doi 10.

1515/ijcre-2012-0023 International Journal of Chemical Reactor Engineering 2013; 11(1): 169–223

Ahmed A. Youssef*, Muthanna H. Al-Dahhan, and Milorad P. Dudukovic

Bubble Columns with Internals: A Review

Abstract: Most industrial bubble column reactors require the industries [1]. Bioreactions, fermentation, wastewater treat-
utilization of internal structures for heat transfer and/or for ment, alkylation, ozonolysis, hydrogenation, chlorination,
controlling the flow structures and back mixing in the sys- oxidation, Fischer–Tropsch (FT) synthesis, and liquid
tem. The internals denote all types of innards added to a phase methanol and dimethyl ether synthesis are among
bubble column, such as perforated plates, baffles, vibrating the main applications of bubble columns [2, 3].
helical springs, mixers, and heat exchanger tubes. In com- Bubble columns are preferred to other types of multi-
mercial scale bubble columns, instrumentation probes, phase reactors in these applications for a number of
down-comers, and risers with heat exchangers are all con- reasons. Compared to fixed beds, their superior heat
sidered. This review presents the state-of-knowledge of bub- transfer properties allow close to isothermal operation,
ble columns with internals. It starts with an introduction. The leading to improved selectivity [4]. Unlike agitated tanks,
second section discusses the horizontal internals, and the they provide good mass and heat transfer without moving
following section examines the studies involving vertical parts. Moreover, their ease of construction and operation
internals. put bubble columns ahead of both fluidized bed (or
ebulated three-phase fluid beds) and fixed bed (or trickle
Keywords: Bubble Columns, Horizontal Internals, Heat bed) reactors. In short, bubble columns are desirable for
exchangers their good heat and mass transfer, for their ease of con-
struction and operation, for their absence of moving
parts, and for their low maintenance costs. Their main
*Corresponding author: Ahmed A. Youssef, SABIC 1 Lexan Ln., shortcomings are significant phase back mixing and dif-
Mount Vernon, IN – 47620, USA, E-mail: [email protected] ficult scale-up and design due to the complex interaction
Muthanna H. Al-Dahhan, Chemical and Biological Engineering
between the phases. Their scale-up and design difficul-
Department, Missouri University of Science and Technology (MST),
Rolla, MO 65409, USA, E-mail: [email protected]
ties have been the subject of several studies in recent
Milorad P. Dudukovic, Washington University in St Louis, 1 Brookings years, but have yet to be tackled systematically enough
Dr St Louis, MO 63130, USA, E-mail: [email protected] to permit optimization of their performance. The advan-
tages and disadvantages of bubble column reactors are
summarized in Table 1.
1 Introduction The flow patterns in bubble columns are complex,
and the hydrodynamic characteristics relevant to trans-
Bubble columns, in their simplest form, are cylindrical port and back mixing are hard to predict quantitatively.
vessels in which gas is injected as bubbles through a Bubble columns for FT synthesis (and many other appli-
distributor (sparger), into a liquid (a two-phase column), cations) are operated at high gas throughputs (in the so-
or into a suspension of fine solids in a liquid (a three- called churn turbulent flow regime) and at high pressures
phase column). The flow in a three-phase column is in vessels of large diameters that approach or exceed 6 m
sometimes approached as a pseudo two-phase flow: the [5]. Research on the fluid dynamics and transport phe-
fine solids follow the liquid phase, so a pseudo homo- nomena involved in such bubble column reactors seeks a
geneous assumption can be made for the slurry (liquid– rational basis for their scale-up and optimization. To
solid) phase. In a continuous flow system, the gas may accomplish this, three different scales need to be care-
flow either with or counter to the liquid flow direction. In fully considered, namely, the molecular scale, the bubble
a semi-batch system, gas is sparged into a static liquid scale, and the reactor scale.
(slurry) medium. In either case, a high interfacial con- At the reactor scale, gas holdup, its radial distribution,
tacting area is provided between the liquid (or slurry) the liquid’s recirculation, and fluid back mixing are impor-
phase and the gas phase. tant measured (or modeled) parameters. Investigations of
Bubble columns as multiphase reactors (or contactors) bubble scale phenomena include quantifying and under-
are favored for a wide range of applications in the standing gas (reactants) to liquid and solid (catalyst) trans-
chemical, biochemical, petrochemical, and metallurgical port steps as well as models of breakup and coalescence

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
170 A.A. Youssef et al.: Bubble Columns with Internals

