Missouri University of Science and Technology

Scholars' Mine

Chemical and Biochemical Engineering Faculty Linda and Bipin Doshi Department of Chemical
Research & Creative Works and Biochemical Engineering

19 Jun 2013

Scale-Up of Bubble Column Reactors: A Review of Current State-

Of-The-Art
Ashfaq Shaikh

Muthanna H. Al-Dahhan
Missouri University of Science and Technology, [email protected]

Follow this and additional works at: https://scholarsmine.mst.edu/che_bioeng_facwork

Part of the Biochemical and Biomolecular Engineering Commons

Recommended Citation
A. Shaikh and M. H. Al-Dahhan, "Scale-Up of Bubble Column Reactors: A Review of Current State-Of-The-
Art," Industrial and Engineering Chemistry Research, vol. 52, no. 24, pp. 8091 - 8108, American Chemical
Society, Jun 2013.
The definitive version is available at https://doi.org/10.1021/ie302080m

This Article - Journal is brought to you for free and open access by Scholars' Mine. It has been accepted for
inclusion in Chemical and Biochemical Engineering Faculty Research & Creative Works by an authorized
administrator of Scholars' Mine. This work is protected by U. S. Copyright Law. Unauthorized use including
reproduction for redistribution requires the permission of the copyright holder. For more information, please
contact [email protected].
 Review

pubs.acs.org/IECR

Scale-up of Bubble Column Reactors: A Review of Current State-of-

the-Art
Ashfaq Shaikh*,† and Muthanna Al-Dahhan‡
†
Eastman Research Division, Eastman Chemical Company, Kingsport, Tennessee 37662, United States
‡
Missouri University of Science and Technology, Rolla, Missouri 65409, United States

ABSTRACT: In multiphase flow reaction systems, in general, an extrapolation of small diameter behavior to larger ones is
always an important and challenging task. The critical issue in such an extrapolation remains to be mixing and hydrodynamic
Downloaded via MISSOURI UNIV SCIENCE & TECHNOLOGY on April 10, 2023 at 16:19:15 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

characteristics. It needs reliable similarity criteria that would result in similar mixing and hydrodynamics and hence transport and
performance in two different scales. Numerous experimental and computational studies have been performed to investigate the
flow behavior of bubble column reactors for a proper design and scale-up. Experimental techniques vary from simple visual
observation to more advanced noninvasive diagnostic techniques. On computational front the progress has been made from
simple reactor models to fundamentally based Computational Fluid Dynamics (CFD). Such studies ultimately provide a
knowledge that help in understanding the hydrodynamic and mixing characteristics of these reactors and would aid in its scale-
up. Based on these studies, various methodologies have been proposed in literature for scale-up and/or to maintain their
hydrodynamic and mixing similarity. This article attempts to review the current state of reported dynamic similarity and scale-up
methods of bubble column reactors. It mostly covers the methods reported in open literature. The scale-up practices in industry
appear to be proprietary for obvious reasons.

1. INTRODUCTION relatively large diameter bubble columns, i.e., homogeneous and

In a bubble column reactor, a gas phase is bubbled through a heterogeneous (Shaikh and Al-Dahhan1).
column of liquid to promote a chemical or biochemical reaction These reactors are of considerable interest in chemical,
in the presence or absence of a catalyst suspended in the liquid petrochemical, and petroleum industries for various processes.
phase (Figure 1). A bubble column offers numerous advantages Examples of such processes are partial oxidation of ethylene to
acetaldehyde, wet-air oxidation (Deckwer2), liquid phase
methanol synthesis (LPMeOH), Fischer−Tropsch (FT) syn-
thesis (Wender3), and hydrogenation of maleic acid (MAC). In
biochemical industries, bubble columns are used for cultivation
of bacteria, cultivation of mold fungi, production of single cell
protein, animal cell culture (Lehmann et al.4), and treatment of
sewage (Diesterweg5). In metallurgical industries, it can be used
for leaching of ores. The most popular present day application
of bubble columns is for energy conversion process where
‘stranded gas’ is being converted to liquids called Gas-to-Liquid
(GTL) Fischer−Tropsch (FT) synthesis, when syngas is
produced from natural gas. Syngas can also be produced
from coal or biomass in which case this conversion is called
Coal-to-Liquid (CTL) and Biomass-to-Liquid (BTL), respec-
tively. The popularity of such a conversion is a response to the
postulated future energy crisis.
Figure 1. Schematic diagram of a bubble column. Different reactor configurations were utilized for FT
processes in earlier days while the design of bubble columns
such as good heat and mass transfer, no moving parts, ease of has been considered for low temperature FT processes since
operation, and low operating and maintenance costs. The main Kolbel’s6 pioneering work in the 1950s. Krishna and Sie7
disadvantage of bubble column reactors is backmixing, which showed that a bubble column reactor may achieve a
adversely affects product conversion. In these reactors, productivity of 2500 times higher than that of the classical
momentum is transferred from the faster, upward moving gas FT reactors used in the 1940s such as fixed bed reactors,
phase to the slower liquid/slurry phase. The operating liquid
superficial velocity (in the range of 0 to 2 cm/s) is an order of Received: August 3, 2012
magnitude smaller than the superficial gas velocity (1 to 50 cm/ Revised: March 22, 2013
s). Hence, the hydrodynamics of such reactors are controlled Accepted: March 28, 2013
mainly by the gas flow. There exists mainly two flow regimes in Published: March 28, 2013

© 2013 American Chemical Society 8091 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

multibed reactors. However, there are considerable reactor literature data, it was argued that the gas holdup is virtually
design and scale-up problems associated with such energy independent of the column dimensions and the sparger layout
conversion processes involving bubble columns. In general, for (for low as well as high pressures) provided that the following
any multiphase reaction system, an extrapolation of small scale three criteria are fulfilled:
behavior to larger ones is always an important and challenging 1) The column diameter has to be larger than 15 cm.
task. The critical issue in such an extrapolation remains to be 2) The column height to diameter ratio has to be in excess of
mixing and hydrodynamic characteristics. It needs reliable 5.
similarity criteria that would result in similar mixing and 3) The hole diameter of the sparger has to be larger than 1−
hydrodynamic which would provide similar transport and 2 mm.
performance (conversion and selectivity). Therefore, numerous Wilkinson et al.8 concluded that a scale-up procedure in
experimental and computational studies have been performed which the gas holdup, volumetric mass transfer coefficient, and
to investigate the flow behavior of bubble column reactors for interfacial area for a large scale industrial bubble columns are
proper design and scale-up. estimated on the basis of experimental data obtained in a pilot
In order to achieve economically high space-time yields in plant bubble column with small dimensions (aspect ratio < 5, D
the GTL-FT reactors, high slurry concentration (typically 30− < 15 cm) or with porous plate type of spargers will generally
50% vol.) needs to be employed. To suspend such a high lead to a considerable overestimation of these parameters.
amount of catalyst, high energy input is needed that can be Wilkinson et al.8 compared the predictions of selected
provided by high superficial gas velocities which make the reported correlations (Hikita et al.;9 Hammer et al.;10 Idogawa
reactor operate in a churn-turbulent flow regime. The process et al.;11 Reilly et al.;12 Idogawa et al.13) with their own data and
operates under high-pressure conditions (typically 10−80 bar). literature data. The reported correlations were developed either
To remove high exothermic heat of reaction, an efficient means at high operating pressure or for gases with different densities.
of heat removal is needed, and also it needs to be operated in a Wilkinson et al.8 found that none of these correlations could
churn-turbulent flow regime that can be generally achieved at predict the gas holdup within statistical confidence. Reilly et
high superficial gas velocities. Finally, the large gas throughputs al.12 correlation showed the maximum error in this comparison.
necessitate the use of large diameter reactors (typically 5−8 m), Wilkinson et al.8 argued that the behavior of gas holdup varies
and to obtain high conversion levels, large reactor heights, according to the operating flow regime; hence, they developed
typically 20−30 m tall, are required. Successful commercializa- a gas holdup correlation that incorporates flow regime
tion of the bubble column reactors is crucially dependent on transition. In addition, the effect of gas density and liquid
the proper understanding of its hydrodynamics and the scale-up phase properties on regime transition and gas holdup was
principles. studied.
Thus, various methodologies and approaches have been used The experimental data were incorporated with a gas holdup
and reported in the open literature for scale-up of bubble model developed by them, which is similar but slightly different
columns. This overview examines current state-of-the-art of than the one proposed by Krishna et al.14 For superficial gas
scale-up of bubble column reactors reported in the literature. velocities less than transition velocity, the gas holdup increases
Most of these studies have focused on developing either scaling proportionally to the superficial gas velocity, hence the
rules or an approach that would provide a platform for design following relationship was written for homogeneous flow
and scale-up of bubble columns. for UG < Utrans:

