You are on page 1of 10

Chemical Engineering Journal 174 (2011) 341–350

Contents lists available at ScienceDirect

Chemical Engineering Journal


journal homepage: www.elsevier.com/locate/cej

CPFD simulation of a fast fluidized bed steam coal gasifier feeding section
Alireza Abbasi a,b , Paul E. Ege b , Hugo I. de Lasa a,∗
a
Department of Chemical and Biochemical Engineering, University of Western Ontario, London, ON, Canada N6A 5B9
b
Reactech Process Development Inc., Markham, ON, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Coal gasification in fluidized beds is a process that can be strongly influenced by gas–solid suspension
Received 20 February 2011 flow patterns. Fast fluidized beds can be designed to maximize coal gasification yields providing an
Received in revised form 2 June 2011 optimum syngas composition with minimum operational upsets. In order to accomplish this, a com-
Accepted 8 July 2011
prehensive model for steam coal gasification is valuable. With this end, a 2D CPFD (computational
particle fluid dynamics) dynamic model is considered in the present study. The proposed model uses
Keywords:
a Lagrangian–Eulerian approach and describes the flow patterns in the gasifier feeding section, as well as
Coal gasification
the local particle velocities, particle solid fractions and gas composition. As a reference and for compari-
Fluidized beds
Computational particle fluid dynamics
son, an ideal PFR (plug flow reactor) model is also considered. It is found that compositions from the 2D
Suspension chocking CPFD model provide close gas compositions with respect to the PFR. It is observed that the 2D CPFD model
is valuable for predicting particle flows in the feeding near region keeping it unaffected by fast fluidized
bed upsets such as suspension chocking. It is also found that the 2D CPFD model is particularly useful for
describing fast fluidized beds operation involving significant amounts of rich ash recycled feeds.
© 2011 Published by Elsevier B.V.

1. Introduction of the combined reactor hydrodynamics and kinetics [6–9]. To


address these simulation issues two approaches have been applied:
Fluidized beds have an important role in many industrial the Eulerian–Lagrangian approach and the Eulerian–Eulerian
processes including petrochemical, pharmaceutical, and mineral approach. In the first approach, the fluid phase is treated as a contin-
industries. A most valuable area of potential applications for flu- uum by solving the time-averaged Navier–Stokes equations, while
idized beds is in coal combustion and coal gasification for power the dispersed phase is accounted for by tracking a large number
generation, chemical production, and synthesis gas. Fast fluid of particles through the calculated flow field. The dispersed parti-
beds also known as “transport line reactors”. These reactors have cle phase can exchange momentum, mass, and energy with the gas
attracted the attention of researchers since the 1970s [1–3]. Fast phase. In the dispersed particle phase, each particle is affected in
fluidized bed coal gasifiers have the potential of processing large its trajectory by three-dimensional forces [10].
coal particle flow in compact units of low diameter/length ratios. In the Eulerian–Eulerian approach, the different phases are
Traditionally, two two-phase models have been proposed to treated mathematically as interpenetrating continua. Since the vol-
describe the performance of dense fluidized bed reactors [4]. The ume of a phase cannot be occupied by the other phases and as a
two-phase model includes a dilute phase and a dense emulsion [5]. result the concept of phasic volume fraction is introduced. These
Each phase has separate governing equations plus a term in each volume fractions are assumed to be continuous functions of space
equation describing mass interchange between the two phases and time and their sum is equal to one. Conservation equations for
[4]. each phase are then derived obtaining a set of equations, which
These models while being suitable for dense phase fluidized have similar structure for all phases. These equations are closed by
beds require significant changes and adaptations when dealing providing constitutive relations that are obtained from empirical
with fast fluidized beds. In fast fluidized beds the existence of information or, in the case of article by application of kinetic the-
phases has to be accounted in a very different manner with ory [11]. The Eulerian–Eulerian approach has wide application in
interpenetrating gas and particulate phases. In this respect, CFD multiphase flow. It has however, important limitations such as not
(computational fluid dynamics) and CPFD (computational parti- being able to account for particle size distribution and/or the shear
cle fluid dynamics) modeling provide valuable tools for simulation stress for the particles.
Given the above and the required simulation features of
accounting for particle size distribution at various stages of the
∗ Corresponding author. Tel.: +1 519 661 2144; fax: +1 519 661 3498. fast coal gasification process, we select for the present study, an
E-mail address: hdelasa@eng.uwo.ca (H.I. de Lasa). Eulerian–Lagrangian approach. This model captures the needed

1385-8947/$ – see front matter © 2011 Published by Elsevier B.V.


doi:10.1016/j.cej.2011.07.085
342 A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350

description of particle–fluid interactions, particle–particle interac- The trajectory calculation of the discrete phase is made by inte-
tions and coal and unconverted coal particle size distribution. Each grating the force balance on the particles. The particle motion as a
phase in this model has its own set of carefully selected governing result is defined as follows:
hydrodynamic equations. d ∇ P ∇ · p
The CPFD and other simplified models simulations of the present (vp ) = Dp (vf − vp ) − − +g (5)
dt p p p
study show that product yields are not too affected by fluid bed
hydrodynamics. Moreover, it is also proven that fast fluid bed where vp is the particle velocity, p is the particle density,  p is the
operation requires good predictions of the choking of the parti- particle volume fraction, and  p is the particle normal stress. Dp is
cle suspension and it is in this area where the CPFD simulation the interphase drag coefficient defined as follows:
 