Table 1 Advantages and disadvantages of bubble column reactors. continuity (the overall conservation equation) to be
ensured, a downward liquid stream flows along the
Pros Cons walls of the reactor.
Good heat transfer ● A strong recirculation cell is thus produced.
Ease of construction and operation
Absence of moving parts Significant phase back mixing Liquid recirculation might be viewed as advantageous in
Low maintenance costs and processes requiring good mass and heat transfer capabil-
Good mass transfer Difficult scale-up and design ities, since it helps in maintaining isothermal conditions
inside the reactor [15]. However, it also has negative
behavior. At the molecular scale, fundamental chemistry is consequences, including a decrease in conversion, pro-
required to study the catalysts and the gas conversion motion of undesired secondary reactions, and problems
processes [1]. with uniform catalyst suspension. Over the last five dec-
Recently, Shaikh and Al-Dahhan [6] reviewed the var- ades, many researchers (Deckwer, Ueyama, Joshi, Fan,
ious flow regimes prevailing in bubble column reactors and Krishna, Dudukovic, and their coworkers, just to name a
noted four patterns: homogeneous (bubbly) flow, hetero- few) have used advanced measuring and modeling tools
geneous (churn turbulent), flow slug flow, and annular to unveil the characteristics of the complex flow
flow. In addition, a grey region exists between the bubbly dynamics in bubble columns. They have investigated
and the churn turbulent flow regimes, which is referred to the effects of operating conditions (gas and liquid flow
as the transition regime. These different regimes exhibit rates, catalyst renewal rates, temperature, pressure, and
very distinctive flow characteristics. For instance, the wet feed composition), design parameters (column diameter,
air oxidation of sewage sludge [7] is one example of many sparger design, catalyst size, and loading), and physical
biochemical processes where the bubbly flow regime is properties on global gas holdup, holdup radial profile,
achieved by applying low superficial gas velocities, bubble dynamics, liquid recirculation profiles and inten-
resulting in uniform bubble sizes and limited liquid cir- sities, liquid turbulent eddy mixing, and gas and liquid
culation. On the other hand, the FT synthesis is an exam- dispersion [16].
ple of a process where the heterogeneous flow regime is In spite of a significant improvement in understand-
crucial [8, 9]. In the latter, as in many other applications, ing the flow dynamics in churn turbulent flows, there is
high superficial gas velocities are necessary [10–12] since general agreement that the journey to a full understand-
such conditions maintain high productivity rates and ing of bubble column performance is still only beginning.
keep the solid phase (catalyst) always suspended [13, 14]. Recent developments in computational tools have
As higher superficial gas velocities are applied, the enabled a substantial increase in the number of publica-
non-uniformity of the interwoven gas holdup and the tions dealing with the computational fluid dynamics
liquid velocity radial profiles become more significant. (CFD) of bubble column flows [17–29). These studies
The following phenomena are observed with increased provide a promising foundation for phenomenological
gas velocity: models [30–33]. Nevertheless, “a prori” prediction of bub-
● A narrow distribution of small spherical bubbles, ble column’s churn turbulent flows is still not possible.
characteristic of the homogeneous flow regime, is Better turbulence closures are needed [34, 35], as well as
replaced by a broader bubble-size distribution. deeper understanding of bubble interactions, including
● Bubbles tend to coalesce and the large ones (no coalescence and breakup phenomena [28, 29, 36]. For
longer spherical) move toward the core region of the these reasons, it is essential that all CFD simulations for
column. these multiphase bubble column flows are validated
● A difference in mixture density between the core and against reliable experimental data [1, 37, 38, Rados,
the wall regions develops due to a parabolic gas 2003]. Validation is essential for the development of reli-
holdup profile along the column. able scale-up methodologies to bridge the gap between
● This density difference results in a difference in the laboratory scale columns and industrial units. It is note-
buoyancy forces in the cross-section of the column. worthy that such studies are rare in the open literature.
The magnitude of these buoyancy forces is much lar- It is also noteworthy that most of the above-cited
ger in this churn turbulent flow regime and induces a studies were performed in empty columns, a circum-
strong liquid recirculation. stance which unfortunately leaves uncertainties about
● Liquid flow is driven upward by the above-mentioned the impact of internals on the above parameters and
large bubbles in the column’s center, and in order for phenomena. Internals are inserted into bubble columns

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 171

for two reasons: they serve as exchangers for heat trans- throughputs, high solid loading, and high temperature
fer and they control the extent of liquid recirculation [39]. and pressure require heat exchanging internals. Most of
In addition, internals compartmentalize the reactor into a the reactions conducted in bubble columns are exother-
series of mixed tanks that approach favorable plug flow mic, and maintaining safe isothermal conditions entails
behavior. Intuitively, when internal barriers are placed as transferring that heat outside of the bubble column.
obstacles inside the flow field, large bubbles break into Sample reactive systems are given in Table 2 [2] to show
smaller ones, ensuring that the interfacial area density the necessity of utilizing such internals in these multi-
between the dispersed gas phase and the continuous phase reactors.
liquid phase remains large [40]. In other words, internals Generally speaking, the following methods of heat
should provide a high gas holdup, a more uniform bub- transfer can be used [42]:
ble size distribution, and high mass transfer rates. ● Direct heat transfer (e.g., cooling by evaporation)
● Indirect heat transfer through the reactor wall or addi-
tional heating or cooling installation (the focus of this
2 Bubble columns with internals – review)
● Indirect heat transfer to and from an external heat
an overview exchanger operating in the recycle loop
● Heat transfer through an intermediary heat carrier
In the literature, most researchers have not reported medium
studying the effect of internals as a major design para-
meter on bubble column performance. However, most Schlüter et al. [2] and Dyer [43] assert that internal instal-
industrial multiphase reactors use internal structures for lations are the most suitable solution since they provide
heat transfer from/to the system and for controlling the both a reasonable ratio of reaction volume to heat trans-
flow structures and back mixing in the column. The fer area ratio and, in the case of highly exothermic reac-
“internals” denote all types of innards added to an tions, high-pressure steam. Moreover, they preclude the
empty (hollow) column, such as perforated plates, baf- need for either an external heat exchanger or a large and
fles, vibrating helical springs, mixers, and heat exchan- expensive slurry pump.
ger tubes. In commercial scale bubble columns, To our knowledge, few studies in the open literature
instrumentation probes, down-comers, and risers with have been conducted to examine the effect of internals
heat exchangers are all considered to be various forms on the performance and hydrodynamics of bubble col-
of internal obstacles [41]. umns, although they are essential in industrial setups.
Let us consider one specific example of an internal. Even fewer open publications describe the configuration
In many industrial bubble column reactors, the high gas of these internals and their geometrical details (size,

Table 2 Sample reaction systems operated in bubble column reactors.

Product Feed Heat of Reaction (kJ/mol) Pressure (bar) Temperature (°C)

Acetaldehyde Ethylene, Oxygen −243 3 120–130

Acetone Propene, Oxygen −255 10–14 110–120
Ethyl Benzene Benzene, ethylene −113 2–4 125–140
Benzoic Acid Toluene, Air (or Oxygen) −628 2–3 110–120
n-, iso-butyraldehyde Propene, Hydrogen, Oxygen −118/−147 7–25 90–120
Cumene Benzene, Propene −113 7 35–70
Cyclohexane Benzene, Hydrogen −214 50 200–225
Cyclohexanol Cyclohexane, Air −294 8–15 125–165

