You are on page 1of 12

Chemical Engineering Science, Vol. 53, No. 10, pp.

1835—1846, 1998
( 1998 Published by Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0009–2509(98)00044–X 0009—2509/98 $19.00#0.00

Verification of Eulerian simulation of


spontaneous bubble formation in
a fluidized bed
A. Boemer,* H. Qi and U. Renz
Lehrstuhl für Wärmeübertragung und Klimatechnik der RWTH-Aachen Eilfschornsteinstr. 18,
D-52056 Aachen, Germany

(Received 6 September 1997; accepted 26 September 1997)

Abstract—A computer code to simulate the fluid dynamics of fluidized beds with the Eulerian
approach is being developed. To verify the results, experimental investigations were carried out
at a two-dimensional lab-scale fluidized bed. The local-time-dependent solid volume fractions
were measured with a video system. Comparison of the simulation with these experiments and
various empirical models found in literature are presented. ( 1998 Published by Elsevier
Science Ltd. All rights reserved.

Keywords: Bubbling fluidized beds; gas/solid flows; Eulerian simulation; granular theory.

1. THE MODEL with


In gas—solid fluidized beds, the particle concentra- s "2k*D #(j*!2 k*) ) tr(D )I (4)
i i i i 3 i i
tion is usually too high to handle each particle separ-
D "1 [+v #(+v )T] . (5)
ately. By using the Eulerian approach, the particulate i 2 i i
phase can be treated as a continuous fluid if its prop- A balance of the solids fluctuating energy is necessary,
erties are described with the kinetic theory of granular whereas gas turbulence can be neglected. The result-
flow. In literature exist different physical models to ing granular temperature # , defined by
s
perform this task. They were investigated and com-
pared in other papers (Boemer et al., 1995; Boemer 1 m v@2"3 m # (6)
2 s s 2 s s
et al., 1997). The present contribution has to be seen as is an important parameter of the models derived from
a continuation of these papers, while a more detailed kinetic theory of granular flow. It can be calculated by
discussion of the complete subject can be found in the partial differential equation
Boemer (1996). The following set of equations was

