APPLICATION OF A DISPERSION MODEL FOR BUBBLE DYNAMICS
IN THREE-PHASE FLUIDIZED BEDS
S.L.P. Lee and H.I. de Lasa
Department of Chemical and Biochemical Engineering
Faculty of Engineering Science
The University of Western Ontario
London, Ontario, Canada, N6A 5B9
STRACT
The local gas hold-up and gas bubble velocity in three-phase fluidized
beds with single injector were measured by spherical head optical fiber
probes. A radial gas dispersion model was applied to analyse the bubble
dynamics in the TPFB. The gas dispersion is high in the grid region and
the dispersion coefficient decreases gradually with the axial height.
There are two different approaches for the hydrodynamic study of three-
phase fluidized beds (TPFB). The first technique (Kim et al,(1)) gives
overall gas hold-up using the pressure profile along the fluidized bed,
the total amount of solids particles and the continuity of the three
Phases. In this case, the phase hold-up is considered uniform throughout
the fluidized bed. The second method of gas hold-up measurement (Rigby
and Capes (2)) and Begovich and Watson (3)) uses in situ measuring
Probes. The advantage of the in-situ measuring probes is that detailed
information of local phase hold-ups can be obtained. The probe
measurements show that there is a distribution, instead of a constant
phase hold-ups i TPFB.
Another important fluid dynamic parameter for TPFB is the gas bubble
rising velocity. The gas bubble velocity can be obtained from either the
in situ measuring probes (Rigby and Capes (2), de Lasa et al. (4), Lee
et al. (3)) or the analysis of bubble images on a flat two-dimensional
bed (Henriksen and Ostergaard (6), Kim et al. (Z)).
Recently, Lee and de Lasa (8) combined the local gas hold-ups and bubble
velocities, measured by optical fiber probes, to test a radial dispersion
model for TPFB. This radial dispersion model was developed to describe
the gas hold-up distribution in a TPFB. The dispersion model adequately
describes the gas hold-ups throughout the colum.
In order to have a thorough assessment of the radial dispersion model,
further investigation is carried out in the present study with a single
gas injector positioned at the centre of cylindrical TPFB. The central
jet provides a well defined initial gas hold-up distribution and an
effective way for assessing the radial dispersion model applicability.
531532 FLUIDIZATION VI
RIMI ET-UI
The experiments were performed in a cylindrical fluidized bed of 0.2 m
internal diameter. Air, water and glass beads of 250 ym were used as the
gas, liquid, and solid phases. The details of the apparatus and the
experimental procedure were described in Lee and de Lasa (5). However,
a different gas injection system was used in this study. A single gas
injector of 1.25 cm0.D. was positioned at the centre of the grid. such
arrangement provides a well defined initial condition for the bubble
dynamics study. A porous metal plate was placed at the tip of the gas
injector to generate small bubbles. The pore size of the plate was 40
wm. The local gas hold-up and bubble velocity were measured by optical
fiber probes. The probe is similar to the one described in de Lasa et
al. (4) and Lee (9). In fact, the optical probe used in this study is
@ special case of the U-shaped probe discussed in Lee (9). Two optical
fibers of 400 ym core diameter were placed in parallel, as shown in Fig.
1. The two ends of the fibers were fused together to form a sphere where
the detecting of gas bubbles took place. The diameter of spherical tip
is approximately 1 mm. This spherical head probe has two advantages
over the U-shaped probe. The first is that the size of the probe is
smaller. The other advantage is that it is easier to fabricate. The
characteristics of the probe signal is the same as that of a U-shaped
Probe. However, the baseline of the signal from the new probe is very
close to zero. Therefore, The baseline can not be used to assess the
solid hold-up in the TPFB.
TA, SIS
A cross correlation technique was applied to measure the average bubble
rising velocity. Two optical fiber probes aligned vertically with a
separation of 0.94 cm were used in this study. The normalized cross-
correlation function (NCCF) was used to measure the time delay between
the signals from the two probes:
and Ry 7 (Cy )1/ yay «
Agy() - £ flxces - Hy) (y(etr) -p Jat seen eee eeeeeeee (2)
where x(t) and y(t) are the signals from the lower and the upper probes
respectively. A time delay r, can be obtained by noting the time shift
at the peak of the cross cofiélogram. The average bubble velocity is
calculated by dividing the distance between probes by the time delay
Tmax: The bubble velocity can also be obtained from the cross-spectral
density function (CSDF). The CSDF of the two signals is obtained from
the Fourier transform of the NCCF
G62) = 2 Jao exp(-J2REr) Or ee eeeeeeeeeeee @)
and GyC£) = GLE) = 4 Qg(£) ...
where
Cy (£) = 2 Fanon cos (2nfr)drS. L. P. Lee and H. I. de Lasa 533
QE) = 2 fat SinQQefr)dr oo oceccccceeeseseeeeeee (6)
0
and
Gy(£) = tan (Qy(E)/Cyy(£)) qa)
The time delay is obtained from r = 4,,(£)/2m£. Consequently, the average
bubble velocity at frequency f can be calculated.
