Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Study of the catalyst distribution in a new designed downer distributor using ECT and
CFDDEM method
Tong ZHAO, Masahiro TAKEI
Department of Mechanical Engineering, College of Science and Technology, Nihon University,
1814, Kanda Surugadai, Chiyodaku, Tokyo 1018308, Japan
polomarslee @hotmail.com
Keywords: Electrical capacitance tomography; Discrete element method; Computational fluid dynamics; Downer
Abstract
An Electrical Capacitance Tomography (ECT) system has been developed to noninvasively measure the particle distributions
in the downflow circulating fluidized bed (downer) with a new designed distributor. The particle concentration distribution
images are obtained under certain parameter combinations. The influence factor of particle distribution and the fluctuation
profiles of particle volume fraction are discussed in details. The increase of particle flow rate played a negative part in the
uniformity of particle concentration. Based on the reconstructed image obtained at different position, the axial and radial
profiles of particle volume fraction and its fluctuation were put forward. The effect of the airparticle mixer in the particle
uniformity was clarified by comparison to the former researcher's working results. As results, the new designed airparticle
mixer was proved to lead to a more uniform particle distribution in downer. Moreover, a numerical model of this downflow
fluidized bed was developed by means of the combined two dimensional model of computational fluid dynamics and discrete
element method (CFDDEM). This CFDDEM model provides the information regarding the particle movement and
distribution, gas velocity, gas holdup, and conversion profiles in the bed under different airparticle mix condition. Specifically,
the particle volume fraction is proportional to air flow rate ratio and decreases as the measurement position moves downstream.
The simulation results show very good agreement with the experiment.
Introduction
In the past decades, due to the significant advantages of the
downflow fluidized bed (downer) such as uniform particle
distribution and short contact time between gas and particles,
downer has been proposed for some processes such as fluid
catalytic cracking process, where extremely short but
uniform contact between gas and solids are required to
prevent overreacting. Various researches have been carried
out in the past to study the flow behaviors in the downflow
fluidized bed (downer) such as gas and solid velocity
distribution (Wang et al., 1992), solids aggregation (Zhang
et al., 2008) and solid concentration distribution (Zhang et
al., 2003; Wu et al., 2007). However, most of these studies
have focused on the fully developed region (the constant
velocity region), the gas and solids flow patterns in the
entrance region of the downer (less than 2m from the
downer inlet), which is very important to the particle
distribution uniformity, have been ignored. Therefore, in
order to fully understand the mechanisms of flow
development, it is important to clarify the particle
distribution behaviors within the entrance region of the
downer.
Electrical Capacitance Tomography (ECT) which has
become increasingly popular for multiphase flow
measurement should be a proper solution to visualize the
particle distribution (Zhao et al., 2007; Tortora et al., 2006).
Recently, the ECT technique has been successfully applied
to many industry processes, such as fluidized bed
coalescence (Wang et al., 1995), pneumatic conveyance
(Dyakowski et al., 1999) and particle coating process (Takei
and Zhao, 2009). Moreover, the numerical simulation of
the gassolid distribution is very necessary, in order to
confirm the accuracy of the experiment results. The discrete
element method (DEM), which has been proved to be an
effective simulation technique to study the particlefluid
flow systems under different conditions, as briefly reviewed
by various investigators (Yu and Wu, 2003; Deen et al.,
2007), is widely used. In the DEM model, the motion of
particles is modelled as a discrete phase, described by the
Newton's laws of motion on an individual particle scale,
while the flow of fluid (gas or liquid) is treated as a
continuum phase, described by the NavierStokes
equations.
In this work, an ECT system, which consists of three
270mm CT sensors, is designed for particle distribution
visualization in a downer with a new designed distributor.
The distribution images of particle volume fraction were
obtained at a 10 milliseconds intervals for different
parameter combinations. After that, the axial and radial
profiles of particle volume fraction information were
extracted from the images. A numerical simulation of the
particle behaviour in the downer was undertaken using
DEM. The simulation data were then compared with the
experiment results.