C D
found to be most promising to simulate bubbling 3 L
(e o # )#+ ) (e o # )v
fluidized beds. 2 Lt s s s s s s s
Mass balance of each phase i (gas or solid):
"(!p*I#s ): +v #+ ) (k*+# )!c*#'* . (7)
s s s # s # #
L
(e o )#+ ) (e o v )"0 . (1) In these equations, there are physical models to be
Lt i i i i i
inserted. In Boemer et al. (1996) it was concluded to
Momentum balance of the gas phase: use the following relations which are given in the
appendix:
L
(e o v )#+ ) (e o v v ) Drag function b of Syamlal et al. (1993).
Lt g g g g g g g z
z Solid pressure p of Lun et al. (1984).
s
"+ ) s #e o g!e +p!b*(v !v ) . (2) Solid shear viscosity k of Gidaspow et al. (1992) in
g g g g g s
z
s
the dilute and intermediate region, and the model
Momentum balance of the solid phase:
derived from soil mechanics by Sokolovski (1965),
L proposed by Syamlal et al. (1993), in the extremely
(e o v )#+ ) (e o v v )"+ ) s #e o g dense region. Without including interparticle fric-
Lt s s s s s s s s s s
tion by means of the soil mechanics model the
!+p*!e +p#b*(v !v ) (3) simulated behavior of the bubbles in the bed is
s s g s
unrealistic.
z Solid bulk viscosity j of Lun et al. (1984)
s
z Granular temperature diffusion function k of
#
*Corresponding author. Gidaspow et al. (1992).
1835
1836 A. Boemer et al.
z Granular temperature dissipation c of Jenkins and correlation of the gray scale of a video pixel and the
#
Savage (1983). solid volume fraction at this location.
z Granular temperature exchange function ' of
#
Ding and Gidaspow (1990). 3. INVESTIGATION OF SPONTANEOUS BUBBLE
z Radial distribution function (needed in the models FORMATION
listed above) of Ding and Gidaspow (1990).
The model described in Section 1 has been imple-
This set of equations has been used to simulate the mented into the FLUENT code, which is presently
process of spontaneous bubble formation in a two- developed in co-operation with Fluent Europe Ltd.
dimensional fluidized bed. The apparatus described in Section 2 was divided into
59]80 computational cells. Both at the inlet and the
outlet a zero-gradient boundary condition has been
2. EXPERIMENTAL SETUP applied for all variables. At the walls adhesion is
In addition to the numerical simulation, the process assumed for both phases, whereas solid pressure and
has also been investigated experimentally. The test granular temperature have a zero-gradient. The max-
case geometry applied here is similar to that used by imum solid volume fraction is set e "0.6. De-
s,.!9
Kuipers (1990). The fluidized bed is 0.57 m wide and pending on the convergence behavior, time steps be-
1 m tall. Front- and backwall are made out of glass. tween 5]10~5 and 2]10~4 s were applied. Anyhow,
To ensure two dimensionality the apparatus is only simulation of 1 s real time needs between 50 and
0.015 m deep. It is vibrated with 60 Hz to reduce wall 200 CPU hours on a HP Apollo 735 workstation.
effects. The bottom is a porous plate. A superficial
gas velocity of 0.676 m/s, which is twice the mini- 3.1. Solids distribution in and around bubbles
mum fluidization velocity, is applied. The apparatus is In literature, different models are proposed for the
initially filled up to 0.5 m with glass beads distribution of particles in and around bubbles in
(o "2545 kg/m3, d "692 km) belonging to Geldart- a gas—solid fluidized bed. All of them are derived from
s s
group D. This fluidized bed is situated in a completely measurements in different systems and do not claim
dark room and homogeneously illuminated with generality. In Fig. 3 three models are compared with
strong halogen floodlights via a screen from the back. results from similar bubbles obtained in experiment
The process of bubble eruption was recorded with and simulation. Most of the experimental data are in
a video camera from the front. the range predicted by the models. The simulation
To quantify the local solid volume fractions from predicts slightly lower solid volume fractions com-
the video images a calibration system was con- pared to the models, but the gradients Le /Lr are
s
structed. It consists of a chamber (shown in Fig. 1) similar. Above e "0.5 the measured gradients seem
filled with various well- defined amounts of particles. s
too steep which could be due to the weak sensibility of
It was possible to disperse these particles uniformly the calibration method in this region (see Fig. 2). In
within the chamber by vibration and fluidization. To the model of Yates et al. (1994) the actual solid volume
ensure identical optical conditions, the fluidized bed fraction in the emulsion cannot be adapted. They
was removed from its mounting and replaced by the fixed e "0.523, whereas in this test case 0.6 is
calibration device. Access to the pixel gray scales and s,.!9
correct. A serious problem of the simulation is the fact
their manipulation was possible by using the NIH- that the solid volume fraction inside the bubble is too
Image package. The resulting calibration curve is high. Either the forces which carry particles out of the
shown in Fig. 2. In the dense region (e '0.3) the bubble are not correctly described or the effect is
s
resolution becomes poor since most of the light a numerical artefact. The steep gradients of solid vol-
gets absorbed in the bed. However, there is a clear ume fraction at the bubble border or the bed surface
are a general issue for the discretisation method, and
the desirable fine resolution of the mesh is limited by
the acceptable computation time.
To distinguish between bubbles and emulsion in the
following sections, a solid volume fraction of e "0.2
s
is chosen. As visible from the gradients in Fig. 3, the
exact value of this definition is not crucial. Gidaspow
(1994) also used e "0.2, whereas Kuipers (1990)
s
chose 0.15.

3.2. Bubble wake


Bubbles which rise in a fluidized bed faster than
the gas in the emulsion may have a wake. This ef-
fect was observed in both experiment and simula-
tion. Fig. 4 shows a comparison of two bubbles of
similar size. The resolution of this bubble in the simu-
Fig. 2. Calibration curve. lation is about 18]18 cells and the grayscales are
Eulerian simulation in fluidized bed 1837

Fig.1. Calibration device.

Fig. 4. Wake of two similar bubbles.


1838 A. Boemer et al.
Eulerian simulation in fluidized bed 1839

Fig. 3. Lateral solids distribution at the bubble center.