3 f vf − vp 
becomes very essential. The proposed fluid dynamic approach of
this study is most appropriate for fast fluidized beds being consis- Dp = Cd (6)
tent with the earlier works [1,3,12]. This model also contains special 8 p (3Vp /4)1/3
description (conditions/closure equations) for fast fluidized beds Cd represents the drag coefficients and depends on the Reynolds
being very essential to accomplish flow regime diagram, particle number.
size distribution, and component concentration profiles.
24
Re < 1000Cd = (1 + 0.15Re0.687 )f−2.65 ,
Re
2. The CPFD mathematical model Re ≥ 1000Cd = 0.44ε−2.65 (7)
f

Using the Eulerian–Lagrangian approach, the fluid phase is


 
described as a continuum by solving the Navier–Stokes equations, 2f vf − vp   3Vp 1/3
whereas the dispersed phase is treated by tracking a large number Re = (8)
f 4
of particles through the calculated flow field. The dispersed phase
is coupled with the continuum phase able to exchange momen- where f is the gas viscosity and Vp is the particle volume. The
tum, mass and energy. In the computational particle fluid dynamics proposed drag coefficient has been shown to have a good agree-
(CPFD) scheme each particle has three-dimensional forces exerted ment with experimental results, indicating that the cluster-based
on them such as fluid drag, gravity, static–dynamic friction, particle drag coefficient model is suitable to describe various statuses in
collision and other forces. Using this method, the collision force on fast-fluidized beds [15].
each particle is modeled as a spatial gradient [13]. However, the Particle-to-particle collisions are modeled by the particle nor-
direct element method (DEM) considers particle-to-particle inter- mal stress,  p . The particle normal stress model used here is from
actions using a spring–damper model and direct particle contact. [13]:
While using the CPFD method a numerical particle is accounted for
via an ensemble of particles displaying the same properties such Ps P
p = (9)
as chemical composition, size, and density. In contrary to the DEM max[cp − p , ε(1 − p )]
method, the dispersed phase in CPFD method is not limited and
where Ps is a positive constant that has units of the pressure,  cp
can vary from a low to a large solid volume fraction. In CPFD, the
is the particle volume fraction at close packing limit. The constant
numerical particle numerical approximation is established within
is recommended to be in between 2 ≤ ≤ 5[16]. The ε constant
a numerical control volume (domain) where the properties of the
is a small number on the order of 10−7 . Thus, the particle stress
fluid are considered constant. The motion of fluid and dispersed
defined by Eq. (9) depends only on the concentration of particles
phases are governed by respective mass and momentum conserva-
and neglects the size and velocity of particles. The proposed particle
tion equations. The volume averaged fluid mass, momentum and
normal stress model has been used successfully with a dependency
energy equations are defined as follows [14]:
on each solid’s velocity vector. The model applies the particle nor-
mal stress to a solid up to the point where the solid reaches the

(f f ) + ∇ · (f f vf ) = Sf (1) particle-mean velocity [9].
∂t

∂ 3. Simplified fast fluidized bed models


(f f vf ) + ∇ · (f f vf vf ) = −∇ P + ∇ ·  + f f g − F (2)
∂t
A continuous multiphase chemical reactor is a vessel where
∂ ∗ chemical reactions take place. The flow patterns in such a reac-
(f f E) + ∇ · [(f E + P)f vf ] = −∇ .q + Q (3)
∂t tor unit are rather complex and require the use of a computational
fluid dynamics approach as the one described in Section. However,
where  f is the fluid volume fraction, vf is the fluid velocity, f is a PFR (plug flow reactor) model offer the opportunity for the simu-
the fluid density, Sf is the fluid source term, P is the fluid pressure, lation of fast fluidized bed reactors as shown in the early studies of
 is the fluid stress tensor, and g is the gravitational acceleration. F de Lasa and Gau [2]. A material balance on the differential volume
represents the momentum exchange rate per volume between the of a fluid element on species i in a PFR can be expressed as follows
fluid and particle phase. q is the energy transfer due to conduction, [5]:

species diffusion and viscous dissipation. Q is the energy source d  vi,j Rj
term and the energy E is defined as follows: (Fi,x ) = A (10)
dx
r,j
  j
P vf vf
E= H− + (4) where Fix is the molar flow rate of species i at position x. vi,j and
f 2
Rj are the stoichiometric coefficient and the reaction rate for step j.
vr,j is the stoichiometric coefficient used as the basis for the Rj and
where H is the enthalpy for ideal gases, and is written as a summa-
 A is the reactor cross-section.
tion of the mass fractions weighted species’ enthalpy. H = Yi Hi The mathematical solution of this relatively simple model
i involves a set of ordinary differential equations that can usually be
A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350 343

Table 1
Reactions considered in the coal gasification model.