Mathematical Modelling
In order to confirm the experiment result, a
mathematical model was presented. In this mathematical
model, the solid phase is treated as a discrete phase that is
described by a conventional DEM. Movement of
Paper No
individual particles is evaluated by the Newton's equation
of motion which includes the effects of gravitational force,
contact force, and fluid force. The translational and
rotational motions of a particle at any time, t, in the reactor
are determined by momentum balance, given by
dv
m = Fa + F+G (1)
dt
do)
I =T (2)
dt
where m is the mass of the particle; v and o is the
translational and rotational velocity; I is the moment of
inertia; F, is the force acting on particle that exerted by
surrounding fluid; Fc is the contact force; G is the
gravitational force; T is the torque caused by the contact
force and the moment of inertia of particle. The contact
force between two spherical particles can be modelled by
the simple concept of spring, dashpot and friction slider.
Thus the model depends on the parameters of stiffness,
dissipation, and friction coefficients which can be obtained
from the physical properties of the particles. The details of
the applications of this model in gassolid systems were
illustrated in previous work (Tsuji et al., 1992; 1993).
The gas phase is treated as a continuous phase and
modelled in a way similar to the one used in the
conventional TFM. The governing equations are the
conservation of mass and momentum in terms of the local
mean variables over a computational cell, given by
(3)
+V.(at)=0 (3)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
width. In the downer, gas and particles flow downward,
same as the gravity. A noslip condition is used for the air
phase at the walls and particles are allowed for frontal
collisions with the wall. The simulation is started with the
random generation of particles without overlaps in the
rectangular bed, followed by a gravitational settling process
to form a packing. Then, gas is injected into the bed
uniformly to fluidize the particles. After a short time,
macroscopically steady flow can be established. The results
presented in this paper are all obtained in such a state.
VP + F
where e is the volume fraction; u is the velocity vector of
fluid; P is the pressure of fluid; Fa is the fluid drag force
exerted on the particles, and 6V is the volume of a
computational cell.
The fluid drag force acting on each particle inside the
computational cell can be calculated by
f,, = (vP,, u) V (5)
The coefficient ( can be determined by the Ergun's equation
( < 0.8 )or Wen and Yu's equation ( > 0.8 ). In the
present simulation, 3 was deduced based on the summarized
equations in previous works as follow (Tsuji et al., 1993;).
Simulation Conditions
The parameter settings of the simulation are summarized in
Table 1. The simulated fluidized bed consists of a special
designed distributor and a rectangular container. Figure 1
shows the calculation domain and the grid arrangements. In
this simulation, the particle distribution behavior in downer
is numerically simulated under the same conditions as those
for the experiment. The parameter settings of the simulation
are summarized in Table 1. Because of the huge particle
number and limited computation speed, the particle
diameter is set as 3 mm, which is around 50 times to the real
particle diameter. The flow of gas and particles is assumed
to be twodimensional since the thickness of the bed is equal
to the particle diameter, which is much less than the bed
(a) Front view
212 mm
S122 mm
Particle inlet
(b) New designed distributor
Figure 1: Calculation domain and grid arrangement
.........
.........
.........
.........
.........
oil 11111111111111111
EM iii i
Paper No
Table 1: Simulation conditions
Gas phase Particle phase
Fluid type Air Particle shape spherical
Density(kg/m3) 1.205 Density(kg/m3) 1200
Bed 0.27 Particle 3 103
geometry(m) (diameter) diameter(m)
2 (length) Number of 107
particles
Air flow rate at
each nozzle 0.118 Spngconstant 80
(m3/s) (N/m)
Viscosity(kg/ms) 2.0x10 Particle flow ra 175
[kg/m2s]
Friction coefficient 0.3
Time step (s) 5 x105
Experiments Using ECT
The experimental equipment consisted of a hopper tank, a
sender, an airparticle distributor, a circulating pipe, a
cyclone separator, and a receiver tank, as shown in Figure 2.
The downer which connected to the bottom of the distributor
had a diameter of 270 mm and a length of 5.3 m. Particle
was supplied from the hopper tank to the distributor inlet.
The distributor consists of one annular particle inlet and five
air nozzles, which include a centre nozzle and four
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
welldistributed side nozzles. The annular solid inlet has an
inside diameter of 122 mm and an outside diameter of 212
mm. The air nozzles are 24 mm in diameter. The angle
between the centre line of the centre nozzle and the side
nozzle is 45 degrees. Three capacitance tomography sensors
were wrapped around the circumference of the downer at
different vertical positions. The length of each sensor is 0.66
m. Figure 2 shows the schematic diagram of the ECT sensor.
