Available online at www.sciencedirect.

com

Catalysis Today 130 (2008) 519–526

www.elsevier.com/locate/cattod

Evaluation of the hydrodynamics of high-pressure ebullated

beds based on dimensional similitude
J.L. Sánchez a, R.S. Ruiz a,*, F. Alonso b,c, J. Ancheyta b
a
Departamento de Ingenierı́a de Procesos e Hidráulica, Universidad Autónoma Metropolitana-Iztapalapa,
Av. San Rafael Atlixco No. 186, Col. Vicentina, México 09340, D.F., Mexico
b
Instituto Mexicano del Petróleo, Eje Central Lázaro Cárdenas 152, México, D.F., 07730, Mexico
c
Instituto Tecnológico de Ciudad Madero, Juventino Rosas y Jesús Urueta, Col. Los Mangos, Cd. Madero, Tam., 89440, Mexico

Available online 11 December 2007

Abstract
The scale-up method for three-phase fluidized beds hydrodynamics proposed by Safoniuk et al. [M. Safoniuk, J.R. Grace, L. Hackman, C.A.
McKnight, Chem. Eng. Sci. 54 (1999) 4961], based upon the principles of dynamic similitude, was tested matching several systems operated at
high pressure with model systems operated at atmospheric pressure. Experiments were carried out to test this technique by comparing global phase
holdups, and flow regime in systems where the five dimensionless groups proposed by the method were matched, but which were operated at
significantly different pressures. The differences observed between the hydrodynamic parameters of the high pressure and model systems suggest
that the five dimensionless groups employed are not enough to fully characterize the global bed behavior. Similar findings have been reported by
Macchi et al. [A. Macchi, H. Bi, J.R. Grace, C.A. McKnight, L. Hackman, Chem. Eng. Sci. 54 (2001)] based on differences observed in the local
dynamic bed behavior between systems operated at atmospheric pressure.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Ebullated bed; Hydrodynamics; Scaling; Similitude; High pressure

1. Introduction hydrodynamic properties is essential to eliminate significant

scale-up effects.
Ebullated beds are three-phase gas–liquid–solid fluidized Some difficulties exist in scaling-up the results from small
beds which differ from slurry bubble columns in that their laboratory equipment to the extreme operating conditions of
particles are sufficiently large so as to behave as a distinct third large industrial units. For two-phase systems, much work has
phase. Ebullated bed reactors operated under high pressures and been carried out to solve scaling problems [4–6] but
high temperatures are applied industrially in the upgrading of comparatively little has been published regarding the applica-
heavy oil and high molecular weight feedstocks. Knowledge of tion of scaling laws to three-phase systems. With the aid of the
the hydrodynamic behavior at high pressure and temperature is Buckingham Pi theorem, Safoniuk et al. [7] proposed a method
critical in predicting the bed performance as a chemical reactor. of scaling hydrodynamics in three-phase fluidized beds based
The operating pressure has been found to have a significant on achieving geometric and dynamic similitude which resulted
effect on the hydrodynamics of these three-phase systems [1– in basically five dimensionless groups:
3]. Higher pressures yield smaller bubbles and more uniform
bubble-size distribution, and thus higher gas velocities are g Dr m4l
M¼ ; Morton number; (1)
required for the transition between the dispersed and coalesced r2l s 3
bubble flow regimes. Since the hydrodynamics control the
degree of fluids–solids contact, thorough understanding of the g Dr dp2
Eo ¼ ; Eötvös number; (2)
s

rl d p U l
* Corresponding author. Tel.: +52 5804 4648; fax: +52 5804 4900. Rel ¼ ; liquid Reynolds number; (3)
E-mail address: rmr@xanum.uam.mx (R.S. Ruiz). ml

0920-5861/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2007.10.002
520 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526

