Você está na página 1de 27

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/222552755

Simulation of particles and gas flow behavior in


the riser section of a circulating fluidized bed
using the kinetic theory approach for the
particulate phase
Article in Powder Technology October 2000
Impact Factor: 2.35 DOI: 10.1016/S0032-5910(99)00302-2

CITATIONS

READS

236

263

4 authors, including:
Hamid Arastoopour

Hajar Massah

Illinois Institute of Technology

University of Isfahan

120 PUBLICATIONS 1,375 CITATIONS

2 PUBLICATIONS 236 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,


letting you access and read them immediately.

SEE PROFILE

Available from: Hamid Arastoopour


Retrieved on: 21 June 2016

Accepted for publication in Powder Technology

Simulation of Particles and Gas Flow Behavior in the Riser Section


of a Circulating Fluidized Bed Using the Kinetic Theory
Approach for the Particulate Phase
S. Benyahia a, H. Arastoopour a,* , T.M. Knowlton b, and H. Massah c
a
b

Illinois Institute of Technology, Chicago, IL 60616


Particulate Solid Research Inc., Chicago, IL. 60632
c
Fluent, Inc., Lebanon, NH 03766

Abstract
Gas/particle flow behavior in the riser section of a circulating fluidized bed (CFB)
was simulated using a computational fluid dynamics (CFD) package by Fluent Inc. Fluid
Catalytic Cracking (FCC) particles and air were used as the solid and gas phases,
respectively.
A two-dimensional, transient and isothermal flow was simulated for the
continuous phase (air) and the dispersed phase (solid particles). Conservation equations
of mass and momentum for each phase were solved using the finite volume numerical
technique. This approach treats each phase separately, and the link between the gas and
particle phases is through drag, turbulence, or energy dissipation due to particle
fluctuation.
Gas and particle flow profiles were obtained for velocity, volume fraction,
pressure, and turbulence parameters for each phase. The computational values agreed
reasonably well with the available experimental results. Our computational results
showed that the inlet and outlet design have significant effects on the overall gas and
solid flow patterns and cluster formations in the riser. However, the effect of the initial
condition tended to disappear after some time. The main frequencies of oscillations of the

Corresponding author. Tel.: +1-312-567-3038; Fax: +1-312-567-8874; E-mail:hamid.arastoopour@iit.edu

Accepted for publication in Powder Technology

system were obtained in different regions of the riser. These frequencies are important in
comparing the computational results with the available time-averaged experimental data.

Keywords: Kinetic theory; Numerical simulation; Circulating beds; Fluidization; FCC particles

Introduction
Gas/solid flow systems are an essential part of many chemical processes, and
contributions to the understanding of the behavior of such flow systems can significantly
enhances the design and, in turn, the productivity of such processes.
The first step in the fundamental understanding of fluidization is usually
attributed to Davidson [1] for his analysis of a single bubble motion in an infinite fluid
bed. The development of Davidson's model was carried by Jackson [2], Murray [3],
Pigford and Baron [4], and Soo [5,6]. Solid viscosity and pressure are closures that can
either be determined experimentally or theoretically by using more fundamental
approaches. In any case, solid viscosity is an important parameter to include in
simulations of the riser section of a circulating fluidized bed (CFB) since wall effects can
be significant.
In recent years, the dynamics of inter-particle collisions based on the concept of
the work of Chapman and Cowling [7] have been considered by Soo [5], Jenkins and
Savage [8], Lun et al. [9], Johnson and Jackson [10], Sinclair and Jackson [11], and Ding
and Gidaspow [12]. The kinetic theory approach uses a one equation model to determine
the turbulent kinetic energy of the particles (granular temperature) and assumes either a
Maxwellian distribution for the particles, or a non-Maxwellian distribution [13], which
considers both dilute and dense cases. Moreover, Kim and Arastoopour [14] considered

Accepted for publication in Powder Technology

the cohesiveness of the particles and extended the kinetic theory model for cohesive
particles. The kinetic theory approach for granular flow allows the determination of the
pressure and viscosity of the solid in place of empirical relations.
The interaction between gas and particles has been considered in the form of drag
force in most of the publications in the literature. Louge et al. [15] were probably the first
to include both gas turbulence and particle-particle interaction. Recently, most of the
investigations have focused on very dilute systems with gas turbulence and, in some
cases, large-scale particulate phase fluctuations [16, 17, 18]. This could be due to the fact
that most gas phase turbulence models [19] have been developed for single-phase flow,
and their application could most probably be limited to a very dilute flow of particles. For
the case of dense particle flows, the interaction between gas turbulence and particle
fluctuation is significant and, in some cases, the flow tends to be similar to laminar flow.
The kinetic theory formulation includes an interaction term between gas turbulence and
particle fluctuations in the granular temperature equation [13]. However, the validity of
this expression to describe the interactions between gas turbulence and particle
fluctuation is yet to be investigated.
FCC units usually operate at high solid throughput of 400-1200 kg/m2s. The riser
section of a CFB can be characterized by a core-annular flow regime with particles
forming structures at the walls in the form of clusters and sheets. Lateral segregation in
high density CFB's was probably first found experimentally by Bader et al. [20] who
simultaneously measured the solid density and mass flux across the diameter of the riser.
Simulation of lateral segregation in the riser is possible only by using at least a twodimensional model. Tsuo and Gidaspow [21] were among the first to predict the core-