Table 1. Wake fraction and angle

Wake fraction Wake angle a


W
Minimum (%) Mean (%) Maximum (%) Minimum (°) Mean (°) Maximum (°)

Rowe und Partrige (1965) 15 27 34 110 134 180


Experiment 15 22 29 91 107 126
Simulation 16 25 34 93 102 123

smoothed. Again, too many particles can be ob- formed in order to be compared with our simulation.
served to be present inside the simulated bubble. Such a comparison is depicted in Fig. 5. Two bubbles
Besides, the agreement between experiment and simu- have been chosen, a small one in the lower region
lation is good. This observation is supported by (Bubble 1) and a larger one close to the bed surface
the data listed in Table 1. Both wake fraction and (Bubble 2). The results are related to the absolute
wake angle are similar in experiment and simulation pressure at the bubble center obtained from the simu-
and, furthermore, agree with the data given by Rowe lation.
and Partridge (1965). It can be concluded that the The agreement between simulation and experiment
model is able to predict the effect of bubble wakes is good. Differences can be explained by the influence
correctly. of neighbor bubbles which did not occur at the experi-
ment. The models assume constant pressure inside the
3.3. Pressure distribution in and around bubbles bubbles which is confirmed by simulation and experi-
One bubble-model frequently quoted in literature is ment. The difference between the two models is small.
that of Davidson and Harrison (1963). Starting from Both models agree with simulation and experiment in
mass and momentum balances and using simplifying the surrounding of the bubble. Due to the underlying
assumptions, they derived an algebraic equation for assumptions they disagree at greater distances from
the hydrostatic pressure distribution around isolated the bubble.
bubbles in fluidized beds. Collins (1989) extended
their model to include the intermediate region be- 3.4 Bubble size
tween bubble and emulsion. Further, there exist In a fluidized bed, tiny bubbles start at the distribu-
measurements of the pressure distribution at isolated tor and then rise into the bed. On their way upwards
bubbles in a two-dimensional bed published by Litt- they grow due to entrained gas from the emulsion and
man and Homolka (1973). Their results have been coalescence with other bubbles. Caused by the in-
normalized with the respective bubble diameter and creasing diameter, they become faster with increasing
the position in the bed, and can thus easily be trans- level in the bed. In literature, there exists a vast
1840 A. Boemer et al.

Fig. 5. Hydrostatic pressure on a vertical line through the bubble center.

Fig. 6. Bubble diameter.