2. OVERVIEW OF REPORTED SCALE-UP εG = AUG (1)

METHODOLOGIES AND PROCEDURES which can be then written in the following form
The following is a summary and review of the current state of
εG = UG/Us , b (2)
the scale-up of bubble column reactors reported in the
literature. We do not discuss development of phenomeno- where Us,b is rise velocity of bubbles.
logical/CFD models, nor do we provide details of experimental On the basis of dimensional analysis, Wilkinson et al.8
techniques used in those studies. proposed the following equation for bubble rise velocity
2.1. Wilkinson et al.8 Wilkinson et al.8 proposed three
criteria discussed below based on matching overall gas holdup Us , bμL ⎡ σ 3ρ ⎤n1⎡ ρ ⎤n2
for scale-up of high pressure bubble column reactors. A = C ⎢ 4L ⎥ ⎢ L ⎥
correlation for overall gas holdup was proposed based on their σ ⎣⎢ gμL ⎦⎥ ⎣⎢ ρG ⎥⎦ (3)
own as well as literature data that accounts for the effect of gas
density and incorporates the flow regime transition. for UG > Utrans:
Wilkinson et al.8 performed experiments for scale-up UG − Utrans
purposes in two different column diameters (15 and 23 cm) εG = Utrans/Us , b +
Ul , b (4)
at operating pressures varying between 0.1 to 2 MPa. Three
different liquids were used: n-heptane (density = 0.684 g·cc−1, In eq 4, Ul,b is the rise velocity of large bubbles, which should
viscosity = 0.00041 Pa·s, surface tension = 20 dyn·cm−1), be greater than Us,b.
monoethylene glycol (density = 1.113 g·cc−1, viscosity = 0.021 To hold this condition, the following dimensionless equation
Pa·s, surface tension = 4 8 dyn·cm−1), and water (density = was chosen:
0.998 g·cc−1, viscosity = 0.001 Pa·s, surface tension = 72
dyn·cm−1). The hydrodynamics were quantified based on μL Ul , b μL Us , b ⎡ μ (UG − Utrans) ⎤n3⎡ σ 3ρ ⎤ ⎡ ρ ⎤
n4 n5

overall gas holdup. In addition, the authors have performed = + C⎢ L

⎥ ⎢ 4⎥ ⎢ ⎥
L L

literature survey regarding the effect of liquid height and σ σ ⎣ σ ⎦ ⎢⎣ gμL ⎥⎦ ⎢⎣ ρG ⎥⎦

sparger design on gas holdup. Based on their own results and (5)

8092 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

On the basis of rational explanation of behavior of transition The liquid axial velocity profile and eddy diffusivities profile
velocity with the change in liquid properties and gas density, an were not available in an industrial scale unit at reaction
empirical equation for transition velocity has been proposed conditions. The experimental measurements of these fluid
dynamic parameters were available only in a laboratory scale
Utrans = F(σ , ρG , ρL , μL ) (6) bubble column (diameter = 14, 19, 44 cm) at ambient
Based on their own experimental data as well as published conditions in an air−water system. Degaleesan17 developed
data from the literature, optimal values of parameters in eq 3, 5, scaling rules to extrapolate available laboratory data to an
and 6 were determined with the aid of nonlinear regression industrial unit, to predict the needed fluid dynamic parameters,
analysis. The regression analysis leads to the following The proposed approach was facilitated by the following
correlations to predict transition velocity and holdup. correlations:
Overall Gas Holdup. To predict overall gas holdup in an
Us , bμL ⎡ σ 3ρ ⎤−0.273⎡ ρ ⎤0.03 atmospheric air−water system, Degaleesan17 developed a
= 2.23⎢ L⎥
⎢ L⎥ correlation using literature as well as her own data in different
σ ⎢⎣ gμL4 ⎥⎦ ⎣⎢ ρG ⎦⎥ (7) column diameters. The proposed correlation is as follows:
εG̅ = 0.07UG0.474 − 0.000626D (cgs units) (10)
⎡ μ (UG − Utrans) ⎤0.757 ⎡ σ 3ρ ⎤
−0.077
μL Ul , b μL Us , b
= + 2.4⎢ L
⎥ ⎢ L⎥
The developed correlation showed a good comparison with
σ σ ⎣ σ ⎦ ⎢⎣ gμL4 ⎥⎦
Reilly et al.12 and Hammer et al.10 correlations at low superficial
⎡ ρ ⎤0.077 gas velocities. While at high superficial gas velocities, it captured
⎢ L⎥ the effect of diameter similar to the data of Nottenkamper et
⎢⎣ ρG ⎥⎦ (8) al.18 However, based on eq 10, the effect of scale on overall gas
holdup appears to be weak.
UGtrans Liquid Recirculation. Experimental evidence shows that
= εGtrans = 0.5 exp(− 193ρG−0.61μL0.5σL 0.11) there exists a single circulation cell in a time-averaged sense
Us , b (9)
with liquid upflow in the center and downflow near the wall. In
Equations 7, 8, and 9 give the values of Us,b, Ul,b, and Utrans. a fully developed flow (L/D > 2), the liquid axial velocity
Using these values in eqs 2 and 4, one can predict overall gas appears to be axially invariant (Degaleesan17). The general
holdup in homogeneous and heterogeneous regimes, respec- picture of a liquid axial velocity profile in a fully developed flow
tively. The average error of prediction using the newly is shown in Figure 2. The velocity is maximum in the center of
developed correlation was found to be 10% with a maximum
error of 40%. The range within which these physical constants
are valid is σ = 20−72 dyn·cm−1; μL = 0.0004−0.055 Pa·s; ρL =
683−2960 kg·m−3; ρG = 0.09−38 kg·m−3. Hence, it was
recommended to use the developed correlation of overall gas
holdup in bubble columns for scale-up purposes.
The method proposed by Wilkinson et al.8 relies completely
on similarity of global hydrodynamics such as overall gas
holdup. As described in the later part of this paper, Shaikh15
and Shaikh and Al-Dahhan16 have shown that such similarity
may not hold in a heterogeneous flow regime.
2.2. Degaleesan.17 Degaleesan17 proposed a scale-up
method that was based on the assumption that any gas−
liquid/slurry would exhibit the similar hydrodynamic behavior Figure 2. Radial profile of liquid axial velocity. Reprinted with
permission from ref 19. Copyright 1979 John Wiley & Sons, Inc.
as an air−water system if both the systems have the same
overall gas holdup. The procedure involves measuring (or
evaluating based on the suitable correlation) the overall gas the column and monotically decreases with negative values at
holdup in a scaled up unit at its operating conditions and then the wall. The dimensionless radial point where velocity reaches
calculating the equivalent superficial gas velocity, uGe that zero is called an ‘inversion point’ and is generally around 0.6−
provides the same overall gas holdup in an atmospheric air− 0.7 (Ueyama and Miyauchi19).
water system as the estimated one in the scaled up unit. Hence, The strength of liquid recirculation is described in terms of
it was suggested that hydrodynamics and mixing at the mean liquid recirculation velocity which is defined as
equivalent superficial gas velocity, uGe, in an atmospheric air−
ϕ*
water system would represent the hydrodynamics and mixing in ∫0 uz(ϕ)[1 − εG(ϕ)]ϕdϕ
a scaled up unit. urec
̅ = r*
To facilitate this, Degaleesan17 developed a two-dimensional ∫0 [1 − εG(ϕ)]ϕdϕ (11)
convection-diffusion model for liquid mixing to interpret the
tracer response in an 18-in. diameter slurry bubble column where ϕ* is the radial position of flow inversion.
reactor for liquid phase methanol synthesis at La Porte, Texas. In eq 11, the value of gas holdup radial profile is obtained
The convection-diffusion model needs knowledge of liquid axial using Computed Tomography (CT) while liquid axial velocity
velocity profile, eddy diffusivities profile, and gas holdup profile profile is obtained using Computer Automated Radioactive
to predict the tracer responses. The available fluid dynamic Particle Tracking (CARPT). The details of CT and CARPT are
measurements in an industrial unit were gas holdup radial available elsewhere (Kumar;20 Devanathan21) and will not be
profile measured using Nuclear Gauge Densitometry (NGD). repeated here. Based on her own CARPT and Kumar20 CT
8093 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

Figure 3. Determination of liquid velocity profile using the 1-D model and the known gas holdup radial profile (reproduced from ref 17).

Figure 4. Comparison of model predictions and experimental data in 18-in. diameter column at liquid phase methanol synthesis conditions and Ug =
36 cm/s. Reprinted with permission from ref 17. Copyright 1997 Washington University.

data and literature data from Nottenkamper et al.,18 the i. For a given process condition, with prior knowledge of
following correlation was developed for mean liquid overall gas holdup (either by measurement or correlation
recirculation velocity in an air−water system prediction), calculate an equivalent superficial gas velocity, UGe,
0.4 0.4 in an air−water system using eq 10.
urec
̅ = 2.2D UG (12) ii. At the equivalent superficial gas velocity, calculate mean
Eddy Diffusivities. Similar to the liquid recirculation velocity, liquid recirculation velocity using eq 12.
correlations were developed for cross-sectionally averaged axial iii. Calculate the radial profile of liquid velocity at equivalent
and radial eddy diffusivities using CARPT data in an air−water superficial gas velocity using the 1-D recirculation model
system as follows: (Kumar20) as explained in Figure 3.
iv. At the equivalent superficial gas velocity, calculate eddy
2325
Dzz
̅ =− + 106.6D0.3UG0.3 diffusivities profiles using eqs 13 to 16.
D0.8 (13) The liquid recirculation in its simplest form was modeled
using the 1-D recirculation model to predict the liquid axial
350
Drr̅ = − + 13.0D0.3UG0.3 velocity profile. This model requires two inputs: a gas holdup
D0.8 (14) radial profile and a closure for the turbulent shear stress (using
The correlations suggest the weak effect of scale on gas either eddy viscosity or mixing length). In the current case, gas
holdup, while the effect of scale on mean liquid recirculation holdup radial profiles were obtained from experimental
velocity and eddy diffusivities appears to be strong. Addition- measurements using CT. The other unknown is mixing length,
ally, the correlations to predict radial profiles of eddy to which the model was found to be very sensitive (Kumar20).
diffusivities using fourth and second order polynomials were Also, Kumar20 showed that there is no universal expression for
developed as follows: eddy viscosity or mixing length that can be used over a wide
4 3 2
range of operating and design conditions.
Dzz (ϕ) = Dzz
̅ {−3.4979ϕ + 3.2704ϕ + 0.4693ϕ Hence, Degaleesan17 used a guessed mixing length and
+ 0.00503ϕ + 0.5847} iterated until mean liquid recirculation velocity calculated from
(15)
the profile generated by 1-D recirculation model compares well
Drr (ϕ) = Drr̅ {−5.0929ϕ2 + 5.0717ϕ + 0.1653} with (i.e., converged to the) mean liquid recirculation velocity
(16)
predicted using the developed correlation [eq 12].
17
Degaleesan developed the following generalized method to Degaleesan17 suggested that the liquid axial velocity profile
estimate liquid axial velocity profile and eddy diffusivity profiles and eddy diffusivities profile estimated using this procedure at
in a scaled up unit using the available laboratory scale data and equivalent superficial gas velocity indicates the mixing
the above developed correlations: characteristics in an industrial scale unit as both the systems
8094 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