Accepted for publication in Powder Technology

annular flow regime using multiphase flow equations. They successfully used an
empirical relation for the solid viscosity that was later used by Bing [22] and Benyahia et
al. [23]. However, theoretically derived solid viscosity and pressure are the preferred
relations to fully predict the gas/solid in the riser.
Recently, Knowlton et al. [24] presented a challenge problem at the fluidization
VIII conference to compare different CFB hydrodynamic models. The experimental data
were produced mainly at the Particulate Solid Research Inc. (PSRI). These experimental
data included the solid concentration and flux at a certain location of the riser, as well as
the axial gas pressure drop. Benyahia et al. [23] conducted a simulation using an
empirical relation for solid viscosity and compared their results with the pilot-scale
experimental data. The analysis in this paper will include the use of the kinetic theory for
solid particles; the main goal is to investigate the effect of initial conditions, inlet and
outlet design, and increased riser diameter on the overall gas/solid mixing in the riser.
Furthermore, frequency analyses of the oscillatory gas/solid flow were performed. This
enables an accurate comparison of the time-averaged computational results with the
experimental data.

Gas/Solid Multiphase Flow Model Description


The complexity of the hydrodynamic equations makes obtaining an analytical
solution very unlikely. Therefore, numerical solutions should be considered. In the Fluent
computer program that was used in this study, the governing equations were discretized
using the finite volume technique [25]. The discretized equations, along with the initial
and boundary conditions, were solved to obtain a numerical solution.

Accepted for publication in Powder Technology

Conservation equations of mass and momentum were developed using the


Eulerian approach, and were solved simultaneously. This approach treats each phase
separately, and the only link between the two phases is through the drag force in the
momentum equations.
The discretization scheme for the convection terms in the momentum equations
uses the power law interpolation scheme which provides a formal accuracy between first
and second order. This scheme is more robust and less computationally intensive than
higher order schemes. The SIMPLE iterative algorithm [25] is used by Fluent to relate
the velocity and pressure corrections to recast the continuity equation in terms of a
pressure correction calculation. The Inter-Phase Slip Algorithm (IPSA) method that uses
the Partial Elimination Algorithm (PEA) developed by Spalding [26] is used by Fluent4.4. The Full Elimination Algorithm (FEA), which is a fully coupled implicit approach
used in the new version of Fluent-4.5, significantly enhances convergence of the
numerical scheme.
Governing Equations
The following are the equations of conservation of mass and momentum for the
gas and particulate phases as well as the particulate phase fluctuating energy [13, 27]:
The conservation equation of mass of phase i (i = gas, solid)

( i i ) + .( i i U i ) = 0
t

(1)

with the constraint


i =1

(2)

Accepted for publication in Powder Technology

The conservation of momentum of phase I (i = gas, solid, k i)

( i i U i ) + .( i i U i U i ) = i P +.Ti (U i U k ) + i i g
t

(3)

The fluctuations that occur in the solid phase were modeled from the kinetic theory of
gases modified to account for inelastic collisions between particles. The equation for the
turbulent fluctuating energy of solid, called granular temperature, s ( s =

1
U i ), may
3

be written as:
3
( s s s ) + . ( s s U s s ) = Ts : U s + . (k s ) s 3 s + C g C s
2t

(4)

The last two terms in Equation (4) represent the interaction between gas turbulence and
particle fluctuation, and their derivation may be found in Gidaspow [13]. Numerically,
these terms can be important. For example, the term before the last term may be about
three times larger than the dissipation term. However, these terms were not included in
the present model since no turbulence model in the gas phase was used.
Constitutive Equations
Constitutive relations are needed to close any governing relations. The following
represent the constitutive relations used in the current model.
Arastoopour et al. [28] used a gas/solid drag force that gives continuous values over all
ranges of solid volume fractions:
g
17.3
=
U g U s (1 g ) g 2.8
+ 0.336
R
d
p
e
Re =

(5)

d p U g U s g

Accepted for publication in Powder Technology

Solid phase stress:


Ts = ( Ps + s b U s )I + 2 s s S s

(6)

Deformation rate:
Ss =

1
Us + ( Us
2

)T

] 13 ( U

)I

(7)

Gas phase stress:


Tg = 2 g g S g

(8)

The solid pressure ( Ps ), viscosity ( s ), and conductivity ( s ) are composed of two parts:
a kinetic term that dominates in the dilute flow regions and a collision contribution that is
significant in the dense flow regions.
Ps = s s s + 2 s ( 1 + e ) s2 g o s

(9)