Cyclohexanone
1,2 – Dichloroethane Ethylene, Chlorine, Oxygen −239 15–20 170–185
Ethylene, Chlorine −180 4–5 40–70
Acetic Acid Acetaldehyde, Oxygen −294 2.3–2.5 50–70
Acetic Acid, Methyl ethyl ketone n-Butane, Air −1270 15–20 180
Vinyl Acetate Ethylene, Ethyl Acid, Oxygen −176 30–40 110–130
Wet air oxidation of sewage sludge Sewage sludge, air −435 50–150 200–300
Fischer–Tropsch synthesis Hydrogen, Carbon Monoxide −210 12–15 250–290

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
172 A.A. Youssef et al.: Bubble Columns with Internals

pitch, etc.). This guardedness can be attributed to the fact

that they are held as patents or trade secrets. For example,
Sasol usually shows a schematic of their 2,500 bbl/day
reactor with a cooling coil unit with no dimension details
[44]. Given the limitations on information in the open
literature, it is clear that the internal structures (horizontal
and/or vertical) required in industrial applications need to
be further studied to accurately account for their effect
on various parameters determining bubble column
performance.
This review assesses studies of bubble columns with
internals. Horizontal internals will be discussed in the
next section, and vertical internals will be considered in
the following section.

Figure 2 Effect of baffles on gas holdup [45].

3 Horizontal internals
As shown in Figure 2, the stationary baffles increased the
In this section, we discuss the studies performed on holdup by some 40–50%. The reciprocating movement of
bubble columns with horizontal internals, mostly perfo- the baffles further increased the holdup by some 25–30%
rated plates (trays) or tube bundles (rows and columns) over the stationary case, as shown in Figure 3.
placed across the flow direction. A wide range of free In addition to the latter effect, baffle movement was
open areas of horizontal perforated plates, covering found to increase the heat transfer by some 10–15%, as
0.5–53%, with a variety of hole diameters and systems, can be seen in Figure 4.
have been utilized. Numerous studies highlight the The design of the horizontal baffles in the latter study
advantages of horizontal internal structures in improving might be a reflection of the background of the leading
flow behavior, leading to higher productivities, and these author, James R. Fair, Leader of the Engineering Research
are pointed out. Section at the Research Center of Monsanto Chemical
As early as 1962, sparged (or bubble) contactors had Company at the time. Fair, in addition to being a notable
become important for gas–liquid chemical reactions. Fair researcher in the field of bubble columns [46], has pub-
et al. [45] carried out experiments at Monsanto Chemical lished numerous articles on distillation columns and the
Co. in Missouri with an 18ʺ Plexiglas vessel equipped development of their tray designs [47, 48]. Furthermore,
with an assembly of 20 internal perforated-plate baffles
with 5.5ʺ of spacing and with varying open areas and
hole diameters of 9.5–33% and 0.125–0312ʺ, respectively
(Figure 1). In some of their tests, the baffles were given a
rapid reciprocating motion.

Figure 1 18ʺ baffled column [45]. Figure 3 Effect of the baffles movement on the gas holdup [45].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 173

cathode of the cell) was mounted, in a 30 cm high column

of square section (10
10 cm2). Different diameters of the
probe were tested (2.5, 4, 8, 12, 16, and 20 mm) to find how
most of the experimental data (up to probe diameters of
8–10 mm) could be fitted with the following relationship:
h i
StD ¼ f ðRe
FrÞ1=3 PrD 2=3 :

Increasing the probe diameter beyond 10 mm led to data

stratification, in which case the authors assumed that StD
could be related to the probe diameter as follows, StD~d−0.2.
Based on these investigations, Khoze et al. [50] confirmed
that the reduction in the heat transfer corresponding to an
increase in probe diameter was due to the introduced
Figure 4 Effect of moving baffles on heat transfer coefficient [45]. variation in the flow conditions in the gas–liquid bed.
In their second series of experiments, Khoze et al.
Fair et al.’s [45] study is one of the few investigations investigated the mass transfer coefficient for individual
performed on a pilot plant scale with internals, as well as glass cylinders in a staggered bundle (seven horizontal
one of the very few studies that have involved dynamic rows and 14 vertical) with tube diameters of 4 mm and a
internals. It did lack, however, quantitative consideration pitch-to-diameter ratio of 3.5. The convective mass trans-
of the power requirement. fer was found to vary independently of the position of the
Mashelkar and Sharma [49] reviewed the available experimental cylinder (to which the probe was mounted).
information on bubble columns at the time and worked The authors concluded that heat transfer in a dynamic
out a design problem for isobutylene absorption in sulfuric two-phase bed depends on the rate of heat transfer
acid to outline the core design features of these multiphase between the heat-transfer surface and the liquid film.
contactors. He noted especially that filling a bubble col- They further concluded that the interphase surface was
umn with packing increases the interfacial area between extensive, and that the evaporation was so rapid that it
phases due to the enhanced bubble break-up, resulting in a did not, practically, affect the overall rate of heat transfer.
higher gas holdup as well. Mashelkar also pointed out that Sekizawa and Kubota [51] studied liquid mixing in
sectionalizing a bubble column reduces the liquid back multistage bubble columns using a wide range of experi-
mixing substantially. Given the exothermic nature of the mental conditions, employing columns with different dia-
absorption problem he studied in his review (10 kcal/mol meters (5, 10, and 20 cm), heights (100, 120 cm), plate
of isobutylene reacted), given the assumed heat transfer spacing (10, 20, and 40 cm), plate hole diameters (0.2,
coefficient (4,000 kcal/m2 hr °C) for the system involved, 0.3, 0.5, 1.0, 1.5, and 2 cm), plate thicknesses (0.3, 0.5,
and given its temperature and operating conditions, he and 1.0 cm), and plate-free area fractions (0.07 up to
found that an adequate heat removal area of 10.5 m2 (in 0.202). Also, liquids of different viscosities were exam-
a reactor with a diameter of 120 cm and a height of ined. Their results show that as the open area of the plate
9,550 cm) was needed. For such heat removal area, he increases, the backflow ratio (κ) (defined as the ratio of
suggested the usage of 1ʺ ID stainless steel tube to be coiled the backflow rate to the net flow rate of liquid through
with a helical diameter of 25 cm and total length of 105 m, the plate) increases. Further, as seen in Figure 5, they
with cooling water circulating inside the coil. The work by show that as the superficial gas velocity increases, κ
Mashelkar is important in providing quantitative values for increases monotonously (except at low free areas where
the heat exchanging surfaces to help in preliminary design its profile exhibits a maximum).
criteria. However, Mashelkar did not provide any systema- The backflow ratio was found to be independent of
tic methodology for the choice of the tube arrangement or both the column’s diameter and the plate spacing.
for the percent open area to be utilized and did not discuss Moreover, Sekizawa and Kubota reported a sharp
in depth the most suitable materials for packing. decrease in κ with a decrease in the plate hole diameter,
Convective mass transfer was investigated using an but a gradual increase in κ at larger values of the hole
electrochemical method by Khoze et al. [50]. Their first diameter (Figure 6).
series of experiments used a single cylindrical tube of The effect of plate thickness was significant only
4 mm diameter to which a nickel probe (constituting the when small-diameter holes were used on the plates. In