amount of models to describe these phenomena. In agrees better than their 2D model. Their models have
this paper, only models explicitly proposed for two- the unrealistic feature of a non-zero bubble diameter
dimensional beds are taken into account. at the distributor level. If these curves would be shif-
Concerning the bubble size, the only models valid ted downwards to the origin, the results would agree
for 2D beds with Geldart-D particles are those of better.
Hilligardt and Werther (1986) and Lim et al. (1993).
Together with results of our experiment and simula- 3.5. Bubble rise velocity
tion they are shown in Fig. 6. In this figure, the The velocity of the rising bubbles was calculated by
relative wide range of scattering in both simulation tracking the motion of the bubble centers manually.
and experiment attracts attention. However, this For this purpose, frames provided by the video
seems to be unavoidable and is reported in other camera every 0.04 s and graphical outputs produc-
studies, e.g. Hilligardt and Werther (1986), too. Obvi- ed by the computer code at the same interval were
ously the local flow conditions may considerably af- used. Since the bed was only 0.5 m heigh, the ascent
fect a bubble (compare Section 3.6). of bubbles is quite short and the evaluation thus
The agreement between simulation and experiment difficult.
is good. In the simulation, bubbles were detected The models found in literature for two-dimensional
above a level of 0.1 m, only, whereas in the experiment beds are depicted in Fig. 7. Again, a strong scattering
they were found directly above the distributor. The was found for both, experiment and simulation. Be-
model of Lim et al. (1993) matches these data. Surpris- side the formula presented by Pyle and Harrison
ingly, the 3D model of Hilligardt and Werther (1986) (1967), they also published the data-set. The respective
Eulerian simulation in fluidized bed 1841
scattering is shown in Fig. 7 by the shaded area. It is in pared their own model with the work of Miwa et al.
the same order of magnitude valid for our data. (1972) and Allahwala (1975). The latter consists of
For small and medium bubbles, the agreement be- empirical equations, whereas the others are derived
tween experiment and simulation is quite good, but theoretically under similar simplifications as the
the large bubbles observed in the experiment are model of Davidson and Harrison (1963) mentioned
much faster than the simulated ones. Obviously near above.
the bed surface the bubbles are considerably disturb- Strictly speaking, all these model are valid only for
ed. The formulas of Collins (1965) and Cranfield and single bubbles produced by a jet under ideal fluidiz-
Geldart (1974) predict too slow bubbles. The same is ation conditions. In the testcase investigated in this
valid for the formula given by Pyle and Harrison paper such undisturbed coalescence did not appear.
(1967), while their range of scattering includes some of Furthermore, the height of the bed is only 0.5 m and
our data. However, for small bubbles all models agree thus too short for clear observation of coalescence.
well with experiment and simulation. However, from both experiment and simulation one
According to the suggestion of Davidson and example of coalescence has been selected to be com-
Harrison (1963), the results of models derived from pared with these models. In Fig. 8, the example from
investigations of isolated single bubbles are correc- the simulation is shown by means of the contour e "
s
ted by adding º !º , that is 0.338 m/s here. Look- 0.2. The corresponding relative velocities are depicted
o mf
ing at Fig. 7, this correction appears reasonable since in the left part of Fig. 9. The right hand side of Fig. 9
without it the models predict too slow bubble vel- shows the example chosen from the experiment. The
ocities. If the Froude number depicted relative bubble distances are defined:
u
"6" *y
Fr" (8) *y " "6" . (9)
Jg ) d "6",3%- R #R
y "6" "6",1 "6",2
is plotted vs the bubble diameter, for the calculation The initial mean radii of the bubbles inserted into the
as well as the experiment a more or less constant value model equations are listed in Fig. 9. Beginning at state
can be seen. This agrees with the findings of Ellenber- (3), the lower bubble 2 gets accelerated and streams
ger (1995). into the upper bubble 1. During the process, bubble
1 becomes flattened while bubble 2 elongates. This
3.6 Bubble interaction behavior has been described by Clift and Grace
Considerable interaction of bubbles can be ob- (1985), too. At state (7), *y is not yet equal to 1,
"6",3%-
served in technical fluidized beds. A bubble may grow but the chosen contour already encloses both bubbles
either by absorbing gas from a neighbor bubble or and the relative velocity starts to decrease. Due to the
due to coalescence with another bubble. The latter has deformation of the bubbles the radii differ from the
been intensively investigated by Clift and Grace initial ones taken to calculate *y . The further
"6",3%-
(1985). Basis of their models is the thesis that ‘the process of coalescence takes place at solid volume
velocity of a bubble may be approximated by adding to fractions lower than 0.2 and can thus not be seen in
its rise velocity in isolation the velocity which the con- Fig. 8. The simulated tendency of increasing relative
tinuous phase would have at the position of the bubble velocity with decreasing bubble distance agrees with
nose if the bubble was absent’. For the velocity in the models, while the absolute values are partly higher
isolation they suggest to insert the formula of Pyle than the model’s predictions. In the experiment, the
and Harrison (1967), see Section 3.5. They com- clearest process of bubble coalescence included only

Fig. 7. Bubble velocity.


1842 A. Boemer et al.
four frames. In this case, the relative velocities are in averaged for each row of cells. It can be observed that
between the range predicted by the models, but the t is always significantly lower than 1, which means
relative velocity is more or less constant. not the complete amount of additional air rises as
bubbles through the bed. Further, since the visible
3.7. »isible bubble flow bubble flow rate increases with the position in the bed,
The most simple theory of two-phase flow as- the bubble growth cannot be caused by coalescence
sumes minimum fluidization velocity in the emulsion alone.
and all additional gas to stream through the bed as The formula of Johnsson et al. (1989) agrees with
visible bubbles (Toomey and Johnstone, 1952). Ac- the simulation below 0.35 m, while above this height
cording to this theory, the dimensionless visible the calculated data agree better with the proposals of
bubble flow rate Hilligardt and Werther (1986) and Lord et al. (1982).
+ A )u Probably due to the unreliable analysis by hand, there
t" "6" "6" (10) is a strong scattering of the experimental results. Fur-
A ) (º !º )
"%$ o mf ther, the experiment shows higher values of t, espe-
is equal to 1. Since this value can be measured by cially below the bed surface. Nevertheless, they are
simply evaluating the size and velocity of all bubbles, still smaller than the results of Peters and Grace
there exist many investigations in literature. They are (1988). The two data by Cranfield and Geldart (1974)
compared with our experimental and numerical re- agree very well with both our simulation and experi-
sults in Fig. 10. Here, the numerical data are time ment.

Fig. 8. Example of a bubble coalescence process from simulation.