Where e is the restitution coefficient, and go is the radial distribution function which
becomes very large as the solid volume fraction of solid approaches the maximum
packing ( s, max = 0.6 in this paper). The following expression was used by Ding and
Gidaspow [12]:

3 s
g o = 1
5 s ,max

1 1
3

(10)

Solid phase shear viscosity:


s =

2 s ,dil

1/ 2

4
s
4

1 + 5 g o s ( 1 + e ) + 5 s s d p ( 1 + e ) g o

(1 + e ) g o

(11)

Solid phase dilute viscosity:


s ,dil =

5 s d p s

(12)

96

Accepted for publication in Powder Technology

The solid bulk viscosity accounts for the resistance of the granular particles to
compression and expansion, and is given by:
4

b = s s d p g o (1 + e ) s
3

1/ 2

(13)

The diffusion coefficient for the solid phase energy fluctuation is:
1/ 2
2
150 s d p s 6
s

2
ks =
1 + 5 g o s ( 1 + e ) + 2 s s d p g o ( 1 + e )
384 (1 + e) g o

(14)

The collisional dissipation of energy fluctuation is:


4 1 / 2

s = 3 s2 s g o s 1 e 2
s . U s
d p

(15)

The dissipation of energy fluctuation due to interaction of particles with the gas phase
was not considered. The term ( .U s ) in Equation (15) can be positive or negative
depending on the direction of the flow in the riser. Since the term ( .U s ) was about two
orders of magnitude less than the positive term on its left, Equation (15) will always be a
dissipative term in the granular temperature Equation (4). It is important to note that the
strong dependency of the dissipation of the granular temperature on the restitution
coefficient (e) appears in term ( 1 e 2 ). An increase in the restitution coefficient from 0.9
to 0.99 would lead to an order of magnitude decrease in the dissipation ( s ). This
decrease in the collisional dissipation yields to a significant increase in the granular
temperature. Pita and Sundaresan [29,30] pointed out the high sensitivity of their
computed solids density and mass flux to the restitution coefficient. They showed that
only a value of e = 1 was appropriate to accurately compute the experimental data of
Bader et al. [20]. However, this value of restitution coefficient is not realistic since the

Accepted for publication in Powder Technology

particle-particle collisions in the FCC riser are not elastic. For particles like FCC
catalyst, the restitution coefficient may be deduced from indirect measurement. Gidaspow
et al. [31] estimated the restitution coefficient to be close to unity from measurements of
the Reynolds stress and granular temperature. In this study, a particle-particle restitution
coefficient (e) of 0.95, and a particle-wall restitution coefficient (ew) of 0.9 were used. In
the literature, the reported measured restitution coefficients were for larger particles than
FCC particles [32].
Boundary Conditions
At the inlet, all velocities and volume fractions of both phases were specified. The
pressure was not specified at the inlet because of the incompressible gas phase
assumption (relatively low pressure drop system).
At the outlet, only the pressure was specified (atmospheric). The other variables
were subject to the Newmann boundary condition.
At the wall, the gas tangential and normal velocities were set at zero (non-slip
condition). The solid normal velocity was also set at zero. The following boundary
equations apply for the solid tangential velocity and granular temperature at the wall:
v s ,w =

6 s s ,max
v s ,w
3 s s g0 s n

(16)

3 s s v s2,slip g 0 s3 / 2
k s s s ,w
s ,w =
+
n
s ,w
6 s ,max s ,w

(17)

These equations were developed by Hui et al. [33], and Johnson and Jackson [10].
According to their work, the slip velocity between particles and the wall can be obtained
by equating the tangential force exerted on the boundary and the particle shear stress
close to the wall. Similarly, the granular temperature at the wall was obtained by equating

Accepted for publication in Powder Technology

the granular temperature flux at the wall to the inelastic dissipation of energy, and to
generation of granular energy due to slip at the wall region. Later, these boundary
conditions were successfully applied to dense riser flow [34].

Simulation and Results of Gas/Solid Flow in a 2-D Riser


System Description
Figure 1 shows the riser section of a CFB used in the present numerical
simulation of gas/solid flow. The geometry of the 20-cm riser is similar to the
experimental set-up used in the challenge problem by Knowlton et al. [24] for the case of
489 kg/m2s solid mass flux inside the riser. Solid particles were fed from both sides of the
riser near minimum fluidization conditions. At these conditions, an expected low value of
granular temperature was assigned to the particulate phase.
It is important to note that the real geometry of the riser is cylindrical with the
solid entering from one side only. To obtain mixing at the entrance zone similar to the 3D experiment, a two-inlet geometry was selected. A one-inlet design for the 2-D riser
could lead the inlet gas to flow to the opposite side of the solid inlet, therefore, creating
low mixing throughout the height of the riser. The effects of inlet design will be
demonstrated in this paper.
Initially, the velocity of both the gas and solid was set at zero. Gas and solid
exited the system through two symmetric side outlets set 0.3 m below the closed top of
the riser with a width of 0.1 m (similar to the side-inlets). The total height of the riser was
14.2 m. The solid particles in this simulation consisted of FCC material of 76 m in