Reaction name Chemical reaction Reaction rate (mol/m3 ·s) × 106


 E
 εs s Xsl PH PCO
Steam gasification [17,18] C + H2 O ↔ CO + H2 R1 = k1 exp − RT1g 12
(PH2 O − PH∗ ), PH∗ = 2
2O 2O exp(17.29−16326/Tg )

E

 εs s Xsl ∗ ∗
P2
CO2 gasification [17,18] C + CO2 ↔ 2CO R2 = k2 exp − RT2g (PCO2 − PCO ), PCO = CO

 
12 2 2 exp(20.92−20282/Tg )
ε  X
Methanation [17,18] 1
2
C + H2 ↔ 1
2
CH4 R3 = exp −7.087 − 8078 s s sl
Tg 12
(PH2 − PH∗ )
 0.5
2

PCH
PH∗ = 4
exp(−13.43+10999/Tg )
2
 E

Carbon combustion [19] 2C + O2 → 2CO R4 = k4 s exp − RT4g PO2
 E5
0.5
 E
0.5
Water gas-shift [20] CO + H2 O ↔ CO2 + H2 R5 = k5 exp − RTg [CO] [H2 O] − k6 exp − RT6g [H2 ] [CO2 ]

solved in a quite straightforward manner even for complex reaction the gas phase. This approximation is considered adequate for the
kinetics. The numerical solutions of Eq. (10) or a set of equations small particles of the present study given the difference, between
such as Eq. (10) offers the opportunity of a simulation that repre- gas temperature and solid temperature is negligible [17].
sents the limiting (ideal) behaviour for a fast fluidized bed unit. Due to the potential variability of the volatiles composition,
there is no standard stoichiometric for the steam gasification of
4. Reaction kinetics volatiles. In the present work, a so called “non-stoichiometric
approach” is used where the volatile matter with an elemental
Description of chemical changes in a coal gasifier normally formula Cx Hy Oz reacts with steam as follows:
requires the consideration of a complex reaction network. One pos-
Volatile + ˇH2 O → ˛1 CH4 + ˛2 H2 + ˛3 CO + ˛4 CO2 (14)
sible approach is to assume that the pyrolytic process of the raw
coal conversion is complete at conditions very close to the gasifier where ˇ, ˛1 , ˛2 , ˛3 , ˛4 coefficients are adjusted to comply with
entry. the hydrogen, oxygen and carbon elemental balances and chemical
equilibrium for water gas shift and methane steam reforming as
Coal → Char + Volatile + H2 O + Ash (12)
proposed by Salaices et al. [22].
with the sum of the mass fraction of products is calculated to be It is further assumed in the proposed model that the sulfur
unity in the raw coal [21]. content in the coal is small and therefore H2 S is not affecting sig-
nificantly either the product composition or the enthalpy balance
YChar + YVolatile + YH2 O + YAsh = 1 (13) in the gasifier.
Thus, char is formed from coal particles and the continue equa-
tion of solid phase can ensure the mass balance of char. Following 5. Enthalpy balances
this, the unconverted coal or fixed carbon fraction of the coal, des-
ignated as char, reacts with steam via the reactions reported in The enthalpy balance is a key and complementary assessment
Tables 1 and 2. Main reactions considered in this network and tak- for the simulation of the gasifier operation. A valuable approach to
ing place after pyrolysis or contributing in Eq. (12) are: (a) steam perform enthalpy balances is the use of Van’t Hoff box approxima-
gasification, (b) CO2 gasification, (c) methanation, (d) carbon com- tion [5] involving the following:

Tref
bustion and (e) water gas-shift reaction. It is assumed that all
reactions except the water gas-shift are heterogeneous reactions H1 = Fi,x Cp,i dT (16)
Tin
taking place throughout the entire volume of the char particles. i
Methanation is considered as well to be a slow reaction [17]. 
Tables 1 and 2 report various suggested reaction rates for these H2 = Hri Ri V (17)
reactions as well as their equilibrium constants. Regarding equi- i

librium constants for these set of reactions, they are in all cases 
Tr

very close to the recent values suggested by Salaices et al. [22]. H3 = Fi,x Cp,i dT (18)
Tref
Concerning kinetic rate parameters for these equations, the con- i
stants reported in Table 2 and selected for the calculations are with
the ones recommended for sub-bituminous coal steam gasification
[17,23]. Regarding the Arrhenius expression for the various gasi- H1 + H2 + H3 = Q (19)
fication reactions, this expression is based on the temperature in
where H1 , H3 specify the enthalpy changes for species coming
and leaving the control volume, V. H2 is the total heat of reaction
Table 2 in the control volume.
Coal gasification kinetics constants.
One should notice that in Eq. (18) all reactants and products
Kinetics constants Values species are considered as molecular species at Tref , which is the
k1 (1/atm s) 930
reference temperature for the heat of reaction (usually 25 ◦ C).
E1 (cal/mol) 45,000 Moreover an overall gasification stochiometry can established with
k2 (1/atm s) 930 the overall extent of gasification as follows:
E2 (cal/mol) 45,000
k4 (1/atm s) 1.79 × 106 Coal + ˇ H2 O → ˛ 1 CH4 + ˛ 2 H2 + ˛ 3 CO + ˛ 4 CO2 (20)
E4 (cal/mol) 27,000
k5 (1/atm s) 7.68 × 1013 where ˛ 1 , ˛ 2 , ˛ 3 , ˛ 4 and ˇ coefficients are defined at each position
E5 (cal/mol) 304.6 of the gasifier.
k6 (1/atm s) 6.4 × 1012 Finally and on this basis, the overall heat of reaction can calcu-
E6 (cal/mol) 326.4
lated with the enthalpy of formation of all species involved in Eq.
344 A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350