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
174 A.A. Youssef et al.: Bubble Columns with Internals

Figure 5 Relations between κ and gas velocity at different values

of free plate area [51].

Figure 7 Effect of liquid properties [51].

Figure 7 shows the effect of liquid properties on the

backflow ratio at a constant liquid flow rate. Finally,
the authors found that the axial (longitudinal) dispersion
coefficient decreased when partition plates were added.
This study by Sekizawa and Kubota recognized the
effects of the details in horizontal plates and their spacing
in various systems on the ability to approach a desired
plug flow pattern for the liquid phase. The authors used
the backflow ratio as a characteristic of the liquid phase
mixing. Sekizawa and Kubota [51] claim that, for the first
time, an accurate measurement technique (thermal tracer)
was utilized to quantify the extent of liquid mixing.
Unfortunately, an empirical correlation was suggested by
Figure 6 Effect of plate hole diameter [51]. the authors, not a fundamentally based model for the
prediction of liquid mixing characteristics.
addition, the researchers observed a significant effect of The rates of heat transfer from a single heat transfer
liquid viscosity but an insignificant one for surface ten- tube and from different tube bundles were investigated by
sion (see Table 3 for properties of the liquids used). Aksel’rod et al. [52] using air–water and air–aqueous

Table 3 Properties of liquids used at 20°C.

Liquids Density [g/cm3] Viscosity Surface tension Key (Figure 11)

[g/cm.sec] [dyne/cm]

Ion-exchanged water 0.998 0.010 72.4

Glycerine aqueous 1.197 0.382 66.0
solution 1.176 0.165 66.5
1.154 0.0945 67.1
1.132 0.0558 68.5
1.083 0.0253 68.7
0.9 wt% iso- 0.997 0.00913 44.9
amylalcohol
aq. solution

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 175

Figure 8 (a) Experimental setup used in the investigation. (b) Configuration of the internals tube bundle – Aksel’rod et al. [52].

glycerin systems in a 0.3 m high column of 0.8
0.9 m2 The authors also concluded that this non-uniformity
cross-section. They used heat exchanger tubes of 25, 38, of flow distribution across the cross-sectional area (CSA)
and 59 mm diameters. At low gas velocities, they observed, increased perceptibly with “tight” bundles with small
a non-uniform operation of the gas sparger, resulting in spacing as well as with the number of rows in the bundle.
some parts of clear liquid and no bubbles. This condition Blass and Cornelius [54] conducted one of the earliest
yielded differences in the heat transfer coefficients between Residence Time Distribution (RTD) studies of solid and
the two zones of almost an order of magnitude. liquid phases in multistage bubble columns. They used
A schematic diagram of their experimental setup, an acrylic glass column of 14 cm diameter and 325 cm in
shown in Figure 8 (a and b), was given in an earlier height, equipped with four perforated plates having 2 and
study by Vorotnikova and Aksel’rod [53]. 4 mm diameter holes, with a percentage-free area (%FA)
They claimed that placing the bundles transversely to ranging from 1.1 to 36%. Air, water, and sand/glass spheres
the liquid flow should increase the total flow resistance were used as the gas, liquid, and solid phases, respectively.
and the aerated height of the bed (Figure 9). The differential distribution function of the liquid phase
(water) was obtained from conductivity measurements at
the reactor exit as a response to NaCl tracer pulse injection
in the liquid feed line at the reactor entry (from the bot-
tom). Solid RTD was, on the other hand, measured with
10–30 g sand with a radioactive Scandium-46 tracer track-
ing the solid phase in the column.
As shown in Figure 10, the authors observed RTD
curves to analyze the case of a single, fully mixed, stirred
vessel only at a high-%FA (36%). With increased gas flow
but decreased liquid flow, a backflow of liquid started to
occur. Gas cushions started to appear underneath the
plate and increased with an increased gas flow rate,
thus further contributing to liquid back mixing. This
phenomenon does not occur at higher liquid flow rates
and very high gas flow rates. Obviously, the number of
effective stages is less than the number of real stages,
Figure 9 Effect of tube bundles on the distribution of static levels according to the following equation:
of clear (bubble-free) liquid along the tray. Hn = 0.1 m, dt = 38 mm,
staggered bundle, Wliq = 6.3 m3/m-h, Wg = 1.1 m/s. (1) Sieve tray
 GD ’ 0:19
m0:19 w 
Neff
without tube bundle (free CSA = 6.7%), (2) Single-row tube bundle ¼ 1:061
(S1 = 1.42), (3) Two-row tube bundle (S1 = 1.42 and S2 = 1.26), (4)
N s w2Ld
Three-row tube bundle (S1 = 1.42 and S2 = 1.26), (5) Single-row
In the above equation wGD is the mean superficial gas
bundle (S1 = 1.13), (6) Two-rod bundle (S1 = S2 = 1.13), (7) Three-row
bundle (S1 = S2 = 1.13). S is the pitch between the tubes in the velocity and wLd, is the orifice liquid velocity of the plates.
staggered bundle (Aksel’rod et al. [52]). This correlation, along with the one developed by Sekizawa