Fig. 9. Relative rising velocity º !º during a coalescence processes.


"6",2 "6",1
Eulerian simulation in fluidized bed 1843

Fig. 10. Visible bubble flow rate (simulation averaged over 3 s).

Fig. 11. Motion of gas and particles relative to a bubble (result of simulation. Vector scale: max. 1.5 m/s;
rising velocity of this bubble: 1 m/s).

Fig. 12. Gas bypass through bubbles.

3.8 Bypass through bubbles or flow through the bubbles as a bypass. In Fig. 11 the
The part of the additional amount of air situation at a typical bubble is shown. The gas velo-
(º !º ) which does not form bubbles must either city in the bubble is much faster than the bubble rise
o mf
lead to gas velocities higher than º in the emulsion, velocity. A considerable amount of gas thus takes its
mf
1844 A. Boemer et al.
way right through the bubble. Fig. 12 shows the K parameters in equation (13), dimensionless
i
relative bypass velocities m mass, kg
p pressure, Pa
u !u
º " g "6" (11) r, R radius, m
": º Re Reynolds number, dimensionless
mf
from literature compared with the results at a certain t time, s
instant from our simulation. The authors presented tr trace (sum of main-diagonal elements), di-
average values for the whole bed, whereas for the mensionless
computational results averages for each row of cells in u upward velocity component, m/s
the bed are depicted. In eq. (11) a bypass velocity is º superficial upward velocity, m/s
defined only in a row of cells where a bubble is present v velocity vector, m/s
at that instant. Therefore, the results of the simulation v@ fluctuating velocity vector, m/s
» ratio of terminal velocity of a group of par-
are not depicted by a continuous line in Fig. 12. r
According to the simple two-phase theory, ticles, to that of an isolated particle, dimen-
º should be zero. However, in all the bubbles occur- sionless
":
ring in this example, a bypass can be observed. The
bypass velocity is 2—3 times the minimum fluidization Greek letters
a wake angle, deg
velocity. This range agrees with the models found in W
literature, see Fig. 12. The bypass effect is additionally b interphase drag function, kg/m3/s
c granular temperature dissipation, Pa/s
visible by the compaction of the gas stream functions #
in the bubbles, as shown in the left-hand side of this e volume fraction, dimensionless
figure. j bulk viscosity, Pa s
' granular temperature exchange, Pa/s
#
4. CONCLUSIONS
t visible bubble flow rate, dimensionless
o density, kg/m3
A comparison of bubble features and behavior in
k shear viscosity, Pa s
a two-dimensional fluidized bed between Eulerian
# granular temperature, m2/s2
simulation, own experiments, and empirical models s
s viscous stress tensor, dimensionless
found in literature was performed and, in general,
a good agreement was observed. However, the simu-
Subscripts
lation gives additional information of local flow con-
bed bed
ditions which cannot be measured with acceptable
bub bubble
expenditure. The Eulerian simulation method has
by bypass
thus been proven to be a useful tool to predict the
g gas phase
fluiddynamics of bubbling fluidized beds.
i i"g, s (gas or solid phase)
max maximum
Acknowledgements
mf minimum fluidization condition
The extension of the FLUENT code for the Eulerian
o overall
multiphase option is being performed in co-operation be-
tween the Lehrstuhl für Wärmeübertragung und Klimatech-
rel relative
nik der RWTH-Aachen, Germany, and Fluent Europe Ltd., s solid phase
Sheffield, UK. It is supported by the European Commission
under contract JOU2-CT94-0452. The contribution of the Superscripts
students M. Heuwes, L. Lorang, J. Schäfer, and A. Schmidt is * parameter multiplied by the volume frac-
gratefully acknowledged. tion: X*"X ) e averaged
i i i
¹ transposed
NOTATION
2D two dimensional, dimensionless REFERENCES
3D three dimensional, dimensionless Allahwala, S. A. (1975) Dissertation at Monash Uni-
A area, m2 versity, Department of Chemical Engineering, Aus-
C drag coefficient, dimensionless tralia.
D
d diameter, m Bar-Cohen, A., Glicksman, L. R. and Hughes, R.
D strain rate tensor, 1/s (1997) Fluid dynamic modelling of fluidized bed
e coefficient of restitution (chosen: 0.61) combustors. Proceedings of the 5th Int. Conf. on
Fr froude number, dimensionless Fluidized Bed Combustion. Washington DC, USA,
g gravitational constant, 9.81 m/s2 pp. 458—470.
Boemer, A. (1996) Euler/Euler Simulation der Fluid-
g radial distribution function, dimensionless
0 dynamik blasenbildender Wirbelschichten. Disser-
I identity tensor, dimensionless tation, RWTH-Aachen.
I second invariant of the deviator of the solids
2D Boemer, A., Qi, H. and Renz, U. (1995) Eulerian
strain rate tensor, s~2 computation of fluidized bed hydrodynamics —
k granular temperature diffusion function, a comparison of physical models. Proc. 13th Int.