10

Accepted for publication in Powder Technology

diameter and 1712 kg/m3 in density. The simulated system was isothermal at 300 K, and
the initial pressure was set at 1 bar.
In this simulation, the computational domain consisted of 18 grids radially and
210 non-uniform grids along the axis of the riser. A total of 3780 fluid cells resulted from
this grid distribution. A constant time step of 5.10-4 sec was used, and the simulation was
conducted on an SGI Origin 200 computer for 40 sec of real fluidization time
corresponding to 3-4 weeks of computational time.
Results and Discussion
The flow patterns in the 20-cm riser section of a CFB consisted of a core-annular
flow regime typical of a high density CFB as shown in Figure 11. The solid density is
higher near the wall region and more dilute at the center of the riser. The solid velocity is
high at the center of the riser and solid down-flow at the wall regions is shown in Figure
11 by the arrow plot of the solid velocity vectors. Moreover, the solid down-flow was
mostly in the form of clusters descending at low velocities at the wall region.
Figure 2 shows the comparison between the calculated time-averaged (from 15 to
40 sec) solid density distribution ( s s ) at 3.9 m height inside the riser with the
experimental data taken at the same height. This figure shows a dilute region in the center
of the pipe and a high concentration of particles at the walls. This distribution reflects the
establishment of a core-annular regime shown by the experiments and the computational
results. However, contrary to the experiments, the computational results showed a smaller
core width in the core-annular system. By means of mass conservation, the solid
concentration at the walls was predicted to be less than the experimental values.
Moreover, it was experimentally proven that bigger particles in a mixture (as in an FCC

11

Accepted for publication in Powder Technology

particle size distribution) have a greater tendency than the smaller particles to
accumulate at the wall regions [35, 36]. This could result in a higher solid concentration
at the wall than that predicted in this study using mono-sized particles. In addition, a
more accurate boundary condition that accounts for the possible electrostatic forces at the
walls could increase the solid density to a value closer to the experimental data.
Figure 3 shows the comparison between the calculated time-averaged solid mass
flux distribution in the riser at a height of 3.9 m with the experimental data taken at the
same height. The computational results agreed reasonably well with the experimental
data. The solid flux was at its maximum value at the center of the riser, although the solid
density was at its lowest value. This is due to the very high solid (and gas) axial velocity
at the center of the riser. Near the wall region of the riser, there was down-flow of solid
mainly in the form of clusters (see Figure 11). The solid down-flow was due to the
weight of the solid in the annular region that exceeded the axial pressure drop. The
velocity boundary condition used in this case successfully captured the velocity
magnitude of solid down-flow at the walls.
Figure 4 shows the time-averaged axial pressure drop in the riser compared with
experimental data. The high pressure drop at the bottom of the riser was due to the effect
of solid feeding in that region. The pressure drop then decreased along the height of the
riser due to the decrease in the solid concentration. The pressure drop was in reasonable
agreement with the experimental data, although it was under-predicted. This could be
mainly due to the prediction of a bigger dilute core in the 2-D riser (see Figure 11), which
yielded lower flow resistance and, in turn, a lower calculated pressure drop.

12

Accepted for publication in Powder Technology

Figure 5 shows the instantaneous variation of the granular temperature with the
solid volume fraction after 10 sec of simulation. Gidaspow and Huilin [37] showed that,
in the dilute regions, the granular temperature is proportional to the solid volume fraction
raised to the power of 2/3. This is similar to the increase of an ideal gas temperature upon
compression. In the dense regions, the decrease in the granular temperature is due to the
decrease of the mean free path of the particles. The trend of the computed results shown
in Figure 5 agreed well with the experimental results of Gidaspow and Huilin [37].
However, the granular temperature was an order of magnitude lower than the
experimental data. This could be due to the neglect of inlet gas and, in turn, large-scale
fluctuations and the relatively low restitution coefficient used in the simulation. An
increase in the restitution coefficient will definitely increase the granular temperature
since the dissipation of fluctuating energy will decrease considerably. Figure 5 also
shows that the granular temperature was very low in the dense flow regions. The most
probable reason could be that the large-scale oscillations that were measured
experimentally were not included in the computational results [38]. The time-averaged
experimental measurements always include small- and large-scale oscillations [39], and
the computational results, in this case, include only the small-scale oscillations.
Therefore, a direct comparison between the computation and experimental data would be
accurate only if the large-scale oscillations are subtracted from the experimental data.
Unfortunately, Knowlton et al. (1995) did not obtain a time series analysis of their
experimental data. Therefore, calculations of the large-scale oscillations for this
experiment were not possible. On the other hand, large-scale oscillations may be