have similar overall gas holdup. She compared the prediction of found the same productivity in both the reactors when the
a two-dimensional convection-diffusion model with the values of βi (= 0.2) were the same.
experimental tracer data obtained in an 18-in. diameter slurry The method proposed by these authors shows that
bubble column for liquid phase methanol synthesis at LaPorte, maintaining a relative contribution of transport parameter the
TX. As shown in Figure 4, the model predictions are close to same in two systems results in the same performance even if
the experimental data. one system is impeller-agitated (CSTR) and the other one is
The proposed method provides a systematic approach to gas-agitated (slurry bubble column). Although the Inga and
characterize recirculation and mixing in an industrial scale Morsi23 procedure combines transport and reaction kinetics,
bubble column using an atmospheric air−water data. However, the use of overall mass transfer coefficient is not enough to
similarity based on only overall gas holdup is not sufficient in a describe the mixing as it represents global transport parameter.
churn-turbulent flow regime as shown by Shaikh.15 In addition, The authors described phase mixing in an industrial scale
this method needs a priori knowledge of gas holdup and its bubble column using ADM. The flow patterns in the bubble
distribution. The method relies on the use of a 1-D model to column are far more complex than plug flow superimposed
estimate circulation parameter. With advances in modeling with axial nonidealities. In addition, ADM needs reliable
techniques, one needs to utilize a more realistic description of prediction of an axial dispersion coefficient in both gas and
the system. slurry phase.
2.3. Inga and Morsi.22 Inga and Morsi22 developed a scale- 2.4. Fan et al.24 Fan et al.24 proposed a criterion for
up/scale-down methodology for bubble and slurry bubble hydrodynamic similarity in bubble and slurry bubble columns
columns based on similarity of the relative importance of mass based on the overall gas holdup. A correlation was developed to
transfer resistance in the overall reaction resistances. They have predict overall gas holdup in bubble and slurry bubble columns
demonstrated their procedure for FT synthesis where it was in terms of three dimensionless numbers. It was argued that
shown that the experimental results obtained in a laboratory maintaining these numbers the same in two systems would lead
scale stirred tank reactor could be extrapolated to design an to similar overall gas holdup and hence similar mixing and
industrial scale slurry bubble column. hydrodynamics.
The relative importance of mass transfer resistance was Fan et al.24 studied the effect of various operating parameters
defined in terms of a dimensionless parameter, βi, which that have a profound effect on overall gas holdup. Based on the
represents the balance between kLa (mass transfer coefficient) vast range of data collected from the literature and their own
and k0 (rate of consumption, pseudokinetic constant for first data, an empirical correlation was proposed to estimate the gas
order) as follows: holdup in bubble column reactors
⎛ u 4ρ ⎞α β

βi =
(1/kLa)
(1/kLa) + (1/k 0) (17)
εG
=
2.9⎜ σG gG ⎟
⎝ L ⎠
ρL () ρG

0.054 0.41
(1 − εG) [cosh(MoSL )] (18)
Accordingly, maintaining the same β in two reactors will
where
result in the same reactant concentration and catalyst activity
and thereby the similar conversion and selectivity in two 0.0079 −0.011
α = 0.21MoSL ; β = 0.096MoSL
reactors.
The procedure suggested by Inga and Morsi22 can be g(ρSL − ρG )(ξμL )4
summarized as follows: MoSL = 2 3
1. Measure kLa and k0 in a stirred tank reactor at the given ρSL σL
operating conditions.
2. Calculate the value of dimensionless parameter, βi, using ln(ξ) = 4.6CV {5.7CV0.58 sinh[− 0.71 exp(− 5.8CV )ln Mo0.22]
eq 17. + 1}
3. For a given industrial scale slurry bubble column reactor,
calculate the value of kLa using a correlation proposed by Inga The average error of prediction of overall gas holdup over the
and Morsi.22 Adjust the operating conditions such that the collected databank was 13% with a maximum error of 53%.
value of βi in the industrial scale unit is the same as the The physical meaning of the dimensionless group of u4GρG/σg
laboratory scale stirred tank. in the above equation was based on the maximum stable bubble
4. Calculate the values of gas and liquid axial dispersion size. Using the Davis-Taylor relation, the rise velocity of the
coefficients for an industrial scale slurry bubble column reactor maximum stable bubble size is
using available correlations. In their case, the correlation
proposed by Field and Davidson23 was used. ⎡ σg ⎤1/4
umax = C⎢ ⎥
5. Predict the performance of industrial scale slurry bubble ⎢⎣ ρG ⎥⎦ (19)
column reactor using an axial dispersion model (ADM) at the
adjusted conditions. If one substitutes this equation into the dimensionless group
Inga and Morsi22 demonstrated their proposed method for it becomes
FT synthesis using a laboratory scale 4-L stirred tank reactor
uG4ρG ⎛ uG ⎞
4
operating at 20 Hz and 5% wt which was simulated to a
conceptual industrial scale slurry bubble column reactor. The α⎜ ⎟
σg ⎝ umax ⎠ (20)
conceptual slurry bubble column reactor with 7 m diameter and
30 m height operating at 30 bar, 523 K, and 20 cm/s was This dimensionless number, therefore, signifies the con-
modeled using ADM. The simulations were performed to tribution of large bubbles to overall gas holdup. As the
maintain similar βi as in a stirred tank reactor. The authors correlation covers the wide range of operating conditions
8095 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

within acceptable statistical error, Fan et al.24 suggested that length, time), five independent dimensionless groups (Morton
hydrodynamic similarity requires three dimensionless numbers number, Mo; Etovos number, Eo; Reynolds number, Re; density
to be the same, i.e., uG/umax, MoSL, and ρG/ρL. ratio; superficial gas and liquid velocity ratio) were formed
This proposed similarity rule has not been evaluated according to the Buckingham-Pi theorem as follows:
experimentally. The suggested dimensionless numbers do not
g ΔρμL4 g Δρdp2 ρL dpuL
account for the effect of column diameter on gas holdup. The Mo = , Eo = , ReL = ,
reason for not accounting for the column diameter lies in the ρL2 σL3 σL μL
reported studies which found that, above 15 cm, column ρp uG
diameter has a negligible effect on the overall gas holdup. βd = , βu =
However, such studies were performed mostly but not ρL uL
exclusively in an air−water system. van Baten and Krishna25 In addition, the sixth group (ratio of gas to liquid density)
reported that in concentrated slurries such an observation does was suggested for high pressure conditions.
not quite hold as shown in Figure 5. Additionally, similarity The experiments were performed in two different column
criteria based only on overall gas holdup is not sufficient in a diameters. One column was an industrially operated cold flow
churn-turbulent flow regime as shown by Shaikh.15 unit of diameter 0.91 m. The gas, liquid, and solid phases used
in this column were hydrogen, kerosene, and ceramic particles,
respectively. The laboratory scale column had a diameter of
0.0826 m. The gas, liquid, and solid phases used in this column
are air, aqueous magnesium sulfate solution, and cylindrical
alumina particles. The geometric similarity between the
particles was maintained by using the particles of the same
length to diameter ratio, i.e., 2.6. The gas holdup was measured
based on the bed expansion and pressure measurements at
various axial locations. In the case of pressure measurements, it
was assumed that solids holdup is independent of height
throughout the column in the expanded bed.
The effect of gas and liquid Reynolds number appears to be
consistent with the previous literature findings (Figure 6).

Figure 5. Overall gas holdup curve in three different column diameters

using air−paraffin oil slurry. Reprinted with permission from 25.
Copyright 2004 American Chemical Society.

2.5. Safonuik et al.26 and Macchi et al.27 Matching the

dimensionless hydrodynamics groups or numbers derived from
mass and momentum balances such as Reynolds (Re) and
Froude (Fr) numbers has been used as a conventional scaling
rule. For example, van den Bleek and Schouten28 reported that
in order to maintain dynamics and geometric similarity, Re, Fr,
and dimensionless geometric numbers such as L/D should be
maintained constant.
Accordingly, based on such an approach, Safonuik et al.26
presented a scale-up method for three phase fluidized beds for
dynamic similarity. In the proposed method, Safonuik et al.26
identified the dimensionless numbers with the aid of the
Buckingham-Pi theorem that have a profound effect on
hydrodynamics of these reactors. The authors suggested that
Figure 6. Bed expansion versus Reynolds number in two columns.
the dimensionless hydrodynamic parameters such as overall gas
Filled-in point: 0.0826 m diameter column, open points: 0.91 m
holdup in two independent systems would be the same if the diameter column. Reprinted with permission from ref 26. Copyright
identified dimensionless numbers were matched in these 1999 Elsevier Ltd.
systems. Although the particle sizes used in their study, i.e.,
0.8 and 1 mm, were outside the range generally employed for
slurry bubble column reactors, this method is presented here as Although the exact operating conditions could not be matched
the concept that can be extended for bubble and slurry bubble for this set of experiments, the trends obtained in both columns
column reactors. are in line with the expectations. Whenever the dimensionless
Safonuik et al.26 identified eight variables that have a numbers were close, the overall gas holdup was found to be
significant effect on the hydrodynamics of three-phase fluidized close to each other. In addition, the gas holdup in the freeboard
beds as superficial gas velocity, superficial liquid velocity, liquid region was measured in both columns. The gas holdups in
viscosity, surface tension, particle diameter, liquid density, these two systems were found to agree well.
particle density, and buoyancy (Δρg). Using these significant Later, Macchi et al.27 tested the scaling approach proposed
variables that involve three fundamental dimensions (mass, by Safounik et al.26 for three phase fluidized beds based on
8096 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