the coalescing behavior of the water–glycerol mixture and that

Nomenclature of a pure monocomponent liquid as silicone oil. It should be
noted that both of these works have evaluated the scaling
A column cross-sectional area (m2)
approach proposed by Safoniuk et al. based on systems
dp particle diameter (m)
operated at atmospheric pressure.
Dc column diameter (m)
In the present work, the dimensional similitude approach
Eo Eötvös number (¼ g Dr dp2 =s)
proposed by Sofaniuk et al. [7] is tested in simulating the
g gravitational acceleration (9.8 m/s2)
hydrodynamics of a three-phase fluidized bed operated at high
H expanded bed height (m)
pressure using a model system at atmospheric pressure and for
L length (m)
both of which the five dimensionless groups (Eqs. (1)–(5)) are
M Morton number (¼ g Dr m4l =ðr2l s 3 Þ)
closely matched.
P pressure (Pa)
DP pressure drop (Pa)
Reg Rel  bU = rldpUg/ml 2. Experimental
Rel liquid Reynolds number (=rldpUl/ml)
T temperature (8C) In order to evaluate the fundamental approach to simulate
U superficial velocity (m/s) bed and bubble hydrodynamics in high-pressure, high-
Ws particle inventory (kg) temperature industrial units using small-scale cold-flow
Dz difference in axial height (m) models, it is important that accurate industrial data with which
to compare the laboratory results are used. In the present work,
Greek letters adequate industrial data were not available for this purpose,
bd ratio of densities (=rp/rl) however, hydrodynamic data for a high-pressure laboratory unit
bU ratio of superficial velocities (=Ug/Ul) reported by Luo et al. [9] was considered as suitable and hence
Dr bouyancy term (=rl  rg) (kg/m3) was chosen as the prototype or reference system. This system
e holdup consisted of a stainless-steel column with an inner diameter of
m viscosity (Pa s) 0.051 m and a 0.8 m total height. Paratherm NF heat-transfer
r density (kg/m3) fluid and nitrogen were used as the liquid and gas phases,
s surface tension (N/m) respectively. Hydrodynamic data are reported in this reference
for 3 mm spherical glass beads (rs = 2520 kg/m3) used as the
Subscripts solid phase, for a constant operating bed temperature of 34 8C,
b bubble and pressures up to 15.6 MPa. The physical properties of
g gas the gas and liquid phases at various pressures, required for
l liquid the calculation of the dimensionless groups were also reported
p particle [9].
The experimental unit employed for simulating the
prototype system (referred to as the model system) is shown
rp schematically in Fig. 1. A cylindrical glass column of 0.1 m
bd ¼ ; density ratio; and (4) internal diameter and 1.4 m height was used. The column is
rl
made of three sections, namely, the gas–liquid distributor
Ug
bU ¼ ; superficial velocity ratio: (5)
Ul
The authors compared gas holdup and bed expansion in two
different columns operating with significantly different liquid
and solid properties with close matching of the five
dimensionless groups. They reported good agreement between
both units for most of the data. Macchi et al. [8] tested the
scaling approach proposed by Safoniuk et al. [7] using two
matched systems, where aqueous glycerol solution and silicone
oil were utilized respectively as the liquid phase of each unit.
They compared statistically hydrodynamic parameters such as
bed expansion, gas holdups, dimensionless transition velocities
between dispersed and coalesced bubble flow regimes, and the
minimum liquid fluidization velocity Reynolds number, and
reported that although the differences between the hydro-
dynamic parameters for the two systems studied were found to
be statistically significant, they were also generally small and
less than 12%. They attributed the mismatch to differences in Fig. 1. Schematic diagram of experimental model system.
 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526 521

0.7 and 2.13

and 0–0.451
6.94  104
(0.08 MPa;
section, the test section, and the gas–liquid disengagement

0–5.655
section. A perforated plate was used as the distributor. Gasoil

33.333
19 8C)
Model
Matched systems VI

2.38

2.79
was used as the liquid phase, which consisted of a mixture of
heavy and light straight-run gasoils that contained 95% in

0.7 and 2.13

and 0–2.451
(15.60 MPa;

6.75  104
volume of the former. Nitrogen was used for the gas phase and

Prototype
3 mm spherical glass beads for the solid phase. Table 1 presents

0–5.655
34 8C)

16.93
the operating conditions and dimensionless groups for both

2.65

2.83
prototype and model systems.
It should be pointed out that while the Buckingham Pi

0.77 and 2.34

and 0–2.451
4.40  104
(0.08 MPa;
theorem also requires equality of Dc/dp for the two units, the

0–7.738
exact match of this dimensionless group, however, can be

33.333
22 8C)
Model

2.44

2.79
relaxed providing that it is sufficiently large that wall effects

Matched systems V
are negligible. For the prototype and model systems Dc/dp

0.77 and 2.34

were 17 and 33, respectively, thus minimizing wall effects

and 0–2.451
(10.10 MPa;

4.53  104
Prototype
[7].