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
176 A.A. Youssef et al.: Bubble Columns with Internals

Figure 10 RTD of liquid phase with various free area on sectionalizing plates [54].

and Kubota [51], indicates that modeling the liquid phase underneath the plates formed only of gaseous bubbles
mixing in bubble columns was still mainly dependent on leaving a clear liquid region below. However, this latter
empiricism. This work, however, surpasses that by provid- observation did not hold at higher liquid flow rates. This
ing a comparison between the system’s RTD and that of an observation is very interesting, since it is in agreement
ideal cascade and an ideal mixing tank, as shown in with that reported by Blass and Cornelius [54].
Figure 10. In their study, Blass and Cornelius found the Unfortunately, their liquid flow ranged from 0 to
solid RTD curves to be well fitted by their modified mixing 6.56 cm/s with no specification as to the threshold, after
cell model. In addition, and based on their experimental which the gas cushion disappears. Since the above study
evaluation of the liquid holdup, they concluded that they was limited to a maximum of 1 cm/s for the liquid velo-
could neglect the effects of the plate geometry and the city, one can assume with caution, due to differences in
distance of separation between plates as factors affecting both systems, that ~1 cm/s would be the breaking point
the liquid holdup. Moreover, they observed no significant above which the gas cushion phenomenon below the
bubble coalescence since the perforated plates regularly perforated tray would not occur.
resized the bubbles into smaller ones. The effect of the solid inserts is shown in Figure 11,
Chen et al. [55] investigated the gas holdup and the which reveals that the screen plate column (first setup)
pressure drop in bubble columns with mounted horizon- provides a higher restraining power on the bubbles’
tal plates. Two multistage columns were used by the movement than the perforated plate (second setup) or
investigators; the first was 7.5 cm in diameter, with an the empty column. The result is a considerably higher
overall height of 3 m. As the internals for the first col- gas holdup in the screen-plate column than that found in
umn, 37 plates made of 6 mesh/inch stainless steel wire the perforated plate or in an empty column.
screen were mounted 5 cm apart. The second column was The pressure drops were measured in both columns
a 5 cm Karr column [56] of about 4 m in height. The at a zero net liquid flow rate and are shown in Figure 12.
perforated plates used in this column were made from a From this data, Chen et al. represented the pressure drop
Teflon sheet with 1.27 cm diameter holes, leaving a free for screen and Karr plates by the following equations,
area of 53%. A total of 84 plates placed at 2.54 cm inter- respectively:
vals were attached to the column’s central axis. The
authors noticed almost spherical bubbles of uniform ΔPc ¼ 1:86
104 VG ; ΔPc ¼ 0:73
104 VG
size rising between stages with no interference, gently Another study, using the same internal geometry
sliding through the openings with no coalescence. They described for the first setup, was undertaken by Chen
also visually reported, at low liquid velocities, a region and Yang [57]. They used three different columns of

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 177

Figure 13 Draft tubes with perforated plates [58].

Figure 11 Gas holdup with and without horizontal internals [55].
In agreement with the findings of Sekizawa and Kubota
[51], Chen and Yang indicated a decrease in the value of
the axial dispersion coefficient due to the use of horizon-
tal screens.
Kawasaki et al. [58] investigated the effect of draft
tubes with perforated plates on gas holdup and mass
transfer coefficient in bubble columns. They used a bubble
column of 0.15 m diameter and 2 m in height in which no,
one, three, and five draft tubes of 0.1 m diameter and of the
same height, with 1, 3, and 5 perforated plates of variable
configurations were used (Figure 13). They noted that for
each N (number of perforated plates), both the gas holdup
and the volumetric mass transfer coefficient increased with
an increase in the superficial gas velocity. Their results also
reveal that as the ratio of the draft tubes to the perforated
plates ratio increased, both the gas holdup and the mass
transfer coefficient increased accordingly. They found that
Figure 12 Pressure drop due to liquid circulation in screen plate (o) the highest gas holdup as well as highest mass transfer
and in Karr-plate (•) [55].
coefficient values were obtained when the smallest hole
diameter was used on the perforated plate.
Contrary to the observations of Blass and Cornelius
0.05, 0.075, and 0.15 cm diameter and focused on the [54] and of Chen et al. [55], the manuscript authors report
hydrodynamic and mass transfer characteristics. Chen the absence of gas pockets (cushions) below the perfo-
and Yang found that when a high gas-flow rate was rated plates. The reason for such an absence might be the
used in a viscous system (air-CMC), the mesh failed to range of gas velocities used by Kawasaki et al. [58]
control the growth of the bubbles, and lower gas holdup (<0.04 m/s), the utilization of a batch liquid, or the pre-
values were obtained than expected. The observed sence of a free annular region between the draft tubes/
increase in the interfacial area was reported as the reason perforated plates and the column’s walls that might help
for the larger mass transfer coefficient (kLa). The investi- the gas bubbles in escaping and prevent an accumulation
gators did formulate a correlation for the mass transfer below the trays. There is no doubt that this phenomenon
coefficient in a dimensionless form, as follows: needs further investigation.
Al Taweel et al. [59] investigated the effect of honey-
Sh ¼ 1:13
104 Res1:76 Sc0:5 Re0:285
L comb inserts on axial mixing and gas–liquid interfacial

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
178 A.A. Youssef et al.: Bubble Columns with Internals

Figure 14 Effect of honeycomb structures on the interfacial area and axial dispersion coefficient [59].