dynamic and geometric similitude with five dimensionless

groups. The similarity of overall gas holdup in two systems was
investigated by maintaining these numbers the same by using a
monocomponent liquid in one system and a multicomponent
liquid in the other one. Essentially, they tested the ability of
Safonuik et al.26 approach to capture coalescing behavior.
Macchi et al.27 used air, 55% wt glycerol solution, and
spherical borosilicate particles as gas, liquid, and solid phases,
respectively, in one system (System I). While in another
system, air, silicone oil, and spherical porous silica−alumina
particles were gas, liquid, and solid phases, respectively (System
II). The experiments with both systems were performed in an
acrylic column of 0.127 m diameter and 2.58 m long. The
overall gas holdup was measured using a differential pressure
drop. They found negligible axial variation of phase holdups in Figure 8. Power spectra in the two systems at operating conditions
the bed and freeboard region at all conditions except the with the same dimensionless numbers. Reprinted with permission
highest gas and liquid velocities. The set of five dimensionless from ref 27. Copyright 2001 Elsevier Ltd.
numbers proposed by Safonuik et al.26 was maintained the
same as in the 0.127 m diameter column.
Figure 7 shows the comparison between the gas holdup in 2.6. Krishna et al.30 Krishna et al.30 developed a strategy
two systems at various gas and liquid Reynolds number. Both for scaling laboratory reactors to commercial ones based on
developing empirical correlations of various hydrodynamic
parameters to be incorporated into Computational Fluid
Dynamics (CFD) to design a commercial size bubble column
reactor. The developed procedure can be described as shown in
Figure 9.
They have performed dynamic gas disengagement (DGD)
experiments in columns with different diameters i.e. 10, 19, and
38 cm. using an air−paraffin oil−silica system at ambient
conditions. Based on DGD results, they have analyzed the
effect of different operating and design parameters on ‘small’
and ‘large’ bubble holdups. The bubble column was divided
into ‘dense’ and ‘dilute’ phases similar to gas−solid fluidized
beds. The ‘dilute’ phase consists of the large bubbles, while the
‘dense’ phase consists of liquid phase along with solid particles
and small bubbles. The ‘dense’ phase holdup was found to be
independent of column diameter but varied with solids loading
as shown in Figure 10. ‘Large’ bubble rise velocity was found to
be the function of column diameter. Using these experimental
results and their earlier findings, Krishna et al.31 proposed
Figure 7. Gas holdup in two systems: filled-in points − System I, open various phenomenological models/correlations to predict
points − System II. Reprinted with permission from ref 27. Copyright bubble diameter, rise velocity, and holdups.
2001 Elsevier Ltd. The dense phase holdup and small bubble rise velocity were
correlated as follows
⎡ ⎤
0.7 ⎥
the systems show the same trends. However, the gas holdup in εdf = εdf ⎢1 − υS
0⎢ εdf ⎥⎦
System II was slightly lower than in System I. The statistical ⎣ 0 (21)
analysis concludes that the gas holdups in the two systems are
within 11% root mean standard deviations. Hence, it does ⎡ 0.8 ⎥⎤
provide a reasonable basis for hydrodynamic similarity in the Vsmall = Vsmall0⎢1 + υS
two systems. However, the pressure fluctuations studies ⎣⎢ Vsmall0 ⎥⎦ (22)
revealed that the power spectra in the two systems are different
where εdf 0 = 0.27, and Vsmall0 = 0.095 for paraffin oil slurries
(Figure 8).
Macchi et al.27 concluded that the mismatch is likely due to (corresponding to υS = 0).
the difference between coalescence in monocomponent and The superficial gas velocity through dense phase can be
multicomponent liquid that results in different bubble calculated as
distribution. In the pure liquids, gas holdup decreases with an udf = Vsmallεdf (23)
increase in liquid phase viscosity, while in liquid mixtures it
goes through maxima. An initial increase in gas holdup with an To calculate the rise of single large bubble in an infinite
increase in viscosity in liquid mixtures is due to less coalescence volume of liquid, Krishna et al.31 modified the classical Davis-
rate than that of pure liquids (Wilkinson29). Hence, they Taylor relation and proposed the following correlation
suggested that more than five dimensionless groups are needed
Vb , single = 0.71(SF)(AF) √ gdb (24)
to fully characterize the system.
8097 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

Figure 9. Scale-up strategy for bubble column reactors. Reproduced from ref 30.

db = 0.069(uG − udf )0.376 (27)

These developed correlations were incorporated into CFD

using Eulerian description of fluid phases. The authors adopted
a two-phase model in this case assuming pseudohomogeneity.
In the experimental studies, they found that air−paraffin oil−
silica particles with 35% vol. solids loading shows gas holdup
curve close to that of air−Tellus oil (Figure 11). Interestingly,
the pseudoviscosity of the slurry is close to the viscosity of
Tellus oil.
Hence, slurry phase was modeled in their simulation as
pseudoliquid phase with Tellus oil properties. No turbulence
model has been used to calculate velocity field inside large
bubbles. The overall gas holdup values predicted from CFD
were compared to experimental values for three different
Figure 10. Effect of column diameter and slurry concentration on column diameters and found to be in a good agreement (Figure
dense phase gas holdup. Reprinted with permission from ref 30. 11). In addition, they performed experiments to measure liquid
Copyright 2001 Elsevier Ltd. velocity profile in an air−water and air−Tellus oil system using
Pavlov tube. The reasonable agreement has been found in
The scale correction factor, SF, which accounts for influence experiments and simulations. Based on this confirmation,
of column diameter was given as
Krishna et al.32 simulated the behavior of column diameters up
⎧1 for db/D < 0.125 to 6 m and found that overall gas holdup decreases and
⎪
⎪ centerline velocity increases with an increase in diameter. The
SF = ⎨1.1 exp( −db/D) for 0.125 < db/D < 0.6 strong reduction in large bubble holdup was also observed with
⎪
⎪ 0.496 D/d an increase in column diameter. However, no comparison of
⎩ b for db/D > 0.6 (25) CFD predictions with experimental data was shown at these
The acceleration factor, AF, that accounts for an increase in conditions. The simulated centerline velocities showed a good
large bubble rise velocity over that of a single bubble rise agreement with Riquart33 correlation. Additionally, to predict
velocity was given as liquid phase backmixing, a correlation proposed by the authors
AF = 2.25 + 4.09(uG − udf ) for air−Tellus oil was recommended. Based on these observations, they
(26) suggested the Eulerian simulation model developed in their
The following empirical correlation was proposed for the study to be a valuable tool to predict the hydrodynamics of
average bubble size in the swarm bubble column reactors at commercial scale.

Figure 11. Gas holdup in air−paraffin oil slurry and air−Tellus oil in three different diameter columns. Reprinted with permission from ref 30.
Copyright 2001 Elsevier Ltd.

8098 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

Figure 12. Variation of overall, large, and small bubble gas holdup with superficial gas velocities at ambient pressure. Reprinted with permission from
ref 35. Copyright 2004 Washington University.

The correlations for prediction of bubble diameter, rise dimensional bubble column. Hence the correlations developed
velocity, and holdups (eqs 21-27) were developed using using DGD need to be revisited.
Dynamic Gas Disengagement (DGD) studies with an In addition, CFD data shown in this work were validated in
assumption that the small bubble holdup remains constant in laboratory scale reactors and were extrapolated to design a 6 m
a heterogeneous flow regime. However, Jordan et al.34 reactor.
performed DGD experiments at atmospheric as well as high 2.7. van Baten et al.37 van Baten et al.37 proposed a new
pressure and analyzed the obtained results using independent method for scale-up of bubble columns where they have
disengagement, sequential disengagement, and constant slip demonstrated how the hydrodynamic behavior in a 1 m
velocity assumptions. In contradiction to previous findings, they diameter column can be estimated using the experimental data
observed that the small bubbles holdup increases with an in a 5.1 cm diameter column. They developed a procedure to
increase in superficial gas velocity. The experiments performed study the flow behavior in commercial scale bubble columns
in 10 cm diameter column using butanol showed 100% and based on CFD in Eulerian framework by considering only a
67% increase in small bubble holdup from 5 to 20 cm/s using momentum exchange term. A method was devised to replace
an independent (and constant slip) and sequential disengage- the need for a closure equation that may mar the predictions of
ment technique, respectively. In addition, Xue35 studied bubble CFD. They calculated a drag coefficient and a bubble diameter
dynamics in an air−water system at atmospheric and high utilizing only overall gas holdup data in a small diameter
pressure using a four point optical probe. As shown in Figure column (5.1 cm).
12, at dimensionless radius locations of 0, 0.6, and 0.9, an Van Baten et al.37 performed experiments in a 5.1 cm
increase in large and small bubble population with an increase diameter column using an air−ethanol−cobalt (Co) system at
in superficial gas velocity was observed. Lee et al.36 studied the various solids loading i.e. 4.4, 8.5, 12.5, and 16.4% wt.
assumptions of the DGD technique in a two-dimensional Insignificant change in the overall gas holdup was found
bubble column and a slurry bubble column with the help of between 12.5 and 16.4% wt solids loading system as shown in
Particle Image Velocimetry (PIV). They found that the Figure 13. The CFD model was developed in an Eulerian
assumptions in DGD that there is no bubble−bubble framework for operating conditions with 16.4% wt solids
coalescence and breakup during disengagement and disengage- loading assuming pseudohomogeneity. Based on the work of
ment of bubble classes are not affected by each other and are Sanyal et al.38 and Sokolichin and Eigenberger,39 the added
not valid particularly in a heterogeneous regime. However, such mass and lift force terms were neglected. The momentum
limitations shown by Lee et al.36 have to be checked in a three- exchange term was written following the work of Pan et al.40 as
8099 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

v. The value of bubble swarm velocity at vanishingly small

superficial gas velocity is Vb,0. Substitute this value in eq 29.
This way, one does not need to know the bubble size in the
momentum exchange term to predict the drag coefficient. The
overall gas holdups predicted from CFD along with the
developed approach was compared to the one obtained from
experiments and were found to be in a good agreement. Based
on this comparison, van Baten et al.37 estimated the behavior of
a 1 m diameter bubble column.
Later, van Baten and Krishna25 demonstrated their approach
for different column diameters. As shown in Figure 15, for

Figure 13. Gas holdup curves in air−ethanol−Co system at various

solids loading in 5 cm diameter column. Reprinted with permission
from ref 37. Copyright 2003 Elsevier B.V.