13

Accepted for publication in Powder Technology

calculated by computing the time-averaged quantity: U s U s using the results of our


simulation.
Figure 6 shows the variations of the viscosity of the solid particles with the solid
volume fraction in the riser after 10 sec. The solid viscosity measured by Gidaspow and
Huilin [37] was also found to be an order of magnitude higher than the computed results
in Figure 6. This was mainly due to the under-prediction of the granular temperature (see
Figure 5). In fact, a correct granular temperature will lead to a more realistic prediction of
viscosity using Equation 11, and a linear increase of the solid viscosity with the solid
volume fraction would be obtained [37].
Figure 7 shows the power spectrum density at different frequencies of oscillations
in the riser. The Fourier transform method has been used to calculate the power spectrum
of the solid volume fraction. This analysis was done in the dilute core region of the riser
at a height of about 12 m with a sampling time of 0.1 sec. The calculated average solid
volume fraction was about 5%. The main frequency of the solid particles was 0.14-0.2
Hz, which corresponds to 5-7 sec time-period. A similar frequency has been reported
experimentally in a different riser by Gidaspow et al. [40] and computed successfully by
Neri [34] using the kinetic theory approach for the solid phase.
Figure 8 shows the power spectrum of different frequencies of oscillations in the
dense region at about 3 m above the bottom of the riser. The main frequency of
oscillations of particles in the dense region was about 0.14 Hz, which corresponded to 7-8
sec time-period. The dense region had an average solid volume fraction of about 13%.
The main reason for calculating the frequency of oscillations of the gas/solid flow was to
know the minimum time required to conduct proper time averaging of the computational

14

Accepted for publication in Powder Technology

data. In the specific simulation that was run, there was not a significant difference
between the frequencies of oscillations in the dilute and dense regions. This could be due
to the small difference between what we called the dense and dilute solid densities. In the
case of larger differences in solid density, Gidaspow et al. [40] showed that large
differences could exist between the main frequencies of the dense and dilute regions. It is
important to mention that the computational values for the frequency analysis were taken
at least 3 m away from the solids inlet zone. In the inlet region, a high concentration of
solids along with solids inlet configuration led to a very small oscillation. Therefore, it is
advised to perform time series analysis away from the inlet region in order to capture the
right oscillatory flow patterns inside the riser. Experimentally, the measurements of
Knowlton et al. [24] did not include a frequency analysis of their system, and comparison
with the experimental data was, therefore, not possible.
Effects of Initial Conditions
A case where initially the solid particles filled the bottom of the riser to a height of 3
m at minimum fluidization conditions ( s ,mf = 0.4) was considered. All the other initial
and boundary conditions are the same as the previous case study.
Figure 9 shows the effect of initial conditions on the solid density distribution in
the riser at 3.9 m height compared with experimental data taken at the same height. At the
beginning of the simulation, the solid initially present in the riser moved toward the wall
region. This enhanced the solid density at the walls in the beginning of the simulation.
However, after a longer period of time, the solid initially present at the bottom moved
toward the top and mixed with the particles that were initially in the bed. The longer
averaging time (10-40 sec) showed no significant difference between this simulation and

15

Accepted for publication in Powder Technology

the previous one starting with an empty riser (see Figure 2). Therefore, the effect of the
initial condition used in this simulation tends to disappear after several seconds of
operation.
Effects of Inlet and Outlet Design
To investigate the inlet and outlet effects, only one inlet and outlet that are
diametrically symmetric are considered. Geometrically, this simulation seems to be in
many features closer to the experiment since the 3-D experiment consists of only one
inlet and outlet. The inlet solid velocities were U = V = 0.505 m/s with a feed inclination
of 45o, and a minimum fluidization solid volume fraction of 0.4. This assured a 489
kg/m2s solid mass flux in the riser similar to the previous cases.
Figure 10 shows the time-averaged solid density at 3.9 m height in the riser
compared with the experimental data at the same height. The solid density was not
symmetric due to inlet and outlet effects. Most of the solid particles are concentrated at
the inlet side of the riser, and the opposite side of the riser was maintained at a relatively
dilute concentration. The typical core-annular flow regime was not clearly observed in
this simulation. It is important to mention that the experimental data were taken from one
side of the riser to its center only. Therefore, it is not certain that the experimental data
were symmetric. In the previous figures, radial symmetric behavior of the data was
assumed for the sake of comparison with the computational values. The exact
experimental inlet and outlet configuration and conditions cannot be implemented unless
a 3-D simulation is conducted in a geometry that is more complex. This requires very
high computational time with today's workstations capabilities.

16

Accepted for publication in Powder Technology

Effects of Doubling the Riser Diameter


To investigate the effect of increasing the riser diameter on the flow profile and
pressure drop along the riser, a 40-cm riser operating at the same solid flux of 489 kg/m2s
as the 20-cm riser was considered. The diameter of the side inlets and outlets was 20 cm.
All other initial, inlet, and outlet conditions are described in Figure 1.
Figure 11 shows a comparison of the flow behavior in a 20- and 40-cm riser after
40 sec. The core-annular flow regime can be seen clearly in this figure since the core of
both risers is more dilute than the near wall regions. The velocity of the particulate phase
is higher at the center, and down-flow of solids occurred at the wall of the two risers
mainly in the form of clusters. It is noticeable that the length of the core of the 40-cm
riser was relatively smaller than that of the 20-cm riser. This was due primarily to solid
side inlet design. A change in the inlet design could significantly change the solid mixing
and flow patterns in the lower region of the riser.
Figure 12 shows a comparison between the time-averaged axial pressure drop
inside the 20- and 40-cm risers. From single-phase flow calculations, it could be shown
[41] that a higher diameter would lead to a lower pressure drop inside the riser. However,
the effects of solid inlet conditions significantly affected the mixing in the inlet region of
the riser and, in turn, resulted in higher pressure drop in the riser with a larger diameter.
The higher pressure drop was also due to the higher solid hold up in the riser as shown in
Figure 11. Therefore, special care should be taken when designing the solid inlets and
outlets.