The details of the ECT sensor have been reported in the
authors' former research paper (Zhao et al., 2007). The
particulate solids used for this study were fluid catalytic
cracking (FCC) catalyst particle that has a particle density of
1,200 kg/m3, a mean diameter of 69.6 pm and a relative
permittivity of 2.7.
In the experiment, the hopper was adjusted to provide a
particle flow rates as 175 kg/m2s. The air flow rate at each
air nozzle was set as 0.118m3/s. After a few seconds delay,
in order to allow the particle to flow in a stable manner, the
capacitances measurement started at bases time using a
combined system consist of a capacitance acquisition device
and a highspeed multiplexer. The time interval to acquire
the capacitances measured by 66 pairs of electrodes in a
crosssection was At = 10.0 ms. The total measurement time
is 5s, and the total frame number of the reconstructed image
Nt is 500. The Generalized Vector Sampled Pattern
Matching (GVSPM) method was used for image
reconstruction (Takei et al., 2006).
Figure 2: Experimental apparatus
Air nozzle 1 106
(Side) x4
Air nozzle
(Center)
270
(Dimensions: mm)
Configuration of the specially
designed distributor
Paper No
Results and Discussion
Figure 3 shows the three dimensional particle distribution
image of case 1 at t0.2s as an example. The dash lines in
figure 3 indicate the positions of ECT sensor in the
experiment. Then, the distribution images of particle volume
fraction at the crosssection of these positions were extracted
and compared with the experimental images obtained by
ECT sensors.
Table 2 shows the distribution image of particle volume
fraction obtained by the ECT system and DEM simulation.
In these images, the red pixels indicate the high particle
volume fraction, 0.2, while the blue one indicates air as
shown by the colour bar. As shown in Table 2, both the
simulation and experiment result presents that near the
entrance of the downer, the high particle concentration part,
which probably represents the solid aggregation (cluster),
always exists in the near wall region and the centre of the
downer. Hence, an annular region of low particle volume
fraction can be observed between the centre area and the
wall. And it seems that the low particle volume fraction
region located at the same place as particle inlet. Then, as the
position moves downstream, due to the force affect by air,
the clusters become smaller or even disappear. In other
words, the particle distribution becomes uniform. The above
radial profiles of particle volume fraction are very
reasonable. By means of the new designed distributor, the air
velocity was inclined with respect to the gravity direction
because of the side nozzle. After dropping from the inlet,
particles are accelerated in not only the axial direction, but
also in the radial direction. Under the affect of this radial
force, particles move away from the annular region to the
center or the wall. As a result of these trends, an annular
region of low particle volume fraction and two regions of
relatively high particle volume fraction in the center and near
the wall are formed.
Figure 4 reveals the experiment and simulation results of the
axial particle volume fraction profile. In figure 4, the
simulation result of particle volume fraction decrease with
the increase of the axial distance from distributor. The
reason of this phenomenon could be lies on the slip velocity
between particle and air phase. Wang et al. put forward that
there are three sections of particle velocity in solid air two
phase down flow, which is the first acceleration section, the
second acceleration section and the constant velocity section
(Wang et al., 1992). Following this theory, along with the
particle drops down, the particle velocity will increase first.
Then, when the air upward drag becomes equal to the
particle gravity, particle velocity will level off. However,
during this process, the air velocity nearly keeps on constant.
Therefore, the slip velocity between particle and air phase
increases as particle drops downstream, and then cause the
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
decrease of particle volume fraction. As shown in Figure 4,
the experiment results also show the same tendency as the
simulation result. However, the experiment result is
obviously large than the simulation one. This phenomenon
might be due to the magnification of the particle size. The
magnification of particle size will influence the drug force
acting on particle exerted by surrounding fluid and the
contact force exerted by other particles, and then cause the
increase of particle velocity. Hence, the particle velocity in
the simulation is faster than that in the experiment at the
same axial position. It means that the slip velocity in the
simulation is bigger than that in the experiment at the same
axial position, and causes the smaller particle volume
fraction.