⎡3 C ⎤
ML , G = ⎢ D
ρL ⎥εGεL(uG − uL)|uG − uL|
⎣ 4 db ⎦ (28)
For a bubble swarm rising in a gravitational field, drag force
balances the difference between weight and buoyancy so that
the bracketed term in eq 28 can be substituted as
3 CD 1 Figure 15. Average bubble-swarm velocity in air−paraffin oil slurry in
ρ = (ρL − ρG )g 2 three different diameter columns. Reprinted with permission from 25.
4 db L Vb,0 (29) Copyright 2004 American Chemical Society.
where Vb,0 is the rise velocity of bubble swarms at low
superficial gas velocities. vanishingly small velocity, practically the same bubble velocity
Equation 29 clearly shows that the knowledge of the bubble was observed for all three different diameter columns. Utilizing
rise velocity in an infinite volume of liquid is the only unknown the developed approach, van Baten and Krishna25 found good
parameter in the model. The authors proposed the following agreement between the overall gas holdup obtained from
method to calculate the bubble rise velocity: experiments and CFD simulations for three different diameter
i. Perform overall gas holdup experiments in small diameter columns (10, 19, and 38 cm). They studied the hydrodynamic
column (5.1 cm, in the current case). behavior of 1 to 10 m diameter bubble column by extending the
ii. Calculate average bubble swarm velocity, Vb as same approach.
Vb = uG /εG (30) The authors proposed a simple approach to study and design
the flow behavior in large diameter bubble columns, as it does
iii. Plot average bubble swarm velocity, Vb versus superficial not require a priori knowledge of bubble diameter and also it
gas velocity (Figure 14). does not need drag force coefficient closure. The only
iv. Extrapolate Vb data to low superficial gas velocity as knowledge needed is simple overall gas holdup experimental
shown in Figure 14. data in a laboratory scale column. However, the validation of
CFD simulation results with experiments in large diameter
columns need to be established. In addition, this method can be
applicable only in cases where the bubble size does not increase
significantly with superficial gas velocity. Vermeer and
Krishna41 showed that the ratio of kLa/εG is constant (∼0.5)
in a churn-turbulent flow regime in the air−Turpentine 5
system. Godbole et al.42 reached the similar conclusions in air−
water and air−Soltrol 130 systems. Vandu and Krishna43 also
found the similar results in three different diameter columns
(10, 15, and 38 cm) using an air−water system (Figure 16) and
concluded that the constant value of ratio of kLa/εG is due to
the fact that the effective bubble diameter is independent of
superficial gas velocity. Based on these findings, van Baten and
Krishna25 rationalized the application of the developed
approach for scale-up of bubble columns. Chaumat et al.44
developed a gas tracer technique to study Residence Time
Figure 14. Average bubble-swarm velocity in air−ethanol−Co. Distribution (RTD) and mass transfer in 0.2 m diameter and
Reprinted with permission from ref 37. Copyright 2003 Elsevier B.V. 1.6 m long bubble column. Various gases (nitrogen, carbon
8100 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

water−quartz sand, nitrogen-actual liquid medium-Cu based

catalyst powder) systems were studied. The actual liquid
medium was sodium methoxide and xylene. Hydrodynamic
parameters studied were overall gas holdup, gas holdup axial
profile, solids holdup axial profile [using Sedimentation-
Dispersion Model (SDM)], and flow regime transition.
Catalyst performance evaluation was accomplished by
performing experiments in a 1-L autoclave where the catalyst
preparation method was studied. In addition, the effect of liquid
phase medium, temperature, and pressure was evaluated. Based
on these experiments, xylene was chosen as liquid phase
medium and temperature was chosen to be lower than 120 °C.
In the process development part, preliminary experiments
were performed in a slurry-batch bubble column reactor of 50
mm × 5 mm diameter and 4500 mm height. They concluded
that reaction and separation of liquid products is essential for a
Figure 16. Variation of kLa/εG with superficial gas velocity in three commercial methanol synthesis process. Hence, a continuous
different diameter columns using an air−water system. Reprinted with slurry bubble column system that consists of a slurry bubble
permission from ref 43. Copyright 2004 Elsevier B.V. column reactor, a separator, and a slurry recycle unit was
recommended for synergy of coupling of two-step methanol
dioxide) and liquids (water, cyclohexane) were used at ambient synthesis. A continuous circulating slurry bubble column
pressure. However, they concluded that the relation between reactor with capacity of 2 tons per year was manufactured.
mass transfer and hydrodynamics appears to be more complex The syngas composition was maintained at CO/H2 = 2. The
than simplified linear relation (Figure 17). In light of such operation was carried out for 100 h. The catalyst started
conflicting results, such an approach may need to be revisited. deactivating after 50 h. The effect of temperature and pressure
2.8. Zhang and Zhao.45 Zhang and Zhao45 proposed a was studied. They the observed uniform catalyst concentration
scale-up strategy for low temperature methanol synthesis in a profile at the methanol synthesis conditions. The experiments
continuous slurry bubble column reactor. It involved studying were also performed in a 1-L autoclave at the same conditions
hydrodynamics in cold flow units, catalyst performance as that of theh circulating slurry bubble column reactor. The
evaluation in an autoclave, and process investigation in pilot- comparison of the results in these two systems showed that the
scale continuous slurry bubble column reactor. scale-up efficiency of the circulating bubble column reactor was
The study was performed over the following three steps: about 80% of that in the autoclave.
1) Hydrodynamics in a cold flow unit with different column The authors provided a comprehensive roadmap for
structure using the three-phase systems development of the low-temperature methanol synthesis in
2) Catalyst preparation and performance evaluation in an the circulating slurry bubble column reactor. The strategy tied
autoclave flow behavior and catalysis studies with that of process
3) Process exploration in continuous slurry bubble column engineering. However, no guidelines were provided regarding
reactor with a capacity of 2 tons per year. hydrodynamic similarity in cold and hot units.
Hydrodynamics experiments were performed in three 2.9. Shaikh15 and Shaikh and Al-Dahhan.16 Shaikh15
different (diameter = 0.042, 0.05, and 0.1 m) continuous slurry and Shaikh and Al-Dahhan16 proposed a new hypothesis for
bubble column reactors and tapered bubble column reactor. hydrodynamic similarity in bubble column reactors. They
The methanol synthesis was characterized by gas volumetric evaluated the proposed hypothesis utilizing advanced diagnostic
contraction; hence the tapered bubble column was explored. It techniques such as gamma-ray Computed Tomography (CT)
has a conical and cylindrical section such that the diameter of and Computer Automated Radioactive Particle Tracking
the conical section increases from 0.1 m at the bottom to 0.2 at (CARPT). The details of these techniques are available
the top. Above the conical section, a cylindrical section of 0.2 m elsewhere (Degaleesan;17 Kumar20) and hence will not be
diameter was placed. The different gas−liquid (air−water, repeated here. Essentially, CT provides time-averaged cross-
nitrogen-actual liquid medium) and gas−liquid−solids (air− sectional distribution of phase holdups and CARPT provides

Figure 17. Relation between mass transfer coefficient and hydrodynamics. Reprinted with permission from 44. Copyright 2005 Elsevier Ltd.