0–7.738
34 8C)
In the model system the height of the expanded bed was

16.93
2.77

2.87
determined from the axial pressure profiles as the position
where there is an abrupt change of the slope of the axial

0.879 and 2.67

pressure profile. For this purpose, pressure tappings were fitted

and 0–2.549
2.79  104
(0.08 MPa;
along the column at every 7.5 cm interval, with the first one

0–6.548

33.333
25 8C)
Model
located 10 cm above the distributor. For a bed with low solids

2.51

2.80
Matched systems IV
entrainment rates and where solids holdups are assumed to be
independent of height throughout the entire expanded bed

0.879 and 2.67

region, the overall bed porosity, e, is commonly calculated from

and 0–2.549
2.55  104
(5.62 MPa;
Prototype
a mass balance on the particles in terms of the expanded bed

0–6.548
34 8C)

16.93
height, H, the cross-sectional area of the bed, A, and total solids

2.84

2.89
inventory, Ws, i.e.

1.32  104
Ws
(0.08 MPa;
e ¼ 1  es ¼ 1  : (6)

0–6.845
rs AH

33.333
30 8C)
Model

0.994
Matched systems III

2.63

2.81
Once the solids holdup, es, is known the overall gas and
liquid holdups can be determined from the following relation- 1.37  104
(2.86 MPa;

ships [10]:
Prototype

0–6.845
34 8C)

0.994

16.93
ðDP=gHÞ  es ðrs  rl Þ  rl
2.79

2.90
eg ¼ ; (7)
rg  rl
9.79  105
(0.08 MPa;

es þ el þ eg ¼ 1; (8)
0–8.250

33.333
32 8C)
Model
Operating conditions and dimensionless groups for matched systems

2.68
1.06
2.81

where DP is the total pressure in the fluidized bed and which is

Matched systems II

obtained from the dynamic pressure drop as measured by

differential manometry corrected by the hydrostatic head of
1.00  104
(1.82 MPa;
Prototype

the liquid [10,13]. For low and moderately gas and liquid
0–8.250
34 8C)

16.93

velocities, holdup predictions based on this pressure gradient

2.77
1.06
2.91

method have been reported to work satisfactory when compared

to other measuring methods [13].
6.28  105
(0.08 MPa;

0–9.524

33.333
35 8C)
Model

3. Results and discussion

2.76
1.15
2.82
Matched systems I

The matching of the dimensionless groups proposed by

6.52  105
(0.79 MPa;

Safoniuk et al. [7] for the experimental systems reported by Luo

Prototype

0–9.524

et al. [9] operated at high pressure and our experimental model

34 8C)

16.93
2.68
1.15
2.91

systems are shown in Table 1. As seen in the table, although the

match of the dimensionless groups is not perfect, it can be
Dimensionless

considered to be acceptable and as good as those reported in the

literature [7,8]. A comparison between some hydrodynamic
Table 1

parameters for both the high-pressure prototype and model

group

Dc/dp
Rel
Eo

bU
bd

systems is shown and discussed next.

M
522 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526

Fig. 3. Bed porosity versus Reg for prototype and model systems with the
former operated at 5.62 MPa and both systems at: (a) Rel = 0.88 and (b)
Rel = 2.67.

porosity of the prototype system and that bed contraction was

not evident from the data. Fig. 2 also shows the results obtained
with the corresponding matching model system. As can be seen,
a reasonable correspondence between the values of the
prototype and the model appears to have been obtained,
especially for the prototype operated at lower pressures. For all
the data in Fig. 2 the deviations between porosities were found
to be within the 2–14% range. It is also evident that the general
effect of gas velocity on the model bed porosity was
Fig. 2. Bed porosity versus Reg for prototype and model systems with the qualitatively similar to those reported for the prototype. The
former operated at three different pressures: (a) 0.79 MPa, (b) 2.86 MPa, and (c)
effect of liquid velocity on bed porosity for the model and
10.10 MPa.
prototype systems is shown in Fig. 3, for the prototype operated
at a pressure of 5.62 MPa. It is clear that both the prototype and
3.1. Bed porosity the model responded similarly to the increase in the liquid
Reynolds number although the bed porosities were compara-
Bed expansion characteristics or the bed porosity has been tively lower for the latter.
studied under high-pressure conditions [2,3,9] and it has been From the above results it should be pointed out that although
reported that, in general, bed porosity is more sensitive to the the observed similarities, the prototype system showed
liquid than to the gas velocity. It has also been found that, at a systematically higher bed porosities than those for the matching
given liquid velocity, upon introduction of the gas phase the bed model. This difference could be due to the foaming character of
porosity can either decrease or increase depending on the the liquid utilized in the model system, which during the
bubble flow pattern. The bed porosities for the prototype, for operation showed a tendency to trap small air bubbles that could
different operating conditions, and for the corresponding model have the effect of diminishing the effective viscosity of the
systems are shown in Figs. 2 and 3. Fig. 2 presents the variation continuous phase and hence the drag force on the particles.
of the bed porosity of the prototype with gas Reynolds number
for three different operating pressures, 0.79, 2.86, and 3.2. Gas and liquid holdups
10.10 MPa, and for an average liquid Reynolds number of
0.97. It is clear from this figure, for all three pressures, that an It has been reported in the literature that the gas and liquid
increase in the gas Reynolds number increases slightly the bed holdups in three-phase fluidized beds can be affected by the
 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526 523