h=0.33m
.pl
     t
h=0.99m
h=1.65m
Figure 3: Three dimensional particle distribution (t=0.2s)
Paper No
 Simulation U Experiment
0.3 0.5 0.7 0.9 1.1
1.3 1.5 1.7 1.9
Distance from the distributor h [m]
Figure 4: Axial solid fraction profile from simulation and experiment
Conclusions
A threesensor ECT system was applied to visualize the
particle distribution during the particleair mixing process in
the downer. Based on the reconstructed image, both the axial
and radial profiles of particle volume fraction are analyzed in
detail and clarified. Then, in order to confirm the experiment
results, a numerical simulation of the gassolid distribution
in the downer is presented using discrete element method
(DEM). The comparison results of the simulation and the
experiment are summarized as follow. The distribution
images of particle volume fraction from simulation agree
well with the experiment ones. The highest particle volume
fractions could be observed near the wall. And an annular
region of relative low particle volume fraction, which
coincides with the particle inlet area, can be observed near
the entrance of the downer. The reason could be lies on the
radial drag force imposed by the air phase. And as particles
descend, the particle volume fraction decreases but become
well distributed.
Acknowledgements
The authors wish to acknowledge the financial support
provided by the Information Center of Powder Technology
Japan, and the Grantinaid for scientific research (C) from
Japan Society for the Promotion of Science. The authors
wish to acknowledge the support provided by the Rflow Co.
Ltd in the DEM simulation.
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Table 2 Distribution images of particle volume fraction
0.0 0 .20
Relative solid concentration
I~
Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Y. TSUJI, T. KAWAGUCHI, T. TANAKA, Discrete
particle simulation of twodimensional fluidized bed,
References Powder Technology 77, (1993) 7987.
Z.W. Wang, D.R. Bai, and Y. Jin, Hydrodynamics of
concurrent downflow circulating fluidized bed (CDCFB),
Powder Technology 70 (1992) 271275.
M.H. Zhang, K.W. Chu, F. Wei, A.B. Yu, A CFDDEM
study of the cluster behavior in riser and downer reactors,
Powder Technology 184 (2008) 151165.
M.H. Zhang, Z. Qian, H. Yu, F. Wei, The solid flow
structure in a circulating fluidized bed riser/downer of
0.42m diameter, Powder Technology 189 (2003) 4652.
B. Wu, J.X. Zhu, L. Briens, H. Zhang, Flow dynamics in a
fourinch downer using solids concentration measurements,
Powder Technology 178 (2007) 187193.
T. Zhao, M. Takei, K. Masaki, R. Ogiso, K. Nakao, and A.
Uchiura, Sensor design and image accuracy for application
of capacitance CT to the petroleum refinery process, Flow
Measurement and Instrumentation 18(56) (2007) 268276.
P.R. Tortora, S.L. Ceccio, T.J. O'Hern, S.M. Trujilo, J.R.
Torczynski, Quantitative measurement of solids distribution
in gassolid riser flows using electrical impedance
tomography and gamma densitometry tomography,
International Journal of Multiphase Flow 32 (2006), 972995
S.J. Wang, T. Dyakowski, C.G. Xie, R.A. Williams, M.S.
Beck, Real time capacitance imaging of bubble formation at
the distributor of a fluidized bed, Chemical Engineering
Journal 56 (3) (1995) 95100.
T. Dyakowski, S.P. Luke, K.L. Ostrowski, R.A. Williams,
Online monitoring of dense phase flow using real time
dielectric imaging, Powder Technology 104 (1999) 287295.
M. Takei, T. Zhao, Measurement of particle concentration in
powder coating process using capacitance computed
tomography and wavelet analysis, Powder Technology
193(1) (2009), 93100
A.B. Yu, B.H. Xu, Particlescale modeling of gassolid flow
in fluidization, Journal of Chemical Technology and
Biotechnology 78 (2003) 111121.
N.G. Deen, M.V.S. Annaland, M.A.Van der hoef, J.A.M.
Kuipers, Review of discrete particle modeling of fluidized
beds, Chemical Engineering Science 62 (2l' r'_'44.
M. Takei, GVSPM image reconstruction for ECT images of
powders in a vertical pipe and comparison with the
conventional method, Measurement Science and Technology
17(8) (2006) 21042112.
Y. TSUJI, T. TANAKA, T. ISHIDA, Langrangian numerical
simulation of plug flow of cohesionless particles in
horizontal pipe. Powder Technology 71, (1992) 239250.