8101 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

instantaneous liquid velocity and its radial profile as well as a profile of liquid axial velocity and turbulent parameters) in
turbulent parameters profile. order to assess the proposed hypothesis.
The combination of two studies was the basis of a new They demonstrated that to be hydrodynamically similar it is
hypothesis for their studies. These studies were detailed necessary to have similarity of both overall gas holdup and gas
analysis of reported scale-up procedures of the bubble column holdup radial profile. Figure 19 shows one such result at a
and the results from Kemoun et al.46 and Ong.47 Based on the similarity condition. Figure 19a shows similar gas holdup
detailed analysis of reported hypotheses and scale-up profiles at the same overall gas holdup (0.35) which results in a
procedures, they found that most of these procedures similar liquid axial velocity profile (Figure 19b) and turbulent
inherently used similarity of global parameters such as overall kinetics energy (TKE) profile (Figure 19c). Thus, the similarity
gas holdup and/or mass transfer coefficient. In addition, they conditions showed that similar overall gas holdup and gas
combined the results from the work of Kemoun et al.46 and holdup radial profiles resulted in close liquid recirculation and
Ong47 as shown in Figure 18. It shows the gas holdup radial mixing intensity in both systems. Figure 20 shows experimental
results at one of the mismatch conditions. Figure 20a shows
mismatch gas holdup profiles at the same overall gas holdup
(0.35) which results in mismatched liquid axial velocity profile
(Figure 20b) and turbulent kinetics energy (TKE) profile
(Figure 20c). Thus, the mismatch experiments had similar
overall gas holdups but mismatched profiles and resulted in
varied liquid recirculation and mixing intensity. This clearly
shows that maintaining similar overall gas holdup alone can
lead to different recirculation and mixing, if gas holdup radial
profiles are not matched.
They also showed that the condition that two systems must
operate in the same flow regime to be hydrodynamically similar
is necessary but not suf f icient. It showed the importance of
matching gas holdup radial profiles or cross-sectional
distributions in two systems, even if both the systems operate
in the same flow regime. Using the available data of Ong,47
Shaikh and Al-Dahhan,49 and Shaikh,15 they showed that
traditionally used criterion for hydrodynamic similarity, based
only on global parameter, can be specifically applicable if both
Figure 18. Comparison of gas holdup radial profile in a 6” column the systems operate in bubbly flow.
using an air−water system at two different operating conditions In addition, dimensionless groups reported for hydrodynamic
[D6U12P7Water: 7 bar, 12 cm/s, an air−water (Kemoun et al.46); similarity were evaluated for available sets of similarity
D6U60P1Water: 1 bar, 60 cm/s, and an air−water (Ong47)] with conditions. They concluded that within the range of studied
similar overall gas holdups (∼0.41). experimental conditions, no consistent set of dimensionless
group exhibit similarity and hence the application and
profile obtained using CT in an air−water system (Ong;47 limitations of dimensionless approach for bubble column
Kemoun et al.46) at different operating conditions in a churn- reactors need to be investigated in detail.
turbulent flow regime with the overall gas holdup of 0.41. Although it was out of scope of their work, Shaikh and Al-
Although these systems have similar overall gas holdups, they Dahhan16 outlined a tentative procedure for extrapolating
have different gas holdup radial profiles that will clearly lead to hydrodynamic behavior. It involved a combination of the
different flow patterns and mixing intensities. proposed hypothesis and state-of-the-art correlations for
The conclusions of Macchi et al.28 and Figure 18 suggested needed hydrodynamic parameters developed using a wide
that the two systems can have similar overall gas holdups but range of operating and design parameters. These hydrodynamic
different flow patterns and mixing intensities. This indicates parameters are overall gas holdup, radial profile of gas holdup
that two systems can be globally similar in nature but have and liquid axial velocity, and center-line axial velocity. The
different local hydrodynamics. Hence, similarity based only on correlation developed for overall gas holdup based on Artificial
overall gas holdup does not appear sufficient. Neural Network (ANN) by Shaikh and Al-Dahhan50 is one
Based on this, Shaikh15 and Shaikh and Al-Dahhan16 such example.
proposed a new hypothesis as follows: The focus of their work was to propose and evaluate the
“Overall gas holdup and its time-averaged radial prof ile or hypothesis for hydrodynamic similarity using CT and CARPT.
cross-sectional distribution should be the same for two The development of reliable hydrodynamic similarity criteria is
reactors to be dynamically similar.” a goal of any scale-up procedure. Although, the proposed
The proposed hypothesis was evaluated using similarity and hypothesis was successfully evaluated at different operating
mismatch conditions. Similarity conditions are the ones that conditions and physical properties, it needs to be evaluated
have similar overall gas holdup as well as gas holdup radial using different column diameters and have the potential to
profile. While mismatch conditions are the ones that have apply in various multiphase reactors.
similar overall gas holdup and mismatch gas holdup radial 2.10. Youssef.51 Youssef51 proposed a scale-up method
profiles. These conditions were identified using overall gas based on the horizontal scale-up (i.e., scale-in-parallel or scale
holdup measurement and gamma-ray CT. Later, CARPT out) approach where the large scale unit consists of a
experiments were performed at identified similarity/mismatch multiplication of the small units and hence, geometry, flow
conditions to measure the detailed hydrodynamics (radial pattern, and flow regime are kept the same. His method was
8102 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

Figure 19. a) Gas holdup and b) axial liquid velocity c) TKE radial
profile in a 0.162 m diameter stainless steel column [D6P1U30 C9− Figure 20. a) Gas holdup and b) axial liquid velocity c) TKE radial
C11 oil: 6 in. diameter column, 0.1 MPa, 30 cm/s (Han48), and an air− profile in a 0.162 m diameter stainless steel column (D6P4U30water: 6
C9−C11 fluid system, D6P4U30water: 6 in. diameter column, 0.4 MPa, in. diameter column, 0.4 MPa and 30 cm/s, an air−water;
30 cm/s, and an air−water system] [overall gas holdup ∼0.35] D6P4U16C9−C11: 6 in. diameter column, 0.4 MPa and 16 cm/s,
(Shaikh15). air−C9−C11) [overall gas holdup ∼0.35] (Shaikh15).

8103 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

based on the approach of Kolbel and Ackermann6 in which the the latter issue, hexagonal type arrangement of the tubes has
reactor space was subdivided using similar vertical shaft, and the been recommended (Kolbel and Ackermann;6 Youssef51).
flow of gas was uniformly distributed among the equally spaced It is noteworthy that whether this method is feasible or not
compartments. The heat exchanging tubes were placed either including the approach of Sasol (Steynberg et al.52), the
inside the shaft or entirely or partly inside the spaces between methodology of Shaikh15 discussed above (Shaikh and Al-
the shafts. Such configuration was made to reduce the intensity Dahhan16) can be used to ensure hydrodynamics similarity
of liquid recirculation encountered in bubble and slurry bubble whether in a column of tube bundle or solid wall and for both
columns. Along the same concept of compartmentalization, the scale-in-parallel (horizontal scale-up) or vertical scale-up by
recent Sasol patent (Steynberg et al.52) proposed a solution for increasing the size. This is because the gas dynamics and its
the scale-up risk of the Fischer−Tropsch slurry bubble column radial profile dictate the hydrodynamics of the columns as
reactor by creating zones to mimic the behavior of small reactor demonstrated above (Shaikh and Al-Dahhan16).
diameter. They used a group of discrete channels inside the
reactor shell separated by heat transfer medium flow spaces. 3. REPORTED STATUS OF SCALE-UP IN INDUSTRY
They claimed the feasibility of scaling process, minimizing The status of scale-up of bubble column reactors in industry is
microscale mixing patterns and improving heat transfer. not widely reported in the literature for obvious proprietary
For relatively high exothermic reactions where bubble and reasons. Tarmy and Coulaloglou54 discussed fundamental
slurry bubble column reactors are favored including Fischer− development and scale-up issues and related them through an
Tropsch synthesis, a large number of heat exchanging tubes example of the coal liquefaction process. The traditional
that would cover a noticeable percent of the reactor cross approach for scale-up is shown in Figure 21 where kinetic
sectional area is needed. For example, Fischer−Tropsch
synthesis requires heat exchanging internals that would cover
∼22−25% of the reactor cross sectional area. Accordingly, for
these conditions Youssef51 proposed an arrangement of the
heat transfer tubes in a form of small bubble/slurry bubble
columns inside the reactor shell. In this case, the reactor is
compartmentalized in a form of small columns using heat
transfer tubes. The question that has escalated is whether such a
small column with a wall consists of tubes mimic the behavior
Figure 21. Traditional linear scale-up development. Reprinted with
of the same column size with a solid wall. Therefore, Youssef51 permission from ref 54. Copyright 1992 Elsevier Ltd.
studied the hydrodynamics of an air−water system in a 6-in.
diameter bubble column with a wall formed by 1 in. diameter
constants obtained from the laboratory data assists in
tubes placed inside an 18-in. diameter column using a 4-point
correlating the small pilot plant results. The pilot plant studies
optical probe that measured the radial profile of gas holdup,
are then followed by demonstration in a relatively larger unit.
specific interfacial area, bubble cord length, and bubble velocity
They presented an evolving interactive scale-up strategy (Figure
(upward and downward). The tube bundles were mounted 5 in. 22) that makes more efficient use of resources, time, and
above the gas distributor. The results were compared with technology.
independent results obtained by Xue35 and Wu53 in a separate
6 in. diameter bubble column setup with a solid wall. A higher
gas holdup was obtained at each corresponding radial location
in the tubes bundles compared to those obtained in a solid wall
column. The average relative difference of about 15% was
reported between them, particularly at the radial locations
within the central region of the columns. However, the shape of
the radial profile of the gas holdup inside the tubes bundle
column is about the same as that inside the solid wall column.
The bubble cord length and bubble velocity distributions at the
center of the tube bundle are close to that of the solid wall
column at 45 cm/s superficial gas velocity with the tubes
bundle mounted at about 5-in. above the gas distributor.
However, at the region close to the tube bundle there is a Figure 22. Evolving scale-up strategy. Reprinted with permission from
clear variation from that close to the solid wall. These variations ref 54. Copyright 1992 Elsevier Ltd.
could have been caused by the interaction between the inside
and outside the tube bundle that represent the column of 6-in.
diameter inside the 18 in. diameter column. Although Youssef51
claimed that such methodology for scale-up seems feasible, he Tarmy and Coulaloglou54 mentioned that understanding of
suggested that further work is required. hydrodynamics and related issues to be the most critical
It is obvious that this method has many uncertainties and element in development and scale-up of these reactors. The
technical difficulties associated with the vertical space between industrial example of scale-up demonstrated by them involves
the bottom edge of the tubes and the gas sparger and when the hydrodynamic studies based on overall phase holdups and
configuration is duplicated inside a larger diameter column; liquid backmixing in terms of the Peclet number measured
there will be gaps between tube bundles that promote mixing using pressure transducers and radioactive tracers. The highest
and interactions between inside and outside the tubes. To avoid superficial gas velocity in the pilot reactor at process conditions
8104 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

Figure 23. Eni/IFP hydrodynamic facilities. Reproduced from ref 59.

