been presented in the study. The model development can be
divided into to two parts. First part is devoted to appropriate modeling of the uidized bed hydrodynamics and the second to the
reacting solid and gaseous phases. Coupling of the two parts is
challenging and essential to obtain meaningful results. The Dense
Discrete Phase Model of ANSYS FLUENT has been used to simulate
the ow of the particulate phase in the coal gasier. Kinetic Theory
of Granular Flow has been utilized to model the particleparticle
interactions. The heterogeneous reactions on coal particles surface
are modeled using the nite rate chemistry (multiple surface reactions). Homogenous gas phase reactions have been modeled using
the nite rate/eddy dissipation approach which led to departure of
the gas composition from the measured values. This problem has
been overcome by articial increase of the water gas shift reaction
rate, however a different approach to the gas phase chemistry
should be applied. It has been also observed that the applied time
step led to wrong prediction of carbon conversion and thus different than measured nitrogen and carbon monoxide mole fractions.
This issue has been addressed by application of a control procedure, which allowed avoiding this behavior in case of air gasication. The difference for the case of air/steam gasication are still
visible. Since the model is under development these problems
should be addressed in our future research.
Acknowledgements
The investigations have been supported by the National Center
for Research and Development (NCBiR) as a research project
Developing a technology of coal gasication for highly efcient
production of fuels and electric power, CzTB 5.2. The support is
gratefully acknowledged.
The research have been supported by Research and Development Strategic Program Advanced Technologies for Energy Generation project No.2 Oxy-combustion technology for PC and FBC
boilers with CO2 capture, supported by the National Centre for
Research and Development, agreement No. SP/E/2/66420/10
References
[1] Myhnen K, Hyppnen T. A three-dimensional model frame for modelling
combustion and gasication in circulating uidized bed furnaces. Int J Chem
React Eng 2011;9(1):154280.
[2] Lun C, Savage S, Jeffrey D, Chepurniy N. Kinetic theories for granular ow:
inelastic particles in Couette ow and slightly inelastic particles in a general
ow eld. J Fluid Mech 1984;140:22356.
[3] Gidaspow D. Multiphase ow and uidization. Boston, MA: Academic Press;
1994.
[4] Marchisio D, Virgil R, Fox R. Quadrature method of moments for aggregationbreakage processes. J Colloid Interface Sci 2003;258:32234.