area in bubble columns. Two types of honeycomb parti- decreased the liquid phase back mixing significantly.
tioning plates (two plates placed 0.5 m apart and eight Their experimental work was carried out in two acrylic
plates placed 0.2 m apart) were studied in a 0.1 m dia- bubble columns, one 6.2 cm in diameter and 77 cm high,
meter glass column operating continuously with liquid and the other 20 cm in diameter and 90 cm high. A co-
fed from the top and with compressed air introduced current air–water system was studied with perforated
from the bottom via a venturi sparger. As shown in acrylic sieve plates of varying free (open) areas (0.5,
Figure 14, an increase in the interfacial area of contact 1.48, 10.8, and 100%), which were used to separate the
by up to 120% was obtained, and the axial dispersion column into several sections. A one-dimensional (1D)
coefficient was found to increase by 60% in the hetero- dispersion model and the method of moments were
geneous flow regime. The authors concluded that the used to gather the RTD information generated from injec-
higher axial dispersion coefficient was compensated for tions of NaCl tracer pulses at the bottom of the column
by the larger interfacial area and the ability to use col- and collected via a conductivity probe at the top. The
umns of smaller diameters. authors found that with a decrease in the %FA of the
To fit their experimental data, the authors have perforated plates, the gas flow was redistributed more
examined the Axial Dispersion Model (ADM) and models uniformly, resulting in a reduction in the liquid recircula-
of tanks-in-series with and without back mixing. They tion velocities and leading to a decrease in the back
found that tanks in series with back mixing best repre- mixing effect. These phenomena are shown in Figure 15
sented the data for the case of two inserts, while tanks in for the larger scale column (20 cm diameter).
series without back mixing gave a better fit for the case of They correlated the dispersion number (1/Pe) with
eight inserts. They decided, hence, to use the ADM to their defined flow parameter (FP) as:
discuss their data, using the Péclet number as a fitting
parameter. Although the latter study was one of the first Ln ðDl =Vl LÞ ¼ 0:0402½lnðFPÞ2 þ 0:4626½lnðFPÞ  1:4203
to provide insight on the effect of internals on the inter-
facial area between the gas and liquid phases via experi- where FP = (Vg*%Free area)/Vl
mental investigation, it is obvious that the operating From the above correlation, it is obvious that the dis-
conditions at which the investigations were performed persion level is a strong function of the superficial gas and
are far from most industrial conditions. A maximum liquid velocities as well as of the %FA of the trays.
superficial velocity of only 1.55 cm/s for the gas phase As the studies above have shown, the %FA is a key
is enough to ensure a bubbly flow regime in their small factor in controlling back mixing. As more area is cov-
diameter (10 cm) setup. Consequently, it is debatable ered, the flow structure tends more toward plug flow
whether the same models would fit the data in case of a behavior, and the reactor is indeed sectionalized.
heterogeneous flow regime and a larger diameter column. However, until the 1990s, no investigation provided the
Liquid phase RTD studies were performed in a sectio- optimal free CSA that should be maintained to retain high
nalized bubble column by Palaskar et al. [60]. They found processing capacity. This appears to be a difficult answer,
that introducing horizontal perforated sieve plates given the wide range of applications for bubble columns.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 179

Figure 15 Variation of the liquid phase dispersion coefficient with (a) 0.5% open area, (b) 1.48%, (c) 100% [60].

A multi-stage bubble column reactor was modeled by benefits were detectable from more than four slurry
Maretto and Krishna [61] for the FT synthesis reaction. phase stages, as shown in Figure 17.
For a column vessel of 7 m diameter, the total dispersion The goal of increasing the number of stages is to
height of 30 m was divided into N stages (N ≥ 1). Staging obtain conversion levels ~0.95, while restricting the
of the slurry phase was achieved by introducing sieve inlet superficial gas velocity to less than 30 cm/s, a
plates, which also acted as spacers for the cooling tubes value beyond which conversion significantly drops. This
(Figure 16). can be clearly observed in the simulation results shown
Although the syngas conversion was found to in Figure 17. Furthermore, the increased productivity
increase as the number of stages increased, no significant associated with an increase of the number of stages

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
180 A.A. Youssef et al.: Bubble Columns with Internals

Figure 16 Schematic of multi-stage bubble column slurry reactor [61].

Figure 17 Syngas conversion as a function of number of stages [61].

necessitated the increase of the specific heat transfer area of bubble column reactors has also included studies on
of the cooling tubes. The study’s simulations show that trayed bubble columns. Kemoun et al. [62] investigated a
the reactor’s horizontal compartmentalization allows continuously operated trayed cold-flow bubble column
approaching plug flow for the two phases. which was scaled down from a commercial unit at
Although Maretto and Krishna’s [61] study was lim- DuPont. Time-averaged gas holdup measurements were
ited to simulation-based results with no experimental performed using γ-ray computed tomography (CT). The
investigation in large-scale reactive systems, it does pro- sieve trays used (the design of which is given in
vide an important and useful conclusion by stating that Figure 17) were made of 6.35 mm thick acrylic sheets
exceeding four stages will not improve the conversion with a total of 42 holes in each, yielding ~5% open CSA.
any further. This is an important statement in cost-related The tray down-comer occupied about 10% of the total
matters for industrial scale units. tray CSA. They found that the gas holdup distribution in
The research in the Chemical Reaction Engineering level 2, shown in Figure 18, was asymmetric (with a more
Laboratory (CREL) at Washington University in the field pronounced asymmetry as the liquid superficial velocity

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 181

Figure 18 (a) Design of sieve trays; (b) CT scans axial positions [62].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
182 A.A. Youssef et al.: Bubble Columns with Internals