was 8 cm/s. Based on Tarmy et al.55 studies, this condition however, these studies with convergent objectives help in
appears to be in bubbly flow. improving our understanding of scaling rules and approaches to
Espinoza et al.56 discussed gas holdup prediction and scale- be applied to bubble column reactors. Due to the complexity of
up for Sasol slurry phase reactor. Espinoza et al.56 modified the flow in bubble columns, the scale-up of these reactors is still
gas distribution theory proposed by Toomey and Johnstone57 unanswered. The purely empirical approach for such a purpose
for slurry bubble column reactors where it was divided into should be strictly avoided. An ideal choice for designing
dense and dilute phases. The minimum fluidization velocity was commercial bubble column reactors would be fundamentally
replaced by transition velocity from homogeneous to based CFD models. However, Rafique et al.60 showed that
heterogeneous flow regime. It was observed that dense phase interfacial closures are still an unresolved issue, and tuning the
holdup was not affected by the column geometry, while for coefficients to a known field is still a state-of-the-art practice.
small diameter columns the dilute phase holdup is determined Additionally, available closures do not account for the effect of
by the column geometry. For large diameter columns (>1 m), turbulence. Chen61 implemented the Population Balance
the dilute phase holdup was constant for all the systems, while approach where the bubble population balance equation was
dense phase holdup was different depending upon gas and solved simultaneously with a solution of flow field. In a churn-
slurry properties. Using the modified two-phase theory, they turbulent flow regime, the breakup rate was enhanced 10 times
developed an approach similar to the one proposed by Krishna the predicted one to match experimental flow field results.
and Ellenberger.58 Espinoza et al.56 mentioned that during However, by increasing the break up rate 10 times for all his
scale-up of a 1 m diameter FT reactor to a 5 m diameter one, simulations, he found a good agreement between model
such an approach was not used. However, an importance of gas predictions and experimental data in different air−liquid
and slurry mixing characteristics in a slurry bubble column was systems. This shows that, even after incorporating detailed
emphasized. science, there are still unresolved issues. Hence, the scale-up of
Zennaro59 has briefly discussed the development of Eni/IFP bubble columns, like other multiphase reactors, is still more art
GTL FT and upgrading technology. Their strategy is based on than the science. However, an application of art needs a
three targets: thorough understanding of prevailing hydrodynamic phenom-
- development of an innovative FT/upgrading technology ena. Therefore, in the absence of fully evaluated CFD, a reliable
based on tailored catalysts, reactor design, and optimized similarity rule needs to be developed based on a phenomeno-
product upgrading logical approach.
- engineering studies of a fully integrated GTL complex Most of the reported scale-up procedures so far utilize the
- minimization of scale-up risk by developing appropriate
similarity of a global parameter such as overall gas holdup in
tools
two columns for hydrodynamic similarity. Such similarity based
As shown in Figure 23, experimental facilities of varied scale
on global parameters is not surprising because over the years
for hydrodynamic studies were developed to facilitate easy and
bubble column hydrodynamics have been quantified mostly
low-risk scale-up of reactor technology. In addition, hydro-
based on the global parameters such as overall gas holdup and
dynamic as well as mass and heat transfer, thermodynamics,
and kinetics have been described in detailed reactor model, mass transfer coefficient. However, as shown by Shaikh15 and
validated up to 2500 bpd equivalent size. Shaikh and Al-Dahhan,16 the generalization of hydrodynamic
similarity based on an overall gas holdup can be fatal and needs
to be exercised with prudence.
4. REMARKS The ultimate hope for design and scale-up of bubble columns
An overview shows that there have been numerous attempts remains on fully validated and verified CFD. This hope turning
toward developing procedures for hydrodynamic similarity and into reality depends on a number of issues related to CFD. In
subsequently for scale-up of bubble column reactors. Every this regard, Grace and Taghipour62 provided a comprehensive
proposed method has its own advantages and disadvantages; outline and comments. On an experimental front, the
8105 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108
Industrial & Engineering Chemistry Research Review

researchers have improved measurement techniques from εdf 0 gas holdup in the dense phase corresponding to no solids
simple visual observation to advanced diagnostic techniques holdup, dimensionless
such as PIV, CARPT, CT, etc. These improved techniques εGtrans overall gas hold up at transition point, dimensionless
provide us with a better understanding of prevailing εG overall gas hold up, dimensionless
phenomena than before. However, the combination of an εL overall liquid hold up, dimensionless
experimentally evaluated scaling rule based on a phenomeno- εs overall solids hold up, dimensionless
logical approach and CFD would provide a promising avenue ξ constant in eq 18, dimensionless
for scale-up of bubble column reactors and needs to be α parameter in eq 18, dimensionless
explored. ϕ ratio of radial location to column radius, dimensionless

■ AUTHOR INFORMATION
Corresponding Author
ϕ*
βd
ρG
radial location at inversion point, dimensionless
ratio of particle to liquid density, dimensionless
gas phase density, kg m−3
*E-mail: [email protected]. βi dimensionless parameter in eq 17
ρL liquid phase density, kg m−3
Notes
ρP particle density, kg m−3
The authors declare no competing financial interest.

■
σL liquid surface tension, N m−1
ACKNOWLEDGMENTS μL liquid viscosity, kg m−1 s−1
νS solids loading, dimensionless
The authors are thankful to the High Pressure Slurry Bubble ρSL slurry phase density, kg m−3
Column Reactor (HPSBCR) Consortium [ConocoPhillips βu ratio of superficial gas to liquid velocity, dimensionless
(USA), EniTech (Italy), Sasol (South Africa), Statoil (Nor- β parameter in eq 18, dimensionless
way)] and UCR-DOE (DE-FG-26-99FT40594) grants that
made this work possible. Abbrevations

■
CD
NOMENCLATURE
drag coefficient, dimensionless
AF
ANN
CARPT
acceleration factor
artifical neural network
computer automated radioactive particle tracking
CV volumetric solids loading, dimensionless CFD computational fluid dynamics
D column diameter, m CT computed tomography
db bubble diameter, m DGD dynamic gas disengagement
dp particle diameter, m FT Fischer−Tropsch
DG gas dispersion coefficient, m2·s−1 GTL gas-to-liquids
DL liquid dispersion coefficient, m2·s−1 LDA laser doppler anemometry
Drr radial eddy diffusivity, m2·s−1 LPMeOH liquid phase methanol synthesis
Dzz axial eddy diffusivity, m2·s−1 PBM population balance model
Eo Etovos number, dimensionless PIV particle image velocimetry
g gravity constant, m·s−2 SF scale factor

■
k0 pseudo-first-order rate constant, s−1
kLa volumetric mass transfer coefficient, s−1 REFERENCES
L column length, m
ML,G interphase momentum exchange term, N·m−3 (1) Shaikh, A.; Al-Dahhan, M. A review on flow regime transition in
bubble columns. Int. J. Chem. React. Eng. 2007, 5, R1.
Mo Morton number, dimensionless (2) Deckwer, W. Bubble Column Reactors; John Wiley & Sons: 1992.
r radial location in the column, m (3) Wender, I. Reactions of synthesis gas. Fuel Process. Technol. 1996,
Re Reynolds number, dimensionless 48, 189.
U superficial gas velocity, m·s−1 [in Figures 13, 14, 15, and (4) Lehman, J.; Hammer, J. Continuous fermentation in tower
16] fermentor, I European congress on biotechnology, Interlaken, 1978; Part
UG superficial gas velocity, m·s−1 1, p 1.
UGe equivalent superficial gas velocity (eq 10), m·s−1 (5) Diesterwerg, G.; Fuhr, H.; Reher, P. Die Bayer-Turmbiologie.
UGtrans superficial gas velocity at flow regime transition, m·s−1 Industrieabwasser 1978, 7.
UL superficial liquid velocity, m·s−1 (6) Kölbel, H.; Ackermann, P. Apparatus for carrying out gaseous
Ul,b large bubble rise velocity, m·s−1 catalytic reactions in liquid medium. US Patent 2, 1958, 853, 369.
(7) Sie, S. T.; Krishna, R. Design and scaleup of Fischer−Tropsch
udf bubble velocity in dense phase, m·s−1 bubble column slurry reactor. Fuel Process. Technol. 2000, 64, 73.
umax rise velocity of maximum stable bubble size, m·s−1 (8) Wilkinson, P. M.; Spek, A. P.; van Dierendonck, L. L. Design
urec liquid mean recirculation velocity, m·s−1 parameters estimation for scale-up of high pressure bubble columns.
Us,b small bubble rise velocity, m·s−1 AIChE J. 1992, 38, 544.
uz liquid axial velocity, m·s−1 (9) Hikita, S.; Asai, S.; Tanigawa, K.; Segawa, K.; Kitao, M. Gas
Vb,single single large bubble rise velocity, m·s−1 holdup in bubble columns. Chem. Eng. J. 1980, 20, 59−67.
Vb bubble swarm velocity, m·s−1 (10) Hammer, H.; Schrag, H.; Hektor, K.; Schonau, K.; Kusters, W.;
VB∞ terminal bubble rise velocity, m·s−1 Soemarno, A.; Sahabi, U.; Napp, W. New sub functions on
Vb,0 bubble rise velocity at vanishingly small velocity, m·s−1 hydrodynamics, heat and mass transfer for gas/liquid and gas/
VL axial liquid velocity, m·s−1 liquid/solid chemical and biochemical reactors. Front. Chem. React.
Eng., [Proc. - Int. Chem. React. Eng. Conf.] 1984; Vol. 1, p 464.
Vsmall small bubble velocity, m·s−1 (11) Idogawa, K.; Ikeda, K.; Fukuda, F.; Morooka, S. Effect of gas and
Greek letters liquid properties on the behavior of bubbles in column under high
εdf gas holdup in the dense phase, dimensionless pressure. Kagaku Kogaku Ronbunshu 1985, 11, 432.