[5] Fan R, Marchisio D, Fox R. Application of the direct quadrature method of


moments to poly-disperse gas-solid uidized beds. Powder Technol
2004;139:720.
[6] Kallio S, Airaksinen J, Gulden M, Hermanson A, Peltola J, Ritvanen J, Seppala M,
Srujal S, Taivassalo V, Experimental and numerical study of hydrodynamics in
a circulating uidized bed. In: Proc of Finnish-Swedish Flame DaysElectronic
version.
[7] Shah S, Ritvanen J, Hyppnen T, Kallio S. Space averaging on a gas-solid drag
model for numerical simulations of a CFB riser. Powder Technol
2012;218:1319.
[8] Hartge E, Ratschow L, Wischnewski R, Werther J. CFD-simulation of a
circulating uidized bed riser. Particuology 2009;7:28396.
[9] Zhang N, Lu B, Wang W, Li J. Virtual experimentation through 3D full-loop
simulation of a circulating uidized bed. Particuology 2008;6:52939.
[10] Almuttahar A, Taghipour F. Computational uid dynamics of a circulating
uidized bed under various uidization conditions. Chem Eng Sci
2008;63:1696709.
[11] Almuttahar A, Taghipour F. Computational uid dynamics of high density
circulating uidized bed riser: study of modeling parameters. Powder Technol
2008;185:1123.
[12] Cloete S, Amini S, Johansen S. A ne resolution parametric study on the
numerical simulation of gas-solid ows in a periodic riser section. Powder
Technol 2011;205:10311.
[13] Zhang N, Lua B, Wang W, Li J. 3D CFD simulation of hydrodynamics of a
150 MWe circulating uidized bed boiler. Chem Eng J 2010;162:8218.
[14] Popoff B, Braun M. A lagrangian approach to dense particulate ows. In:
International conference on multiphase ow, Leipzig, Germany.
[15] Cloete S, Johansen S, Braun M, Popoff B, Amini S. Evaluation of a lagrangian
discrete phase modeling approach for resolving cluster formation in CFB risers.
In: 7th international conference on multiphase ow, Tampa, FL, USA.
[16] Cloete S, Johansen S, Amini S. Performance evaluation of a complete lagrangian
KTGF approach for dilute granular ow modelling. Powder Technol.
2012;226:4352.
[17] Adamczyk W, Klimanek A, Biaecki R, Wecel G, Kozoub P, Czakiert T.
Comparison of the standard euler-euler and hybrid euler-lagrange
approaches for modeling particle transport in a pilot-scale circulating
uidized bed, Particuology, in press, available online <http://dx.doi.org/
10.1016/j.partic.2013.06.008>.
[18] Adamczyk W, Wecel G, Klajny M, Kozoub P, Klimanek A, Biaecki R, Modeling
of particle transport and combustion phenomena in a large-scale circulating
uidized bed boiler using a hybrid euler-lagrange approach, Particuology, in
press, available online <http://dx.doi.org/10.1016/j.partic.2013.10.007>.
[19] Andrews M, ORourke P. The multiuid particle-in-cell (MP-PIC) method for
dense particulate ows. Int J Multiphase Flow 1996;22(2):379402.
[20] Snider D. An incompressible threedimensional multiphase particle-in-cell
model for dense particle ows. J Comput Phys 2001;170:52349.
[21] ORourke P, Zhao P, Snider D. A model for collisional exchange in gas/liquid/
solid uidized beds. Chem Eng Sci 2009;64:178497.
[22] Snider D, Clark S, ORourke P. Eulerianlagrangian method for threedimensional thermal reacting ow with application to coal gasiers. Chem
Eng Sci 2011;66:128595.
[23] Institute of Chemical Processing of Coal, <www.ichpw.zabrze.pl>.
_
[24] Chmielniak T, Popowicz J. Fluidalne zgazowanie wegla przy uzyciu
co2 dla
produkcji gazu syntezowego. Chemik 2013;67(5):41522 (in polish).
[25] Chmielniak T, Sciazko M, Tomaszewicz G, Tomaszewicz M. Pressurized CO2 enhanced gasication of coal, J Therm Anal Calorimetry, published online:
<http://dx.doi.org/10.1007/s10973-014-3879-z>.
[26] Sciazko M, Sobolewski A, Tomaszewicz G, Czaplicki A, Sowik K, Tomaszewicz
M. Co2 -enhanced coal gasication in a circulating uidized bed
reactor. Beijing, China: Institute of Process Engineering, Chinese Academy of
Sciences; 2014. p. 0705..
_
[27] Sciazko
M. Studium aerodynamiki cyrkulacyjnego reaktora uidalnego w
szczeglnosci do pirolizy wegla, Wydawnictwo Politechniki Slaskiej, CHEMIA
Z. 143, Gliwice, Poland; 2001 [in polish].
[28] ANSYS, Inc. ANSYS FLUENT 14.0 Documentation, 2011, Software
documentation.
[29] Gidaspow D, Bezburuah R, Ding J. Hydrodynamics of circulating uidized beds,
kinetic theory approach. In: Fluidization VII, proceedings of the 7th
engineering foundation conference on uidization; 1992. p. 7582.
[30] Schaeffer D. Instability in the evolution equations describing incompressible
granular ow. J Differ Eq 1987;66:1955.
[31] Yan L, He B, Pei X, Li X, Wang C, Liang H. Kinetic model and prediction for coal
hydrogasication. Int J Hydrogen Energy 2013;38:451323.
[32] Yoon H, Wei J, Denn M. A model for moving-bed coal gasication reactors.
AIChE J 1978;24:885903.
[33] Syamlal M, Bisset L. METC gasier advanced simulation (MGAS) model DOE/
METC-92/4108 (DE92 001111).
[34] Vascellari M, Arora R, Pollack M, Hasse C. Simulation of entrained ow
gasication with advanced coal conversion submodels. Part 1: Pyrolysis. Fuel
2013;113:65469.
[35] Vascellari M, Arora R, Hasse C. Simulation of entrained ow gasication with
advanced coal conversion submodels. Part 2: Char conversion. Fuel
2014;118:36984.

Please cite this article in press as: Klimanek A et al. Towards a hybrid EulerianLagrangian CFD modeling of coal gasication in a circulating uidized bed
reactor. Fuel (2014), http://dx.doi.org/10.1016/j.fuel.2014.10.058

You might also like