increased). The gas holdup distribution was symmetric at The solids’ concentrations in the bottom of the col-
level 3 (the midpoint between the trays). They also con- umn were found to be higher than those at the top. This
cluded that the sieve trays functioned as obstacles for was explained by claiming that when the ascending par-
uniform bubbles, inducing bubble coalescence and thus ticles suspended in the slurry collide with the lower part
yielding some larger bubbles. The authors, however, of the tubes bank, they return back to the bottom of the
claimed that the overall gas holdup values were nearly column, causing an increase of solids’ concentration in
the same in both cases, with and without internals, even that part of the column.
though the axial and radial profiles were not. These con- The authors warned against the use of the available
clusions are, however, unexpected since the presence of correlations for the prediction of the gas holdup [64, 65],
horizontal barriers to the flow field would, intuitively, since these do not take into account the presence of
inhibit bubble coalescence, leading to an increase in internals. It is noteworthy that such a warning was pub-
gas holdup. The utilization of the CT technique at axial lished only in the 2000, since most of the work involving
positions right below the perforated trays indicates asym- bubble columns hitherto was conducted in columns with
metry, with gas holdup higher below the tray than in the no internals, resulting in correlations which might not be
down-comer region. This implies an agreement with ear- applicable in reactors equipped with internals.
lier visual observations regarding the accumulation of A study of the liquid phase back mixing in bubble
gas bubbles below the plates, with resulting contribu- columns equipped with partition plates was performed by
tions to increased gas holdup. Dreher and Krishna [66]. The main objective of their
Colmenares et al. [63] investigated the effect of air study was to investigate the effect of horizontal internals
velocity and solid loading, as well as of the solid particles’ on the liquid mixing behavior in columns of different
diameter on the gas holdup. They conducted their experi- diameters at high superficial gas velocities. RTD curves
mental study in a column with a square cross-section that of the liquid phase were obtained using a pulse tracer
contained a bank of 49 aligned tubes arranged in the technique, and four conductivity probes mounted axially
central part. The rods were of 2.54 cm diameter each, along columns of diameters 10, 15, and 38 cm.
installed in a squared, in-line arrangement (7 rows and 7 Predictably, uniform bubbles dispersed through the per-
columns) with a pitch-diameter ratio of 1.75 (Figure 19). forated plates prevailed at low gas velocities. In the churn
The researchers observed collisions between the bub- turbulent flow regime, gas caps were observed under-
bles inside the tube bank, causing the rupture of large neath the plates. However, the extent of these gas “cush-
bubbles into smaller sized ones and causing the increase ions” decreased with column diameter increase. The axial
in the gas holdup. Hence, Colmenares et al. agreed with dispersion coefficient values were found to increase with
both Palaskar et al. and Aksel’rod’s findings. gas velocities and column diameter. The investigators

Figure 19 Experimental setup and internals [63].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 183

He used a nitrogen-acidified dichromate solution system

in a (12
12
80 cm) column, in which a sintered glass
distributor was used as the gas sparger. The internal tube
bundle shown in Figure 21 had the following configura-
tion: five arrays of copper tubes, each of 8 mm diameter
with a transverse and longitudinal pitch of 16 mm, each
row of tubes aligned in the direction of Nitrogen flow.
Each array consisted of six tubes of 12 cm length. A
correlation for the estimation of the mass transfer fitting
the experimental conditions was developed and is given
as follows:

J ¼ 0:15 ðFr:ReÞ0:15

The author found that increasing the number of rows in

the tube bundle decreased the mass transfer coefficient.
Nosier claimed that the presence of tube bundles
decreases the axial dispersion (back mixing) relative to
Figure 20 Effect of gas velocity and column diameter on the liquid
that prevailing in empty bubble columns.
exchange velocities [66].
Such a conclusion shows consistency with earlier
findings discussed above in terms of the diminution of
reported the superficial liquid exchange velocity through back mixing when inserting horizontal internals.
the perforated plate (between compartments), in other However, the proposed empirical correlation was never
words the liquid circulation velocity at plate, to be inde- tested in other systems and was compared with
pendent of the column diameter, to marginally increase gas–liquid contactors of different nature (packed beds
with gas velocity, and to significantly increase with the [68, 69]), which renders uncertainty for universal utiliza-
increase in the partition plates open area, as shown in tion. Fortunately, Nosier compared his mass transfer
Figure 20. correlation with tube bundles to that in an empty bubble
Once again, following the work by Al Taweel et al. column [70], and a noticeable deviation is observed. This
[59], Dreher and Krishna [66] used the ADM to fit their comes as a confirmation of our noted caution when using
experimental data using the liquid exchange velocity as correlations developed in bubble columns without
an adjusted parameter. internals.
The mass transfer at an in-line tube bank immersed Doshi and Pandit [71] studied the effect of internals
in a rectangular column was investigated by Nosier [67]. on mixing behavior in a sectionalized bubble column.

Figure 21 Arrangement of tube bank in the bubble column [67].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
184 A.A. Youssef et al.: Bubble Columns with Internals

The %FA of the sectionalizing plates varied in the range The above correlation is remarkable since it attempts
of 4–23%. They noted that, due to sectionalization, each to provide an estimation of the gas holdup in both cases
compartment is isolated and behaved as a CSTR with an with and without internals.
inter-cell exchange velocity that decreased as %FA Alvaré and Al-Dahhan [15, 73] continued the work on
decreased. They reported an increase in both the mixing trayed bubble columns conducted at Washington
time and the pressure drop across the sectionalizing plate University in CREL. Their studies were carried out in a
with %FA decrease. In agreement with Kemoun et al. 19 cm Plexiglas glass column in which both air and water
[62], they concluded that an increase in gas holdup were fed concurrently. In the first study, they investigated
with the percent open area decreases (up to 15% the effect of superficial gas velocity in both empty and
increase in gas holdup when decreasing the %FA from trayed bubble columns as well as the effect of hole dia-
23 to 4%). They related the gas holdup increase to the meter and percentage of open area on the tray on the
visualized bubble coalescence and accumulation occur- overall gas holdup. Three different tray designs and their
ring below the sectionalizing plates at high superficial details, as shown in Figure 22, were used.
gas velocities. They reported little or no effect of the trays on holdup
In parallel to the study above, Pandit and Doshi [72] in the bubbly flow regime, in which case the bubble
discussed in more detail the effect of sectionalizing bub- diameter is smaller than the holes, and the bubbles
ble columns. In comparison with the bubble column with could easily flow through. For the churn-turbulent flow
no internals, they pointed out that the bubble size in the regime, the authors noticed an increase in the overall gas
sectionalized column was governed by the plate hole holdup due to the trays’ insertion, such increase agrees
diameter as well as by the %FA of the plate. The sectio- with the findings from the study by Chen et al. [55]
nalizing plates were responsible of the re-breakage of the mentioned above. The effect of different trays on gas
bubbles, which reduces the average bubble size. The holdup can be simplified as follows: an increase in the
authors confirmed experimentally the decrease in back gas residence time, a smaller average bubble size, and a
mixing in the sectionalized bubble column via a lower lower rate of bubble coalescence, which agrees with Chen
axial dispersion coefficient measured by their tracer tech- et al.’s findings. In addition, the tray hole diameter was
nique. An explanation for the increase in pressure drop found to have a stronger effect on gas holdup as com-
across the plates as the %FA decreases was provided as pared to the tray open area.
follows: the resistance offered by the hole to the gas flow In the second study, Alvaré and Al-Dahhan [15]
increases as less open area is available. A generalized investigated the liquid phase mixing to find that the
correlation for the prediction of fractional gas holdup for placement of perforated plates reduces the overall liquid
both cases (with and without horizontal internals) was back mixing by decreasing the liquid circulation velocity
attempted in their study and was given by: inside the stages. Also, the tray’s open area was found to
have a stronger effect on the extent of liquid back mixing
"G ¼ ð0:61 þ ð0:045
nÞÞ
ðVG Þ0:68 reduction than the tray’s hole diameter. The N-CSTR with
back mixing model was statistically proven to match the
where n is the number of sections (equivalent to Hc/D experimental data significantly better than the ADM.
ratio, with Hc being the static height of the bed), and VG It is obvious that there was much focus on modeling,
is the superficial gas velocity. but no conclusions were reached as to which is the best