Fig. 5. Gas and liquid holdups versus Reg for prototype and model systems with
the former operated at 5.62 MPa and both systems at: (a) Rel = 0.88 and (b)
Rel = 2.67.

matching model systems are also shown in Fig. 4. For all three
cases shown the gas holdup of the model increases with gas
velocity and does so with values higher than for the
corresponding prototype system. For the liquid holdups, the
model systems show a qualitatively similar behavior to those
observed for the prototype systems, however they were always
found to be smaller than the latter. The larger gas holdups
obtained for the model can be attributed to the noncoalescing
behavior of the oil fraction employed for such systems, and in
Fig. 4. Gas and liquid holdups versus Reg for prototype and model systems with which gas bubbles tend to rise as small and slow entities that
the former operated at three different pressures: (a) 0.79 MPa, (b) 2.86 MPa, produce relatively high gas holdups. On the other hand,
and (c) 10.10 MPa.
Paratherm NF heat transfer liquid was employed in the
prototype system and for which no foaming behavior has been
operating pressure through its effect on the bubble character- reported [2,9,11].
istics and on the flow regime [1,9]. An increase in pressure at The effect of liquid velocity on gas and liquid holdups for the
constants gas and liquid velocities produces a reduction of the model and prototype systems is shown in Fig. 5, for the
size and velocity of the gas bubbles which in turn increases the prototype operated at a pressure of 5.62 MPa. The gas holdups
gas holdup in the fluidized bed. for the prototypes are found to decrease slightly with the
The bed gas and liquid holdups for the prototype, for increase in liquid velocity. Similarly, the gas holdups for the
different operating conditions, and for the corresponding model models also decrease with liquid velocity, but does so at larger
systems are shown in Figs. 4 and 5. Fig. 4 presents the variation rates than in the prototypes, and this is particularly noticeable at
of the holdups for the prototype with gas Reynolds number for high gas velocities for which the gas holdups of the prototypes
three different operating pressures, 0.79, 2.86, and 10.10 MPa, exceed those of the models. Luo et al. [9] reported that the
and for an average liquid Reynolds number of 0.97. A pressure effect on gas holdup is more pronounced at high gas
continuous increase in gas holdup is observed with an increase velocities. As gas velocity is increased larger bubbles can be
in gas Reynolds number, which seems to be more pronounced at expected to occur and hence the effect of pressure on bubble
higher pressures. The liquid holdup, on the other hand, breakup phenomena becomes more evident. From comparison
decreases with an increase of gas flow rate and operating of the gas holdups for the prototype and the model systems it
pressure. The gas and liquid holdups for the corresponding seems that the response of the prototype at high pressure and
524 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526