17

Accepted for publication in Powder Technology

Conclusions
A two-dimensional transient model incorporating the kinetic theory for the solid
particles used in the Fluent code was capable of predicting reasonably well the complex
gas/solid flow behavior in the riser section of a CFB.
The core-annulus flow observed in the dense riser flow was predicted by this
model. The flow regime was significantly affected by the solid down-flow in the form of
clusters near the walls of the riser.
The calculated solid flux and the axial pressure drop inside the riser compared
reasonably well with the available large-scale experimental data. However, the calculated
solid density deviated from the experimental data at the wall region. We believe that
experimentally verified boundary conditions that account for particle structures due to
electrostatic effects, particles cohesiveness, and multi-sized particles are needed to
accurately predict the solid density at the wall region. The kinetic theory should be
extended for multi-sized particles with the proper interaction terms between particles of
different sizes.
The kinetic theory for solid particles predicts well the trends and behavior that are
experimentally observed for the granular temperature and solid viscosity. The differences
are due to the restitution coefficient, large-scale fluctuations, and use of proper gas phase
turbulence interaction with the particulate phase.
The fluctuations in gas/solid flow predicted by this model showed that the main
frequency of the oscillations was about 0.15 Hz in both the dilute and dense regions (e.g.,
5% and 13% of solid volume fraction). Therefore, an adequate averaging time is
necessary for comparison with the time-averaged experimental results.

18

Accepted for publication in Powder Technology

The initial presence of solids inside the riser did not have a significant effect on
the solid density distribution. The initial condition used in this simulation tended to
disappear after several seconds of simulation.
The effects of inlet and outlet design are significant in the overall solid flow
patterns and pressure drop along the riser. Real inlet and outlet conditions can be
implemented only when using a 3-D simulation with complex geometry. However,
computational time is still the major limiting factor to simulate gas/solid flow in complex
3-D geometry.

Nomenclature
2-D, 3-D - Two-dimensional, three-dimensional
C i - Instantaneous velocity vector of phase i
d p - Particle diameter
e - Particle-particle restitution coefficient
ew - Wall-particle restitution coefficient
g o - Radial distribution function
k s - Diffusion coefficient of granular temperature
P - Pressure
Ps - Solids pressure
Re - Particle Reynolds number
Ti - Stress tensor of phase i
U i - Velocity vector of phase i
- Inter-Phase drag coefficient

- Normal gradient to the wall


n

i - Volume fraction of phase i


- Specularity coefficient ( = 0.01)
s - Collisional dissipation of granular temperature
i - Viscosity of phase i
i - Density of phase I
s - Granular temperature

19

Accepted for publication in Powder Technology

Gas + Solid
Outlet

350

0.2 m

300
Solid Density (kg/m3)

14.2 m

At Time = 0
Vs = Vg = 0
g = 1.
s = 1e 05 m 2 / s 2

200
150

Us = Ug = 0.476 m/s
s = 0.4,

0.3 m
Uniform Gas Inlet
Distribution, Vg = 5.2 m/s

50
-0.10

-0.05

s = 1e 05 m 2 / s 2

0.00

0.05

0.10

Radial Position (m)

Figure 2 Solid Density Distribution


in the Riser at 3.9 m Compared
with Experimental Data.

Figure 1 Schematic Drawing of a 2-D Riser


with Inlet and Initial Conditions.
1000

8000

800

7000
Pressure Drop (Pa/m)

Solid Mass Flux (kg/m2s)

250

100

Gas + Solid Inlet

0.1 m

600
400
200
Experiment
Compuation

Experimental
Computation

Experimental
Computation

6000
5000
4000
3000
2000
1000

-200

0
-0.10

-0.05

0.00

0.05

0.10

Radial Position (m)

Figure 3 Solid Mass Flux Distribution in


the Riser at 3.9 m Height Compared
with Experimental Data.

10

12

14

Height (m)

Figure 4 Time-Averaged Axial Pressure


Drop in the Riser.

20

Accepted for publication in Powder Technology

0.02

Solid Viscosity (Pa.s)

Granular Temperature (m /s )

0.6

0.4

0.2

0.01

0.00

0.0
0.00

0.05

0.10

0.15

0.0

0.20

0.1

0.2

0.4

0.5

0.6

Solid Volume Fraction

Solid Volume Fraction

Figure 5 Granular Temperature Variation


with Solid Volume Fraction in the Riser
After 10 sec.