MODEL DEVELOPMENT
The following model was developed for the study of gas dispersion in TPFB
by Lee and de Lasa (8).
ae aiid ae
Vg 7 5 Go rae, BRE ae aera att eta eta ate (8)
with boundary conditions
a &g06%y) Vz = Vgq (lg) veceeeeeee een ees (9)
(deg) / (ar,)] ccc renee cence eee ee ences (10)
Cy TO rece eee tee ete e teen eet e teeta eect anaes ay
Equation 8 can be solved by separation of variables to give :
fy Ey Mn FolGaty) exP(-L 2B agl/ V)eceeeeeeeeeeeeeees a2)
where the coefficients A, are
1
J 7 ego(Xq) Jolanty ary
anu 25
: Fy Tolenty) expC-[a 2 D zy1/ Vio) Sry
where J, is the Bessel function of the first kind of order zero and
are the positive zeros of J,(a,) +
The gas hold-up, ¢,, and the average bubble velocity, V,, were measured
at various axial and radial positions in the TPFB. The experiments were
carried out at 0.225, 0.375 and 0.525 m above the distributor. At each
axial level, the measurement were performed at nine radial positions, r,
= 0, 40.25, +0.5, #0.75 and 40.875. Various operating conditions used
in this investigation are listed in table 1.
The gas hold-up was defined as the fraction of total measuring time when
the probe was in contact with gas bubbles. The average bubble velocity
was assessed through the cross-correlation technique. Once the required
experimental data were collected, the dispersion coefficient, D, was
obtained by the Marquardt method of optimization (10). Note that both
the gas hold-up, ¢,, and the average bubble velocity, V,, were assessed
through analyses of the same time series signals. The data analysis was
performed to obtain the dispersion coefficient at each individually axial
height under various operating conditions.SM FLUIDIZATION VI
SIO!
The average bubble velocities were assessed through the analyses of the
cross correlation and the phase angle calculation. The results show good
consistency between the two methods. In addition, the phase angle
analysis indicates that the average bubble velocity is independent of
bubble frequency. The results on bubble velocity shown here were
obtained from the phase angle analysis
The profiles of bubble rising velocities in the fluidized bed were found
to be parabolic. Typical average bubble velocities in the TPFB are shown
in Fig. 2 and Fig. 3. Since single gas injector was used, bubbles tond
to rise in the centre of the column. A flow pattern of upward centre
graft and downward annular flow was established. Therefore, bubbles in
the centre received extra lift and those bubbles in the annular Yegion
were hindered by the circulation. This phenomenon is more pronounced as
gas velocity is increased. For example, bubble velocities in the centre
region at low gas velocity were in the range of 40 to 60 cm/s, shown in
Fig. 2. Bubble velocities of 80 to 100 cm/s were observed at high gas
velocity in the case shown in Fig. 3. Furthermore, the radial bubble
velocity profiles in the grid region are narrower than those at higher
axial heights.
Examples of the gas hold-up distribution in the fluidized bed are shown
in Fig. 4 and Fig. 5. The gas hold-up profile was very steep in the grid
region due to the jet effect. Most of the gas bubbles were found in che
centre of the column in the range of r, = £0.25. The gas hold-up
decreased sharply to a very low level at midway, r, - 40.5, and very few
bubbles were observed from there to the wall. At higher axial positions,
the gas hold-up profiles broadened from delta to parabolic. In any case,
near zero gas hold-up was observed close to che wall of the column. This
confirmed the boundary condition at r, = 1 in Eq.(11). The gas hold-up
profiles were found to be symmetric although a little imbalance vas
observed in the grid region.
The relationship between the gas dispersion coefficient and the
dimensionless bed height are shown in Fig 6 to Fig. 8. The dispersion
coefficient was quite large near the grid and decreased gradually as the
axial height increased. The standard deviations of gas hold-ups
calculated from the dispersion model are listed in table 1 for
experiments performed under various operating conditions. Some examples
of model estimations in comparison with experimental data are shown in
Fig. 4 and Fig.5. The model prediction matches quite well with the
experimental data
The values of gas dispersion coefficient in the present study are higher
than those obtained in fluidized beds with multiple injectors (Lee and
ge Lasa (8)). Since the injectors were not far apart (6.7 em
separation), the dispersion effect was diminished by the interference
between injectors. Furthermore, the interference increased with the
axial height due to the conical projections of the injectors. Therefore,
the dispersion coefficient decrease rapidly with axial height in multiple
injector system.