Table 3 Zone 1: an 18 cm air and steam mixing section where air and steam
The input parameters for the simulation.
are concurrently contacted at 1250 K and distributed uniformely
Computation particle per cell, Np 3 across the radius of the gasifier section, (b) Zone 2: a 25 cm recycled
Time step, t 5−4 s char feeding section where recycled particles with high ash content
Particle–wall normal retention coefficient, en 0.1
designated simply in the upcoming manuscript sections as “ash
Particle–wall tangential retention coefficient, et 0.99
Diffuse bounce, Df 0
rich” are fed, mixed and contacted with the raising steam and air,
Dimensionless constant of the solid-phase stress model, 3 (c) Zone 3: a 73 cm coal–char feeding section where coal is fed and
Dimensionless constant of the solid-phase stress model, ε 10−8 coal particles are gasified.
Pressure constant of the solid-phase stress model, Ps 1 Pa This configuration which is selected as the basis of the sim-
Solid volume fraction at closest packing,  cp 0.6
ulation of the present study is a modified version of gasifier
Gravitational acceleration, g −9.8
Particle feed per average volume, Nf 20,000 arrangement reported in [24] and recommended by the Cen-
Particle/fluid slip ratio, ς 1 tral Research Institute the Electrical Power Industry (CRIEPI). This
Maximum volume iterations, Iv 1 entrained fluidized bed with an expected total height of 5–7 m and
Volume residual, rv 10−6
an anticipated feeding section of 1.2 m height operates in a recircu-
Maximum pressure iterations, IP 2000
Pressure residual, rP 10−8
lating mode. Under this type of operation a cyclone separates the
Maximum velocity iterations, Iu 50 unconverted char (ash rich particles) from the gas and a down pipe
Velocity residual, ru 10−7 returning them to the bottom of the gasifier [23,24].
Maximum momentum redirection from collision, 40% Regarding the proposed simulation of the feeding gasifier sec-
Total number of clouds, Nc 231,052
tion there are a number of model assumptions: (a) air and steam are
distributed uniformly at the bottom of the reactor, (b) the ash rich
(20). Thus, given Eq. (19) and assuming that the gasifier is properly particles provide a good fraction of the heat required for coal gasi-
insulated and heat losses from the gasifier can be neglected (Q = 0) fication, (c) coal is fed to the reactor from a side port, (d) the bed is
the temperature at every axial position can be calculated using Eq. initially formed of a 97% N2 and 3% char (volume fraction) suspen-
(19). sion, (e) N2 is used to simulate the carrier gas assisting the feeding
the ash rich and coal particles, (f) solid particles are assumed to
6. System set-up and simulation parameters remain unchanged in size with this assumption considered reason-
able given the properties of sub-bituminous coals [25], (g) the coal
The coal gasification model was established in a computational particles are hypothesized to remain with the same density given
domain of 15 cm × 120 cm × 1 cm (Fig. 1). A grid of 18 × 120 × 1 is the relatively low coal conversion expected in the gasifier feeding
used to mesh the domain. The input parameters for the simula- section.
tion are reported in Table 3. This adopted dimensioning with a The operating conditions and stream fed to the gasifier unit used
total height of 1.20 m represents a possible feeding section con- in this simulation are reported in Table 4. The particle sizes were
figuration for a fluidizer entrained gasifier. It is in this section of chosen based on the coal dust particle size survey of underground
the gasifier where one could notice the early signs of suspension mines on Pittsburgh [26]. Converted char and ash size distribution
choking. particles are reported in Table 5.
One can observe in Fig. 1 that the feeding section of the gasi- Large eddy turbulence model was used in the simulation given
fier to be modeled is configured with three zones as follows: (a) this is an expected fluid dynamic pattern in fast fluidized beds. As

Fig. 1. (a) The schematic representation of the entrained fluidized bed gasifier. (b) Selected configuration for the simulation of the entrained fluidized bed gasifier in the near
feeding section.
A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350 345

Table 4
Operating conditions and stream fed to the Gasifier unit.

Inlet Mass fraction Temperature, K Pressure, MPa

Air + H2 O = 2.5 g/s 0.180 O2 , 0.550 N2 , 0.270 H2 O 1250 1.1


C + Ash + N2 = 100 g/sDesignated as “Rich ash” particle flow 0.005 C, 0.993 Ash, 0. 002 N2 Case 1 = 1150, Case 2 = 1350 1.1
Coal + N2 = 2.1 g/s 0.377 C, 0.311 volatile, 0.229 H2 O, 0.051 ash, 0.031 N2 298 1.1
Volatile 0.550 CH4 , 0.156 H2 , 0.220 CO, 0.073 CO2

well a quasi-second order upwind scheme was applied for convec-


tion terms. This scheme gives an accurate finite difference stencil
for the approximation of spatial derivative.
The process with all described assumptions was simulated with
the BARRACUDA-CPFD® package for char recycle flow at 1150 K
and 1350 K. The selected temperatures are to maintain the tem-
perature profile recommended gasifier arrangement given the
reference [24]. Each simulation was then run over 10 s to achieve
pseudo-steady state. No major changes were observed in effluent
conditions after 3 s.
Furthermore to compare the CPFD simulation with a simplified
approach, the same feeding reactor section dimensions were con-
sidered in PFR (plug flow reactor) model. This PFR model with the
same coal and steam feed was simulated in Matlab® software. The
kinetics described via equations provided in Table 1 and assum-
ing the carbon combustion reaction forming CO, R4 , is completed
in the bottom reactor section. As well as the gasifier, the plug flow
is separated in three zones: (a) the air and steam mixing section,
(b) the recycled ash rich particles feeding section, (c) the coal feed-
ing section. The selected incoming 1150 K and 1350 K temperatures
were considered for the recycled ash rich section on the basis of
expected desirable operating range conditions for coal gasifier units
as reported by Watanabe and Otaka [24].