Figure 6 Solid Viscosity Variation with


Solid Volume Fraction in the Riser
After 10 sec.

25

Power Spectrum Density

Power Spectrum Density

0.3

20

15

10

5
0

0
0

Frequency (Hz)

Figure 7 Power Spectrum Density


Analysis of the Solids Density
Fluctuation in the Dilute Region
of the Riser (12 m Above the Inlet).

Frequency (Hz)

Figure 8 Power Spectrum Density


Analysis of Solid Density Fluctuation
in the Dense Region of the Riser
(3 m Above the Inlet).

21

Accepted for publication in Powder Technology

450

450

Experimental
Computation (10-15 sec)
Computation (10-40 sec)

350

Experimental
Computation

400
Solid Density (kg/m3)

Solid Density (kg/m3)

400

300
250
200
150

350
300
250
200
150

100
100

50
-0.10

-0.05

0.00

0.05

50
-0.10

0.10

Figure 9 Effect of Initial Conditions


on the Solid Density Distribution in
the Riser at 3.9 m Height.

7000
20-cm Riser
40-cm Riser

Pressure Drop (Pa/m)

5000
4000
3000
2000
1000
0
2

10

0.05

0.10

Figure 10 Solid Density Distribution at


3.9 m Height in the Riser with One Inlet
Compared with Experimental Data.

8000

0.00

Radial Position (m)

Radial Position (m)

6000

-0.05

12

14

Height (m)

Figure 12 Comparison of the Time Averaged


Axial Pressure in a 20 and 40-cm Riser.

22

Accepted for publication in Powder Technology

20-cm Riser

40-cm Riser

Figure 11 Comparison of the Solid Volume Fraction and Velocity


Profiles in a 20 and 40-cm Diameter Riser After 40 sec.

23

Accepted for publication in Powder Technology

References

[1] Davidson, J.R., 1961. Symposium on Fluidization-Discussion. Trans. Inst. Chem.


Eng. 39, 230-232.
[2] Jackson, R., 1963. The Mechanics of Fluidized Beds. Trans. Inst. Chem. Eng., 41,
13-28.
[3] Murray, J.D., 1965. On the Mathematics of Fluidization. Part I. Fundamental
Equations and Wave Propagation. J.F.M. 21, 465-493.
[4] Pigford, R.L. and Baron, T., 1965. Hydrodynamic Stability of Fluidized Beds.
Indust. Eng. Chem. Fundam. 4, 81-87.
[5] Soo, S.L., 1967. Fluid Dynamics of Multiphase Systems. Blaisdell Publishing Co,
Waltham, MA.
[6] Soo, S.L., 1980. Equations of Multi-Phase, Multi-Domain Mechanics. In
Multiphase Transport, Edited by Veziroglu, T.N. Hemisphere, 291-301.
[7] Chapman, S. and Cowling, T.J., 1961. The Mathematical Theory of Non-Uniform
Gases. Cambridge University Press, London.
[8] Jenkins, J.T. and Savage, S.B., 1983. A Theory for the Rapid Flow of Identical,
Smooth, Nearly Elastic Spherical Particles. J.F.M. 176, 67-93.
[9] Lun, C.K.K., Savage, S.B., Jeffrey, D.J. and Chepurnity, N., 1984. Kinetic Theories
for Granular Flow: Inelastic Particles in Couette Flow and Slightly Inelastic Particles in a
General Flow Field. J. Fluid Mech., 140, 223-256.
[10] Johnson, P.C. and Jackson, R., 1987. Frictional-Collisional Constitutive Relations
for Granular materials, with Application to Plane Shearing. J. Fluid Mech., 176, 67-93.
[11] Sinclair, J.L. and Jackson, R., 1989. Gas-Particle Flow in a Vertical Pipe with
Particle-Particle Interactions. AICHE J., 35(9), 1473-1486.
[12] Ding, J. and Gidaspow, D., 1990. A Bubbling Fluidization Model Using Kinetic
Theory of Granular Flow. AICHE J., 36, No. 4, 523-538.
[13] Gidaspow, D., 1994. Multiphase Flow and Fluidization: Continuum and Kinetic
Theory Description. Academic Press.