fluid velocities is reproduced only partially by the model increased regime transitions occurs at higher bed porosities,
system. It has been reported that pressure affects the and hence at higher gas velocities. The gas drift flux for the
hydrodynamics through the variation on the physical properties model systems is shown in Fig. 6(b), and in which the symbols
of the gas and liquid phases, therefore, in the prototype both the utilized for the data are the same to those of the prototypes in
gas and the liquid are being affected. On the other hand, the Fig. 6(a) that are being matched. In order to match the dimen-
model system was designed based on the five dimensionless sionless groups between the model and the prototype systems
groups presented in Eqs. (1)–(5), none of which considers the the operating temperature of the model systems had to be
properties of the gas phase. For the prototype operated at high modified so as to produce the necessary changes in liquid
pressure and gas flow rates it therefore seems necessary to properties. As it is shown in Fig. 6(b) an increase in the
utilize the extra dimensionless group suggested by Safoniuk transition gas velocities in the prototypes is also qualitatively
et al. [7] given by the gas to liquid density ratio, rg/rl. predicted by the model systems but in the latter it is produced
by decreasing the operating temperature.
3.3. Flow regime The prevailing bubble flow regime depends on both the gas
and liquid velocities. Fig. 7 shows the gas drift flux for both
The transition between the dispersed and coalesced bubble prototype and model systems at two different liquid velocities
regimes in high-pressure ebullated beds has been analyzed by for a given prototype pressure of 5.62 MPa. For the range of gas
many researchers utilizing the gas drift flux concept [3,9] The porosities in Fig. 7, and for both liquid velocities the prototype
gas drift flux, jcd, for three-phase fluidized beds can be systems seem to operate in the dispersed bubble flow regime.
expressed by [12] On the other hand, the model systems seem to show different
regime transition conditions for the two liquid velocities
1  eg
jcd ¼ ðU g el  U l eg Þ (9) considered. As the liquid Reynolds number is increased the gas
el
holdup and hence the transition gas velocity decreases. It is then
which seems to increase with gas holdup at a much higher rate clear from this figure that the prototype systems are maintained
at the coalesce bubble flow regime as compared to that at the in the dispersed bubble flow regime for a broader range of gas
dispersed bubble flow regime. An increase of pressure has been and liquid velocities as compared to the matching model
reported to lead to an increase in the transition gas velocity from systems. The effect of pressure on the hydrodynamics of the
the dispersed to the coalesced bubble regimes. The gas drift flux prototypes have been modeled in the present work by matching
for the prototype operated at different pressures and the corre- of five dimensionless groups appear to be insufficient when
sponding matching model systems are presented in Fig. 6. As operating at high gas and liquid velocity conditions. Pressure
can be seen in Fig. 6(a), for the prototype system as pressure is affects hydrodynamics by affecting both liquid and gas

Fig. 6. Gas drift flux versus gas holdup for (a) the prototype systems operated at Fig. 7. Gas drift flux for both prototype and model systems with the former
different pressures and (b) the corresponding matching model systems. operated at 5.62 MPa and both systems at: (a) Rel = 0.88 and (b) Rel = 2.67.
 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526 525