7. Results and discussion

The main objectives of present study was to investigate the


effect of various operational parameters in the fluid dynamics of Fig. 2. The contour of particle volume fraction and particle size distribution at 10 s
simulation. Operating conditions: (a) 100 g/s ash rich particle recycle flow at 1150 K,
the near gasifier feeding section using the CPFD model and to com-
(b) 2.1 g/s coal feed at T = 298 K, (c) 2.5 g/s air/steam flow (1.648 molar ratio) at
pare product compositions predictions from both the CPFD and PFR T = 1250 K. Total feeding pressure: 1.1 MPa.
models.
Fig. 2 reports the contour of particle volume fraction and particle
as well as the fluid velocity display local variations confined to
size distribution at 10 s for recycles flow temperature of 1150 K or
the bottom of the gasifier feeding section. This local densification
Condition 1. Condition 1 was specifically selected for the numerical
of the bed provides of early signs of choking. Furthermore tem-
calculation given it is at the bottom of the column feeding sec-
perature contours in Fig. 3a show in most cases relatively small
tion where the superficial gas velocity supersede several times the
particle terminal velocity or Gs /Gg ratio of the solid fluxes is 27.06
and choking of the suspension may be promoted [27]. Observed
numerical results show that the solid distribution and flow, while
relatively uniformly throughout the unit start, display local varia-
tions with some densification in the bottom feeding of the reactor.
Fig. 3 displays fluid temperature, dp/dz and fluid velocity con-
tours for Condition 1. One can see again that the dp/dz contours

Table 5
Rich ash and carbon particles size distribution.

Cum percentage Radius (m)

0 6.00E−06
3 1.20E−05
6 1.70E−05
9 2.00E−05
20 2.50E−05
45 3.50E−05
65 4.50E−05
75 4.70E−05
Fig. 3. The contour of fluid temperature, pressure drop and fluid velocity at 10 s sim-
85 5.00E−05
ulation. Operating conditions: (a) 100 g/s ash rich particle recycle flow at 1150 K,
95 6.00E−05
(b) 2.1 g/s coal feed at T = 298 K, (c) 2.5 g/s air/steam flow (1.648 molar ratio) at
100 7.00E−05
T = 1250 K. Total feeding pressure: 1.1 MPa.
346 A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350

Fig. 4. The contour of gas species at 10 s simulation. Operating conditions: (a) 100 g/s ash rich particle recycle flow at 1150 K, (b) 2.1 g/s coal feed at T = 298 K, (c) 2.5 g/s
air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding pressure: 1.1 MPa.

fluid temperature variations across the column radius. These data bed losing stability with choking, the particle volume fraction was
supports a close to PFR model which neglects radial temperature calculated at 1 m column height which corresponds to about the
variations across the unit section. middle of the gasifier bed feeding section as reported in Fig. 5.
Fig. 4 reports product composition at various radial and axial Results in Fig. 5 show that the particle volume fraction for the
positions at 10 s of simulation for Condition 1. These results show flow with ash rich recycled stream fed at 1150 K (Condition 1)
that the product fraction can change with both axial and radial are 0.02 and the fluctuations standard deviations are limited to
position. As well simulation shows local variation with time, ±0.008. This represents a bed with mild local solid particle den-
though time average values appear to be fairly stable. Regarding sification. For instance, the choking conditions for the FCC type A
O2 consumption, the contour of Fig. 4g, supports the simplifying particles are about 0.04 [28]. Thus the CPFD model is valuable to
assumption that all O2 reacts in the very bottom of the column confirms that in the case of Condition 1 bed choking is unlikely to
feeding section (Zone 1). This oxygen depletion takes place con- take place with this providing a stable operating condition for the
currently with pyrolysis as a preamble to the gasification taking gasifier.
place in the rest of the gasifier unit. Using the same methodology the particle volumetric concentra-
Choking is commonly defined as a phenomenon where a sud- tion for an ash rich particle stream recycled at 1350 K or Condition
den change in the solids holdup occurs in a gas–solid fluidization 2 is also reported in Fig. 5. A similar average solid concentration
system. The gas flow under these conditions is not capable of sus- of 0.02 is observed with solid concentrations fluctuations limited
taining the gas solid suspension [27]. to fluctuations with a standard deviation of ±0.014. This also con-
To study the flow behaviour of the gas–solids in a fast flu- firms that in the case of Condition 2 the gasifier is operated under
idized bed in the near feeding region and the possibility of the conditions where chocking of the bed is avoided.
A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350 347

Fig. 8. PFR gas species mass fractions profile along the bed feeding section. Oper-
Fig. 5. Transient particle volume fraction distributions at 1 m. Operating conditions: ating conditions: (a) 100 g/s char flow at 1150 K, (b) 2.1 g/s coal feed at T = 298 K,
(a) 100 g/s ash rich particle recycle flow at 1150 K and1350 K, (b) 2.1 g/s coal feed (c) 2.5 g/s air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding pressure:
at T = 298 K, (c) 2.5 g/s air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding 1.1 MPa.
pressure: 1.1 MPa.