8106 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

(12) Reilly, I. G.; Scott, D. S.; DeBruijn, T. D.; Jain, A.; Diskorz, J. (34) Jordan, U.; Saxena, A.; Schumpe, A. Dynamic gas disengage-
Correlation for gas holdup in turbulent coalescing bubble column. ment in a high-pressure bubble column. Can. J. Chem. Eng. 2003, 81,
Can. J. Chem. Eng. 1986, 64, 705. 491.
(13) Idogawa, K.; Ikeda, K.; Fukuda, F.; Morooka, S. Effect of gas and (35) Xue, J. Bubble velocity, size, and interfacial area measurements
liquid properties on the behavior of bubbles in column under high in bubble columns. D.Sc. Thesis, Washington University, St. Louis,
pressure. Int. Chem. Eng. 1987, 27, 93. MO, 2004.
(14) Krishna, R.; Wilkinson, P. M.; Van Dierendonck, L. L. A model (36) Lee, D. J.; Luo, X.; Fan, L. S. Gas disengagement technique in a
for gas holdup in bubble columns incorporating the influence of gas slurry bubble column operated in the coalesced bubble regime. Chem.
density on flow regime transitions. Chem. Eng. Sci. 1991, 46 (10), Eng. Sci. 1999, 54 (13−14), 2227.
2491−2496. (37) van Baten, J. M.; Ellenberger, J.; Krishna, R. Scale-up strategy for
(15) Shaikh, A. Bubble and Slurry Bubble Column Reactors: Mixing, bubble column slurry reactors using CFD simulations. Catal. Today
Flow Regime Transition, and Scaleup. D.Sc. Thesis, Washington 2003, 79−80, 259.
University, St. Louis, MO, 2007. (38) Sanyal, J.; Vasquez, S.; Roy, S.; Dudukovic, M. P. Numerical
(16) Shaikh, A.; Al-Dahhan, M. A new methodology for hydro- simulation of gas-liquid dynamics in cylindrical bubble column
dynamic similarity in bubble column reactors. Accepted Can. J. Chem. reactors. Chem. Eng. Sci. 1999, 54, 5071.
Eng. (Special Issue on Gas-Liquid and Gas-Liquid-Solid Reactor (39) Sokolichin, A.; Eigenberger, G. Applicability of k-epsilon
Engineering), 2010, 88, 503, Invited Paper. turbulence model to the dynamic simulation of bubble column. Part
(17) Degaleesan, S. Fluid Dynamic Measurements and Modeling of − I: detailed numerical simulations. Chem. Eng. Sci. 1999, 54, 2273.
Liquid Mixing in Bubble Columns. D.Sc. Thesis, Washington (40) Pan, Y.; Dudukovic, M. P.; Chang, M. Dynamic simulation of
University, St. Louis, MO, 1997. bubbly flow in bubble columns. Chem. Eng. Sci. 1999, 54, 2981.
(18) Nottenkamper, R.; Steiff, A.; Weinspach, P. M. Experimental (41) Vermeer, D. J.; Krishna, R. Hydrodynamics and mass transfer in
investigation of hydrodynamics of bubble columns. Ger. Chem. Eng. bubble columns in operating in the churn-turbulent regime. Ind. Eng.
1983, 6, 147−155. Chem. Process Des. Dev. 1981, 20, 475.
(19) Ueyama, K.; Miyauchi, T. Properties of recirculating turbulent (42) Godbole, S. P.; Joseph, S.; Shah, Y. T.; Carr, N. L.
two phase flow in gas bubble columns. AIChE J. 1979, 25, 258. Hydrodynamic and mass transfer in a bubble column with an organic
(20) Kumar, S. B. Computed Tomography Measurements of Void liquid. Can. J. Chem. Eng. 1984, 62, 440.
Fraction and Modeling of the Flow in Bubble Columns, Ph.D. Thesis, (43) Vandu, C. O.; Krishna, R. Influence of scale on the volumetric
mass transfer coefficient in bubble columns. Chem. Eng. Proc. 2004, 43,
Florida Atlantic University, Boca Raton, 1994.
(21) Devanathan, N. Investigation of Liquid Hydrodynamics in 575.
(44) Chaumat, H.; Billet-Duquenne, A. M.; Augier, F.; Mathieu, C.;
Bubble Columns via Computer Automated Radioactive Particle
Delmas, H. Mass transfer in bubble columns for industrial conditions
Tracking (CARPT), D.Sc. Thesis, Washington University, St. Louis,
− effects of organic medium, gas and liquid flow rates, and column
MO, 1991.
design. Chem. Eng. Sci. 2005, 60, 5930.
(22) Inga, J. R.; Morsi, B. I. Effect of catalyst loading on gas/liquid
(45) Zhang, K.; Zhao, Y. A scale-up strategy for low-temperature
mass transfer in a slurry reactor: a statistical experimental approach.
methanol synthesis in a circulating slurry bubble reactor. Chem. Eng.
Can. J. Chem. Eng. 1997, 75, 872−881. Sci. 2006, 61 (5), 1459−1469.
(23) Field, R. W.; Davidson, J. F. Axial dispersion in bubble columns. (46) Kemoun, A.; Ong, B. C.; Gupta, P.; Al-Dahhan, M. H.;
Trans. Inst. Chem. Eng. 1980, 58, 228−236. Dudukovic, M. P. Gas holdup in bubble columns at elevated pressure
(24) Fan, L.-S.; Yang, G. Q.; Lee, D. J.; Tsuchiya, K.; Luo, X. Some
via computed tomography. Int. J. Multiphase Flow 2001, 27, 929.
aspects of high-pressure phenomena of bubbles in liquids and liquid- (47) Ong, B. C. Experimental investigation of bubble column
solid suspensions. Chem. Eng. Sci. 1999, 54, 4681. hydrodynamics −effect of elevated pressure and superficial gas
(25) van Baten, J. M.; Krishna, R. Eulerian simulation strategy for velocity, D.Sc. Thesis, Washington University in St. Louis, USA, 2003.
scaling up a bubble column slurry reactor for Fischer−Tropsch (48) Han, L. Hydrodynamics, back-mixing, and mass transfer in a
synthesis. Ind. Eng. Chem. Res. 2004, 43, 4483. slurry bubble column reactor for Fischer−Tropsch alternative fuels,
(26) Safoniuk, M.; Grace, J. R.; Hackman, L.; McKnight, C. A. Use of D.Sc. Dissertation, Washington University in St. Louis, 2007.
dimensional similitude for scale-up of hydrodynamics in three-phase (49) Shaikh, A.; Al-Dahhan, M. Characterization of hydrodynamic
fluidized bed. Chem. Eng. Sci. 1999, 54 (21), 4961. flow regime in bubble columns via computed tomography. Flow Meas.
(27) Macchi, A.; Bi, H.; Grace, J. R.; McKnight, C. A.; Hackman, L. Instrum. 2005, 16 (2−3), 91 (Special Issue on Tomographic
Dimensional hydrodynamic similitude in three-phase fluidized beds. Techniques for Multiphase Flow Measurements).
Chem. Eng. Sci. 2001, 56 (21−22), 6039. (50) Shaikh, A.; Al-Dahhan, M. Development of an artificial neural
(28) Van den Bleek, C. M.; Schouten, J. C. Deterministic chaos: a network correlation for prediction of overall gas holdup in bubble
new tool in fluidized bed design and operation. Chem. Eng. J. 1993, 53, column reactors. Chem. Eng. Process. 2003, 42, 599−610 (Special Issue
75−87. on Application of Neural Networks to Multiphase Reactors). Invited
(29) Wilkinsen, P. M. Physical Aspects and Scale-up of High Pressure Paper.
Bubble Columns, Ph.D. Thesis, University of Groningen, The (51) Youssef, A. Fluid Dynamics and Scale-up of Bubble Columns
Netherlands, 1991. with Internals, Ph.D. Thesis, Washington University, St. Louis, MO,
(30) Krishna, R.; van Baten, J. M.; Urseanu, M. I.; Ellenberger, J. 2010.
Design and scale-up of the bubble column slurry reactor for Fischer− (52) Steynberg et al. Coproduction of Hydrocarbons and Dimethyl
Tropsch synthesis. Chem. Eng. Sci. 2001, Vol. 56, 537. ether. US Patent 0230024A1, 2009.
(31) Krishna, R.; de Swart, J. W. A.; Ellenberger, J.; Martina, G. B.; (53) Wu, C. Heat Transfer and Bubble Dynamics in Slurry Bubble
Maretto, C. Gas holdup in slurry bubble columns: effect of column Columns for Fischer-Tropsch Clean Alternative Energy, Ph.D. Thesis,
diameter and slurry concentrations. AIChE J. 1997, 43, 311. Washington University in St. Louis, Saint Louis, MO, USA, 2007.
(32) Krishna, R.; van Baten, J. M. Scaling up of bubbling column (54) Tarmy, B.; Coulaloglou, C. Alpha-omega and beyond industrial
reactors with the aid of CFD. Chem. Eng. Res. Des. 2001, 79 (Issue 3), view of gas/liquid/solids reactor development. Chem. Eng. Sci. 1992,
283−309. 47 (13/14), 3231.
(33) Riquarts, H. P. Straumungsprofile impulsautausch durchmi- (55) Tarmy, B.; Chang, M.; Coulaloglou, C.; Ponzi, P. Hydro-
schung der flussigen phase in blasensaulen. Chem. Ing. Tech. 1981, 53, dynamic characteristics of three phase reactors. The Chemical Engineer
60. 1984, 18.

8107 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108

Industrial & Engineering Chemistry Research Review

(56) Espinoza, R. L.; Styneberger, A. P.; Jager, B.; Vosloo, A. C. Low

temperature Fischer−Tropsch synthesis from a Sasol perspective.
Appl. Catal., A 1999, 186, 13.
(57) Toomey, R. L.; Johnstone, H. F. Chem. Eng. Proc. 1952, 48, 220.
(58) Krishna, R.; Ellenberger, J. Gas holdup in bubble column
reactors operating in the churn-turbulent flow regime. AIChE J. 1996,
42, 2627−2634.
(59) Zennaro, R. From R&D to a successful scale-up: the Eni/IFP/
Axens GTL technology. Oil Gas Eur. Mag. 2007, 33, 88−91.
(60) Rafique, P. C.; Dudukovic, M. Computational modeling of gas-
liquid flow in bubble columns. Rev. Chem. Eng. 2004, 20 (3−4), 225−
375.
(61) Chen, P. Modeling the fluid dynamics of bubble column flows.
D.Sc. Thesis, Washington University, St. Louis, MO, 2005.
(62) Grace, J.; Taghipour, F. Verification and validation of CFD
models and dynamic similarity for fluidized beds. Powder Technol.
2004, 139, 99−110.

8108 dx.doi.org/10.1021/ie302080m | Ind. Eng. Chem. Res. 2013, 52, 8091−8108