#
Pa s Conf. on FBC. Orlando, USA, pp. 775—786.
Eulerian simulation in fluidized bed 1845
Boemer, A., Qi, H. and Renz, U. (1997) Eulerian bubble in a fluidized bed. Chem. Engng Sci. 28,
simulation of bubble formation at a jet in a two- 2231—2243.
dimensional fluidized bed. Int. J. Multiphase Flow Lord, W. K., McAndrews, G., Sakagami, M., Valen-
23 (5), 927—944. zuela, J. A. and Glicksman, L. R. (1982) measure-
Buyevich, Y. A., Yates, J. G., Cheesman, D. J. and Wu, ments of bubble properties in fluidized beds. Dept.
K.-T. (1995) A model for the distribution of voidage Mech. Engng Massachusetts Institute of Techno-
around bubbles in a fluidized bed. Chem. Engng Sci. logy, Cambridge, USA.
50 (19), 3155—3162. Lun, C. K. K., Savage, F. B., Jeffrey, D. J. and Chepur-
Clift, R. and Grace, J. R. (1985) Bubble interaction niy, N. (1984) Kinetic theories for granular flow:
and coalescence. In Fluidization, eds J. F. Davidson, inelastic particles in couette flow and slightly inelas-
R. Clift, and D. Harrison. Academic press, ¸ondon tic particles in a general flowfield. J. Fluid Mech.
pp. 90—132. 140, 223—256.
Collins, E. (1965) An extension of Davidson’s theory of Miwa, K., Mori, S., Kato, T. and Muchi, I. (1972)
bubbles in fluidised beds. Chem. Engng Sci. 20 (1965), Behaviour of bubbles in a gaseous fluidized bed.
747—755. Int. Chem. Engng 12, 187—194.
Collins, R. (1989) A model for the effects of the void- Peters und G. (1988). Without a detailed source
age distribution around a fluidization bubble. quoted. Gas ¹hroughflow, Mass ¹ransfer and bed
Chem. Engng Sci. 44, 1481—1487. Expansion in the IEA-Model. In: J. E. Johnsson ed.,
Cranfield, R. R. and Geldart, D. (1974) Large particle 7th IEA-FBC Meeting, San Francisco, USA.
fluidization. Chem. Engng Sci. 29, 935—947. Pyle, D. L. and Harrison, D. (1967) The rising velocity
Davidson, J. F. and Harrison, D. (1963) Fluidised of bubbles in two-dimensional fluidised beds. Chem.
Particles. Cambridge University Press, London. Engng Sci. 22, 532—535.
Davidson, J. F. and Harrison, D. (1966) The behav- Rowe, P. N. and Partridge, B. A. (1965) An X-ray
iour of a continuously bubbling fluidised bed. study of bubbles in fluidised beds. ¹rans. Instn.
Chem. Engng Sci. 21, 731. Chem. Engrs 43.
Ding, J. and Gidaspow, D. (1990) A bubbling fluidiz- Sokolovski, V. V. (1965) Statics of Granular Media.
ation model using kinetic theory of granular flow. Pergamon Press.
A.I.Ch.E. J. 36 (4), 523—538. Syamlal, M., Rogers, W. and O’Brien, T. J. (1993)
Ellenberger, J. (1995) Analogies in multiphase reactor MFIX documentation, theory guide. ¹echnical
hydrodynamics. Dissertation at the University of Note DOE/METC-94/1004. Morgantown, West Vir-
Amsterdam, The Netherlands. ginia, USA.
Geldart, D. (1973) Types of gas fluidization. Powder Toomey, R. D. and Johnstone, H. F. (1952) Gaseous
¹echnol. 7, 285—292. fluidization of solid particles. Chem. Engng Progr.
Gidaspow, D. (1994) Multiphase Flow and Fluidiz- 48, 220—226.
ation: Continuum and Kinetic ¹heory Descriptions. Wein, J. W. (1992) Das Expansionsverhalten von
Academic Press, San Diego, USA. Gas/Feststoffwirbelschichten bei höheren Gasge-
Gidaspow, D., Bezburuah, R. and Ding, J. (1992) schwindigkeiten. Dissertation, TU Hamburg-Har-
Hydrodynamics of circulating fluidized beds, kin- burg.
etic theory approach, fluidization VII. Proceedings Yates, J. G., Cheesman, D. J. and Sergeev, Y. A.
of the 7th Engineering Foundation Conference on Experimental observations of voidage distribution
Fluidization, pp. 75—82. around bubbles in a fluidized bed. Chem. Engng Sci.
Hilligardt, K. (1986) Zur Strömungsmechanik von 49 (12), 1885—1895.
Grobkornwirbelschichten. Dissertation, TU Ham-
burg-Harburg. APPENDIX A
Hilligardt, K. and Werther, J. Local bubble gas hold-
up and expansion of gas/solid fluidised beds. Ger. Physical models mentioned in Section 1:
Chem. Engng 9 (1986), 215—221. Drag function of Syamlal et al. (1993):
Jenkins, J. T. and Savage, S. B. (1983) A theory for the
eo
rapid flow of identical, smooth, nearly elastic b*"3C ) s g ) e Dv !v D (A1)
4 D »2d g g s
spherical particles. J. Fluid Mech. 130 187. r s
Johnsson, F., Anderson, S. and Leckner, B. (1989) Bed
expansion. 7th IEA-FBC Meeting. San Francisco, with
USA.
» "1 (K !0.06 Re
Kuipers, J. A. M. (1990) A two-fluid micro balance r 2 1
model for fluidized beds. Dissertation at the Uni-
#J(0.06 Re)2#0.12 Re(2K !K )#K2) (A2)
versity of Twente, The Netherlands. 2 1 1
Lim, K. S., Gururajan, V. S. and Agarwal, P. K. (1993)
K "e4.14 (A3)
Mixing of homogeneous solids in bubbling 1 g
fluidized beds: theoretical modelling and experi-
G
0.8e1.28 for e *0.15
mental investigation using digital image analysis. K " g s . (A4)
2 e2.65 for e (0.15
Chem. Engng Sci. 48, 2251—2265. g s
Litka, T. and Glicksman, L. R. (1985) The influence of