Figs. 8 and 9 report the axial concentration profile in the PFR. The
total char conversion in Condition 1 is 26.3% while it is found to be
27.7% for Condition 2. On this basis of the results reported in Fig. 8
one can postulate that for an ash rich particle recycle flow at 1150 K
or Condition 1, there is little change of various gases once combus-
tion and devolatilazation is complete. Fig. 9 reports the product
distribution changes in the case of Condition 2 when the recycle
flow temperature is increased up to 1350 K. In contrast with the
operation at 1150 K, at 1350 K both CO and H2 O are now reduced
with CO2 and H2 fractions increased suggesting a significant gain
in importance of the water gas shift reaction.
Fig. 10 compares the axial temperature profiles using both PFR
and CPFD models. Figs. 11 and 12 describe the chemical species
product fractions at 1.10 m total unit height which corresponds to
a close to the expected total height gasifier feeding section. Data
Fig. 6. The outlet gas species mass fractions. Operating conditions: (a) 100 g/s ash reported in these figures was obtained in the case of the CPFD using
rich particle recycle flow at 1150 K, (b) 2.1 g/s coal feed at T = 298 K, (c) 2.5 g/s radial and time averages values.
air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding pressure: 1.1 MPa. One can notice that there is close agreement for both tempera-
tures and molar fractions of various chemical species. Thus, these
results confirm that there are minor discrepancies in H2 , H2 O, CO
The effluent product compositions as calculated for the two dif- and CO2 concentrations and temperatures using either the CPFD
ferent recycles temperatures and the same operating conditions or the PFR at 1.10 m axial position. Based on this analysis, one can
(recycle flow is flow of 100 g/s, the coal feed is 2.1 g/s with T = 298 K conclude that the PFR model is very effective for the description of
and air steam flow is 2.5 g/s with T = 1250 K) are reported in Figs. thermal and concentration changes in the gasifier feeding section.
6 and 7. In both cases, the simulations take approximately 3 s to In summary, a PFR simplified model is adequate to predict con-
reach a stable state numerical solution. It is understood the cal- centration of species and thermal gradients in a gasifier as the type
culated product concentrations will fluctuate over time with their proposed by the CRIPI of Japan. This pseudo homogeneous PFR
average value at a given axial position being very stable. model is simple to handle and can be solved even for fairly complex

Fig. 9. PFR gas species mass fractions profile along the bed feeding section. Oper-
Fig. 7. The outlet gas species mass fractions. Operating conditions: (a) 100 g/s ash ating Conditions: (a) 100 g/s ash rich particle recycle flow at 1350 K, (b) 2.1 g/s coal
rich particle recycle flow at 1350 K, (b) 2.1 g/s coal feed at T = 298 K, (c) 2.5 g/s feed at T = 298 K, (c) 2.5 g/s air/steam flow (1.648 molar ratio) at T = 1250 K. Total
air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding pressure: 1.1 MPa. feeding pressure: 1.1 MPa.
348 A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350

gasification kinetics as the one of the present study. Model differ-


ential equations can be solved equation solvers such as the ones
available in Matlab® and Mathcad® . There are also other software
packages that provide solutions for general PFR reactors with user
defined kinetics such as Comsol® .
However these simplified models experience severe limitations
to properly account for the reactor fluid dynamics such is the one
observed in the feeding section of the gasifier of the present study.
It is in this respect, that more comprehensive simulation tools such
the CFD or the CPFD techniques of this study are needed to add
critical information to the gasifier model description.
In this respect, CPFD is called to provide detailed information at
every point inside the reactor mesh an at any time step during the
simulation. Thus in this context, simulation results are very use-
ful for investigating gas-suspension flow phenomena, suspension
dynamic behaviour and local suspension of flow effects. However
Fig. 10. CPFD and PFR temperature profile along the bed for recycles flow temper-
even if very valuable CPFD simulations may be time consuming.
atures of 1150 K and 1350 K with a ash rich particle recycle flow of 100 g/s. The coal Thus, CPFD simulation is recommended for testing specific condi-
feed is 2.1 g/s with T = 298K and air steam flow is 2.5 g/s with T = 1250 K. Total feeding tions where gasification and gasifier design are well established.
pressure: 1.1 MPa. Furthermore CPFD simulations can also surpass PFR modeling
when there is a need for evaluating the effect of design changes to
existing systems. In the specific case of the present study CPFD is
called to describe the fluid dynamic of a gasifier in the near feeding
region, a critical region where early warnings of suspension choking
will manifest. It is in fact with the help of CPFD that one can establish
conditions for stable and free of choking gasifier operation.
Finally one can also argue that PFR simulation provides a valu-
able support to the calculations with CPFD. Time and cross-section
averaged values for CPFD gasifier models should provide close com-
position and temperatures with respect to the ones resulting from
PFR if the numerical calculations with CPFD model converged ade-
quately.