24

Accepted for publication in Powder Technology

[14] Kim, H.S. and H. Arastoopour, 1995. Simulation of FCC Particle Flow Behavior
in a CFB Using Modified Kinetic Theory. Canadian Journal of Chemical Engineering,
73 603-611.
[15] Louge, M., Mastorakos, E. and Jenkins, J., 1991. The Role of Particle Collisions in
Pneumatic Transport. J. Fluid Mech., 231, 345.
[16] Bolio, E.J. and Sinclair, J.L., 1995. Gas Turbulence Modulation in the Pneumatic
Conveying of Massive Particles in Vertical Tubes. Int. J. Multiphase Flow, 21 N. 6.
[17] Bolio, E.J., Yasuna, J.A. and Sinclair, J.L., 1995. Dilute Turbulent Gas-Solid Flow
in Risers with Particle-Particle Interactions. AICHE J., 41 N.6.
[18] Hrenya, C.M. and Sinclair, J.L., 1997. Effects of Particle-Phase Turbulence in GasSolid Flows. AICHE J. 43 (4), 853-869.
[19] Speziale, C.G., 1991. Analytical Methods for the Development of Reynolds-Stress
Closures in Turbulence. Annu. Rev. Fluid Mech., 23, 107-157.
[20] Bader, R., Findlay, J. and Knowlton, T., 1988. Gas/Solid Flow Patterns in a 30.5cm Diameter Circulating Fluidized Bed. Circulating Fluidized Bed Technology: II, P.
Basu and J.F. Large, eds., Pergamon Press, P. 123.
[21] Tsuo, Y.P. and Gidaspow, D., 1990. Computation of Flow Patterns in Circulating
Fluidized Beds. AICHE J., 36, 885.
[22] Sun, B., 1996. Simulation of Gas-Liquid and Gas-Solid Two-Phase Flows. Ph.D.
Thesis, Illinois Institute of Technology, Chicago, Illinois.
[23] Benyahia, S., Arastoopour, H. and Knowlton, T., 1998. Prediction of Solid and Gas
Flow Behavior in a Riser Using a Computational Multiphase Flow Approach.
Fluidization IX, Editors: L.S. Fan and T. Knowlton, Durango, Colorado, P. 493-500.
[24] Knowlton, T., Geldart, D., Matsen, J. and King, D., 1995. Comparison of CFB
Hydrodynamic Models. PSRI Challenge Problem - Presented at the Eighth International
Fluidization Conference, Tour, France, May, 1995.
[25] Patankar, S.V., 1983. Numerical Heat Transfer and Fluid Flow. Hemisphere.
[26] Spalding, D.B., 1980. Numerical Computation of Multi-Phase Fluid Flow and Heat
Transfer. Recent Advances in Numerical Methods in Fluids, edited by, C. Taylor and K.
Morgan, Pineridge Press Limited, Swansee, U.K.
[27] Fluent 4.4, 1997, User's Guide. Fluent Incorporated, Lebanon, NH.

25

Accepted for publication in Powder Technology

[28] Arastoopour, H., Pakdel, P. and Adewumi, M., 1990. Hydrodynamic Analysis of
Dilute Gas-Solids Flow in a Vertical Pipe. Powder Technology, 62, 163-170.
[29] Pita, J.A. and Sundaresan, S., 1991. Gas-Solid Flow in Vertical Tubes. AICHE J.,
37(7), 1009-1018.
[30] Pita, J.A. and Sundaresan, S., 1993. Developing Flow of a Gas-Particle Mixture in a
Vertical Riser. AICHE J., 39(4), 541-552.
[31] Gidaspow, D., Huilin, L., Neri, A., Wu, Y. and Mostofi, M.R., 1997. Turbulence,
Viscosity and Numerical Simulation of FCC Particles in CFB. Los Angeles AICHE
Annual Meeting, Fluidization Preprint.
[32] Foerster, S.F., Louge, M.Y., Chang, H. and Allia, K. (1994). Measurements of the
Collision Properties of Small Spheres. Phys. Fluids, 6 (3), 1108-115.
[33] Hui, K., Haff, P.K. and Jackson, R., 1984. Boundary Conditions for High-Shear
Grain Flows. J.F.M. 145, 223-233.
[34] Neri, A., 1998. Multiphase Flow Modeling and Simulation of Explosive Volcanic
Eruptions. Ph.D. Thesis. Illinois Institute of Technology, Chicago.
[35] Zhang, Y.F. and H. Arastoopour, (1995). Dilute Fluidized Cracking Catalyst
Particles-Gas Flow Behavior in the Riser of a Circulating Fluidized Bed. Powder
Technology, 84, 221-229.
[36] Mathiesen, V., 1997. An Experimental and Computational Study of Multiphase
Flow Behavior in Circulating Fluidized Beds. Dr. Ing. Thesis. Norwegian University of
Science and Technology, Porsgrunn.
[37] Gidaspow, D. and Huilin, L.,1996. Collisional Viscosity of FCC Particles in a CFB.
AICH J. 42, 9, 2503-2510.
[38] Gidaspow, D., 1999. Private Communications. Illinois Institute of Technology,
Chicago, IL.
[39] Mudde, R.F., Lee, D.J., Reese, J. and Fan, L.S., 1997. Role of Coherent Structures
on Reynolds Stresses in a 2-D Bubble Column. AICHE J., 43 (4), 913-926.
[40] Gidaspow, D., Huilin, L. and Therdthianwong, A., 1995. Measurement and
Computation of Turbulence in a Circulating Fluidized Bed. Fluidization VIII, Vol. 1, pp.
81-88 (preprints), Tours, France.
[41] Panton, R.L., 1984. Incompressible Flow. John Wiley & Sons, Inc.

26

Você também pode gostar