A S B
particle mechanical properties on bubble character- » 2
C " 0.63#4.8 ) r (A5)
istics and solid mixing in fluidized beds. Powder D Re
¹echnol. 42, 231—239.
Littman, H. and Homolka, G. A. J. (1973) The d o ) Dv !v D
Re" s g g s (A6)
pressure field around a two-dimensional gas k
g
1846 A. Boemer et al.
z Solid pressure of Lun et al. (1984): z Granular temperature diffusion function of Gidaspow et
al. (1992):
p*"2o g e2# ) (1#e)#e o # . (A7)
s s 0s s s s s

S
#
z Solid shear viscosity of Gidaspow et al. (1992) in case of k*"2e2o d g (1#e) s
# s s s 0 n
0(e (0.59:
s
75 o d Jn#
S
# s s s (1#6 (1#e) g e )2 .
k*"4 e2o d g ) (1#e) ) s # (A11)
s 5 s s s 0 n 192 (1#e) g 5 0s
0

5Jn z Granular temperature dissipation of Jenkins and Savage


2) o d J# (1983):
96 s s s
# ) [1#4 g e ) (1#e)]2 (A8)
(1#e) ) g 5 0s

AS B
0 4 #
c*"3(1!e2)e2o g # s !+ ) v . (A12)
and the model derived from the tensor given by Sokolovski # s s 0 s d n s
s
(1965) in case of 0.59(e (0.6:
s z Granular temperature exchange function of Ding and
p* ) sin / Gidaspow (1990):
k*" s . (A9)
s 2JI
2D '*"!3b*# . (A13)
# s
z Solid bulk viscosity of Lun et al. (1984): z Radial distribution function of Ding and Gidaspow (1990):

C A B D
e 1@3 ~1
S
# 3
s
j*"4 e2o d g ) (1#e) ) s . (A10) g " ) 1!
0 5
(A14)
s 3 s s s 0 n e
s,.!9

You might also like