8. Conclusion

The following are the conclusions of the present study:

(a) The PFR and the CPFD simulations of the present study provide
valuable and complementary information about the operation
Fig. 11. Comparison of effluent compositions for CPFD and PFR models. Operating of the feeding gasifier section.
conditions: (a) 100 g/s ash rich particle recycle flow at 1150 K, (b) 2.1 g/s coal feed (b) The PFR model gives good estimations of temperature and
at T = 298 K, (c) 2.5 g/s air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding
chemical composition of the various species.
pressure: 1.1 MPa.
(c) The PFR model also revealed itself as a valuable tool for
checking adequate CPFD gasifier simulations and calculation
convergence. This is accomplished showing that the time and
cross-section averaged temperature and concentrations for the
CPFD closely approximate the ones for the PFR.
(d) The CPFD simulation provides valuable description of local and
dynamic particle concentration, gas and particle velocities.
(e) The CPFD simulation gives important information for the oper-
ation and design of coal gasifiers. The CPFD model allows
predicting early signs of suspension choking in the gasifier feed-
ing section. As a result, using CPFD one can determine safe
operating conditions for fast fluidized bed coal gasifiers.

List of symbols
A reactor cross section (m2 )
Cd drag coefficient, dimensionless
Cp,i specific heat i (J/mol K)
Df diffuse bounce
Dp interphase drag coefficient (m/s2 )
en particle–wall normal retention coefficient
Fig. 12. Comparison of the effluent results from CPFD and PFR. Operating conditions:
et particle–wall tangential retention coefficient
(a) 100 g/s ash rich particle recycle flow at 1350 K, (b) 2.1 g/s coal feed at T = 298 K,
(c) 2.5 g/s air/steam flow (1.648 molar ratio) at T = 1250 K. Total feeding pressure: E defined by Eq. (4) (J/kg)
1.1 MPa. F momentum exchange rate per volume between the fluid
and particle phase (N s/m3 )
A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350 349

Fcoal molar flow rate of coal (mol/s) vr,j reference stoichiometric coefficient based on Rj , dimen-
Fix molar flow rate of species i at position x (mol/s) sionless
g gravitational acceleration (m/s2 ) vp particle velocity (m/s)
Gg gas flux (kg/m2 s)
Gs solid flux (kg/m2 s) Species designations
H enthalpy for ideal gases (J/kg) Ash rich solid particles with a high ash content as described in
Hi enthalpy for ideal gas of species i (J/kg) Table 4
IP maximum pressure iterations Char solid particles at the outlet of the gasifier
Iu maximum velocity iterations Coal solid particles with content of char, volatile, H2 O, and ash
Iv maximum volume iterations as described in Eq. (12)
Nf particle feed per average volume
Np computation particle per cell
Nc total number of clouds Acronyms
P fluid pressure (N/m2 ) CPFD computational particle fluid dynamics
Pi partial pressure (N/m2 )
Ps positive constant (N/m2 ) Acknowledgments
q energy transfer due to conduction, species diffusion and
viscous dissipation (J/m2 s) We would like to acknowledge the valuable support of the
Q heat losses from the PFR model (J/s) MITACS program and the Reactech Process Development Inc. who