Table 2 Table 3
RMSD and Fm between bed porosities for the matched systems Average RMSD and Fm between hydrodynamic parameters for the matched
systems
Matched system Rel RMSD Fm
Hydrodynamic parameter RMSD Fm
I 1.15 0.063 1.055
II 1.06 0.085 1.091 Bed porosity 0.108 1.115
III 0.99 0.093 1.098 Gas holdup 2.107 0.575
IV 0.88 0.101 1.108 Liquid holdup 0.237 1.284
2.67 0.129 1.147
V 0.77 0.126 1.140
2.34 0.119 1.133
VI 0.70 0.132 1.151 global hydrodynamics but it is clear that they seem to depend on
2.13 0.112 1.121 the coalescing nature of the liquids in the prototype and model
Mean 0.108 1.115 systems. The gasoil mixture employed in the models showed a
noncoalescing behavior and a tendency to foam as temperature
was decreased and the liquid velocity increased. This behavior
produced significantly larger gas holdups in the models as
properties and the aforementioned groups do not consider the compared to the prototypes even for relatively low gas
gas properties. As mentioned earlier, the use of a denser gas in Reynolds numbers. It is also apparent that the gas to liquid
the model system that would match the densities ratio, rg/rl, density ratio is a group that needs to be matched when modeling
could account to the effect of pressure on the gas phase. For the high-pressure systems, especially under relatively high gas and
operating conditions considered in Fig. 7, the density ratios for liquid flow rates for which the effect of pressure becomes more
the prototype and model systems were 7.02  102 and evident. Finally, it is evident that differences between the
9.79  104, respectively. Therefore, in order to match the coalescing behaviors of the liquids of the matched systems play
density ratio between both units gas with a density of 62.4 kg/ an important role that has to be accounted for in future work
m3 is required for the model system, and which is about two when selecting the liquid for the model system.
orders of magnitude larger than that of nitrogen at atmospheric
pressure and ambient temperature. As can be expected, 4. Conclusions
availability of a safe to use gas with such a density could be
a limitation to the exact matching of the density group. The scale-up method for three-phase fluidized beds
The differences in the hydrodynamic parameters between hydrodynamics proposed by Safoniuk et al. [7], based upon
the prototype and the model systems have been characterized the principles of dynamic similitude, was tested matching
by the root mean standard deviation (RMSD) and bias factor several systems operated at high pressure with model systems
(Fm) operated at atmospheric pressure.
rffiffiffi k  Bed porosities were qualitatively similar for prototype and
2
1X model systems and the quantitative differences were generally
RMSD ¼ ð prototype  ModelÞ= prototype ; less than 13%, however, such differences appeared to increase
k i¼1
X  as the pressure of the prototype system was increased. In
k
Fm ¼ exp Inð prototype=ModelÞ : (10) general the gas holdup of the model systems overestimated
i¼1
those for the prototype systems, behavior that seams to be
attributed to the foaming character of the oil fraction used in
Table 2 presents these quantities for the bed porosity and six model systems that showed a tendency to accumulate small
different prototype operating pressures (see Table 1). The gas bubbles even at low gas velocities. The drift fluxes
results show that the differences between prototype and model calculated for both prototype and model systems showed that,
tend to be smaller for the lower pressure systems and that the although most of them operated in the dispersed bubble
largest RMSD obtained corresponds to the matching of the regime, the high-pressure systems appear to have larger gas
prototype operated at the highest pressure considered transition velocities between the dispersed and coalesced flow
(15.6 MPa). It is also evident from this table that Fm is always regimes. It seems that the use of denser gases in the model
larger than unity and which means that the bed porosities for the systems could help extend their respective gas transition
prototype are also always underestimated by the model. velocities.
The average RMSD and Fm values obtained for the The differences observed between the hydrodynamic
hydrodynamic parameters for all the matched systems parameters of the high pressure and model systems suggest
considered are presented in Table 3. The data are clearly that the five dimensionless groups employed are not enough to
skewed (0.575 < Fm < 1.284) and the best RMSD corresponds fully characterize the global bed behavior and that, although
to the bed porosity, of the order of 0.11, which is followed by matching the additional group given by gas to liquid density
that for the liquid holdup, and then, with a significantly large ratio could probably help to extend the dispersed bubble flow
deviation, by the gas holdup. regime in the model system, there is still lack of information
From the present results it seems that the dimensional regarding the coalescing behavior of the liquids used in the
similitude approach provides a useful tool for estimation of matched systems.
526 J.L. Sánchez et al. / Catalysis Today 130 (2008) 519–526

Acknowledgement [6] T.M. Knowlton, S.B.R. Karri, A. Issangya, Powder Technol. 150 (2005)
72.
[7] M. Safoniuk, J.R. Grace, L. Hackman, C.A. McKnight, Chem. Eng. Sci.
The authors thank Instituto Mexicano del Petróleo for the 54 (1999) 4961.
financial support. [8] A. Macchi, H. Bi, J.R. Grace, C.A. McKnight, L. Hackman, Chem. Eng.
Sci. 54 (2001) 6039.
References [9] X. Luo, P. Jiang, L.-S. Fan, AIChE J. 43 (1997) 2432.
[10] N. Costa, A. De Lucas, P. Garcı́a, Ind. Eng. Chem. Process Des. Dev. 25
[1] P. Jiang, D. Arters, L.-S. Fan, Ind. Eng. Chem. Res. 31 (1992) 2322. (1986) 849.
[2] P. Jiang, X. Luo, T.-J. Lin, L.-S. Fan, Powder Technol. 90 (1997) 103. [11] X. Luo, J. Zhang, K. Tsuchiya, L.-S. Fan, Chem. Eng. Sci. 52 (1997)
[3] R.S. Ruiz, F. Alonso, J. Ancheyta, Catal. Today 109 (2005) 205. 3693.
[4] M. Horio, A. Nonaka, Y. Sawa, I. Muchi, AIChE J. 32 (1986) 1466. [12] R.C. Darton, D. Harrison, Chem. Eng. Sci. 30 (1975) 581.
[5] L.R. Glicksman, M.R. Hyre, P.A. Farrell, Int. J. Multiphase Flow (1994) [13] L.-S. Fan, Gas–Liquid–Solid Fluidization Engineering, Butterworths,
331. Boston, 1989.