Q energy source term (J/m3 .s) supported a PDF position for Dr. Alireza Abassi at the University
rP pressure residual of Western Ontario. We are also grateful to the Natural Sciences
ru velocity residual and Engineering Research Council of Canada (NSERC) who also con-
rv volume residual tribute financially to the development of this research.
Rj reaction rate of species j (mol/m3 s)
Sf fluid source term (kg/m3 s) References
Tg gas temperature (K)
Tref reference temperature for the heat of reaction (K) [1] Molerus, Hydrodynamische Stabilität des Fließbetts, Chem. Ing. Tech. 39 (1967)
341–348.
V volume of the reactor (m3 ) [2] H. de Lasa, G. Gau, Influence des agrégats sur le rendement d’un réacteur à
Vp particle volume (m3 ) transport pneumatique, Chem. Eng. Sci. 28 (1973) 1875–1884.
x position in the PFR model (m) [3] Y. Yousfi, G. Gau, P. Le Goff, Réacteur multitubulaire à transport pneumatiqu,
Int. Symp. One fluidization Toulouse Soc. Chem. France (1973).
Xsl mass fraction of the fixed carbon in the solids, dimension- [4] W.C. Yang, Handbook of Fluidization and Fluid-Particle Systems, Marcel Dekker,
less Inc., 2003.
Yi mass fraction of species i, dimensionless [5] U. Mann, Principles of Chemical Reactor Analysis and Design—New Tools for
Industrial Chemical Reactor Operations, John Wiley & Sons, Inc., 2009.
[6] S. Zimmermann, F. Taghipour, CFD modeling of the hydrodynamics and reac-
tion kinetics of FCC fluidized-bed reactors, Ind. Eng. Chem. Res. 44 (26) (2005)
Greek letters 9818–9827.
˛1 and ˛ 1 stoichiometric coefficient of CH4 , dimensionless [7] J. Jung, I.K. Gamwo, Multiphase CFD-based models for chemical looping
˛2 and ˛ 2 stoichiometric coefficient of H2 , dimensionless combustion process: fuel reactor modeling, Powder Technol. 183 (3) (2008)
401–409.
˛3 and ˛ 3 stoichiometric coefficient of CO, dimensionless [8] B. Dou, V. Dupont, P.T. Williams, Computational fluid dynamics simulation of
˛4 and ˛ 4 stoichiometric coefficient of CO2 , dimensionless gas–solid flow during steam reforming of glycerol in a fluidized bed reactor,
ˇ and ˇ stoichiometric coefficient of H2 O, dimensionless Energ. Fuel. 22 (6) (2008) 4102–4108.
[9] D.M. Snider, S. Banerjee, Heterogeneous gas chemistry in the CPFD
ς particle/fluid slip ratio
Eulerian–Lagrangian numerical scheme (ozone decomposition), Powder Tech-
maximum momentum redirection from collision nol. 199 (1) (2010) 100–106.
constant number, dimensionless [10] FLUENT 6.3 User’s Guide; FLUENT Inc.: Lebanon, NH, 2006.
[11] L. Huilin, H. Yurong, D. Gidaspow, Hydrodynamic modelling of binary mixture
ε constant number, dimensionless
in a gas bubbling fluidized bed using the kinetic theory of granular flow, Chem.
εs solid porosity, dimensionless Eng. Sci. 58 (7) (2003) 1197–1205.
H1 enthalpy changes for species coming the control volume [12] L.S. Leung, Vertical pneumatic conveying: a flow regime diagram and a
(J/s) review of choking versus non-choking systems, Powder Technol. 25 (2) (1980)
185–190.
H2 total heat of reaction in the control volume (J/s) [13] D.M. Snider, Three fundamental granular flow experiments and CPFD predic-
H3 enthalpy changes for species leaving the control volume tions, Powder Technol. 176 (1) (2007) 36–46.
(J/s) [14] C.E. Brennen, Fundamentals of multiphase flow, Cambridge University Press,
2005.
Hr heat of reaction (J/mol) [15] L.M. Zou, Y.C. Guo, C.K. Chan, Cluster-based drag coefficient model for sim-
t time step (s) ulating gas–solid flow in a fast-fluidized bed, Chem. Eng. Sci. 63 (4) (2008)
V control volume (m3 ) 1052–1061.
[16] F.M. Auzerais, R. Jackson, W.B. Russel, The resolution of shocks and the effects
f gas viscosity (N s/m2 ) of compressible sediments in transient settling, J. Fluid Mech. 195 (1988)
f fluid density (kg/m3 ) 437–462.
p particle density (kg/m3 ) [17] M. Syamlal, L.A. Bisset, METC Gasifier Advanced Simulation (MGAS) Model,
DOE/METC–92/4108, DE92 001111, 1992.
 cp particle volume fraction at close packing limit, dimen-
[18] M. Syamlal, C. Guenther, A. Gel, S. Pannala, Advanced coal gasifier designs using
sionless large-scale simulations, J. Phys.: Conf. Ser. 180 (2009) 012034.
f fluid volume fraction, dimensionless [19] Q. Li, M. Zhang, W. Zhong, X. Wang, R. Xiao, B. Jin, Simulation of coal gasifica-
tion in a pressurized spout-fluid bed gasifier, Can. J. Chem. Eng. 87 (2) (2009)
p particle volume fraction, dimensionless
169–176.
Re Reynolds number, dimensionless [20] F. Bustamante, R.M. Enick, R.P. Killmeyer, B.H. Howard, K.S. Rothenberger,
 fluid stress tensor (N/m2 ) A.V. Cugini, B.D. Morreale, M.V. Ciocc, Uncatalyzed and wall-catalyzed forward
p particle normal stress (N/m2 ) water–gas shift reaction kinetics, AIChE J. 51 (5) (2005) 1440–1454.
[21] L. Yu, J. Lu, X. Zhang, S. Zhang, Numerical simulation of the bubbling fluidized
vf fluid velocity (m/s) bed coal gasification by the kinetic theory of granular flow (KTGF), Fuel 86 (5–6)
vi,j stoichiometric coefficient of species j, dimensionless (2007) 722–734.
350 A. Abbasi et al. / Chemical Engineering Journal 174 (2011) 341–350

[22] E. Salaices, B. Serrano, H. de Lasa, Biomass catalytic steam gasification ther- [26] K.L. Cashdollar, M.J. Sapko, E.S. Weiss, M.L. Harris, C.K. Man, S.P. Harteis, G.M.
modynamics analysis and reaction, experiments in a CREC riser simulator, Ind. Green, Recent Coal Dust Particle Size Surveys and the Implications for Mine
Eng. Chem. Res. 49 (15) (2010) 6834–6844. Explosions, Centers for Disease Control and Prevention, October, National
[23] C.Y. Wen, C.Y., H. Chen, M. Onozaki, User’s Manual for Computer Simu- Institute for Occupational Safety and Health, Pittsburgh Research Laboratory,
lation and Design of the Moving Bed Coal Gasifier, DOE/MC/16474-1390, Pittsburgh, PA, 2009.
NTIS/DE83009533, 1982. [27] B. Du, L.S. Fan, Characteristics of choking behavior in circulating fluidized beds
[24] H. Watanabe, M. Otaka, Numerical simulation of coal gasification in entrained for group B particles, Ind. Eng. Chem. Res. 43 (18) (2004) 5507–5520.
flow coal gasifier, Fuel 85 (12–13) (2006) 1935–1943. [28] D. Bai, A.S. Issangya, J.R. Grace, A novel method for determination of choking
[25] Slezak, J.M. Kuhlman, L.J. Shadle, J. Spenik, S. Shi, CFD simulation of entrained- velocities, Powder Technol. 97 (1) (1998) 59–62.
flow coal gasification: coal particle density/size fraction effects, Powder
Technol. 203 (1) (2010) 98–108.

You might also like