Figure 22 Configurations and dimensions of the trays [15, 73].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 185

Figure 23 Different designs of sectionalizing plates [74].

model to use. Some argued that the ADM is preferred, as well as the dimensions of the internal heat exchanger,
while others have been in favor of the mixed tanks in which consisted of a bundle of tubes hung vertically, are
series (with or without back mixing). Furthermore, it is shown in Figure 24. The heat of reaction is removed by the
clear that CFD tools were not used for simulating flows in water circulating under pressure through a cooling tube
bubble columns with horizontal internals. bundle suspended in the reactor.
Mecaial and Sadik [74] also studied the effect of In Hall and Taylor’s study, the tube bundle had a
sectionalizing bubble columns on the hydrodynamics total external cooling surface of 19 ft2, with a central tube
and the RTD. Using liquid tracer (NaCl) similar to that (2 3/8ʺ OD) surrounded by eight tubes (1 1/16ʺ OD
associated with the conductivity probes technique and arranged in a 5 3/8ʺ diameter circle). It is clear from the
the bed expansion method, they investigated the effect figure that the header is subdivided within the bundle of
of horizontal tray internals in a 10 cm diameter and 54 cm cooling tubes. The authors claimed that it was easy to
high PVC column. Five types of trays were used and are remove the tube bundle to alter the cooling surface.
shown in Figure 23. They found an increase in the overall Hall and Taylor’s setup, shown in Figure 24, is
gas holdup with the presence of the partitioning plates considered one of the earliest that included such sus-
with the tray hole diameter as a significant factor, espe- pended internals immersed in the slurry phase. Hence,
cially in such a non-coalescing gas-liquid system. The the pilot scale reactor’s schematic diagram in Figure 24
transition from a bubbly to a churn-turbulent flow regime provides insight into the configuration, geometry, and
was found to occur at a higher gas velocity compared to position of heat exchanging internals inside the column
empty columns. They also developed a mechanistic sub- reactor.
model for the liquid phase velocity profile that matched Herbert Kölbel, Professor of Technical Chemistry at
well with most of the published experimental data. TU Berlin, is probably one of the first developers of
A summary of the above studies related to horizontal bubble column reactors. His name is associated with a
internals is given in Table 4 below. large number of publications (both papers and patents)
in the field. He was deeply involved in studies of the FT
process (being a student of Franz Fischer himself at the
4 Vertical internals Kaiser-Wilhelm Institute), covering both transport phe-
nomena and reaction kinetics. It is noteworthy that
The FT process has attracted the attention of many Kölbel was the director of the Institute of Technical
researchers since its discovery in the 1920s. For that Chemistry where Wolf-Dieter Deckwer obtained his PhD
process, Hall and Taylor [75] designed and operated a degree as well as his Habilitation. The following part of
fluid catalyst pilot plant to develop a basis for a suitable this review highlights some of the work accomplished by
reaction system for carrying out the exothermic and Herbert Kölbel and his co-workers.
highly temperature-sensitive reaction on a commercial In 1958, Kölbel and Ackermann [76] patented a slurry
scale. The details of the reactor they used in their study, reactor design for carrying out the FT process. Their

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 Table 4 Summary of research work on bubble columns with horizontal internals.
186

Research System Column and sparger features Internals Study

Fair et al. [45] Air–water Commercial scale 42ʺ diameter – 10ʺ 20 internal perforated-plate baffles with 5.5ʺ of Heat transfer
height/9ʺ ring with 0.03ʺ orifices spacing and open area and hole diameter of
9.5–33% and 0.125–0.312ʺ respectively (sometimes
baffles were reciprocally moving)
Khoze et al. [50] Nitrogen-NaOH solution Alkali-resistant Plastic: 10
10 cm2 CSA Glass cylinder tubes in a staggered tube bundle Heat and mass
with 30 cm height/dielectric gas (seven horizontal rows and 14 vertical) s/d=3.5 transfer
distributing grid of 2.4% open area (s is the pitch between the tubes and d=4 mm)
Sekizawa and Gas phase: Air Liquid 5 cam diameter (100 cm height)/ Perforated plates: 10 and 20 cm of spacing, 0.5 cm Liquid mixing and
Kubota [51] phase: ion-exchanged perforated plate distributor hole diameter and plate thickness with 0.07 free backflow
water, Glycerine aqueous area fraction of the plate
solution (4 different 10 cm diameter (100 and 120 cm height)/ 10, 20, and 40 cm of spacing, 0.2, 0.3, 0.5, 1, 1.5,
properties variations), and perforated plate distributor and 2 cm of hole diameter, 0.3 cm 0.5, and 1 cm of
0.9 wt.% iso-amylalcohol plate thickness, and 0.07 to 0.202 free area fraction
aq. solution of the plate
20 cm diameter (100 cm height)/ 20 cm of plate spacing, 0.5 cm hole diameter and
perforated plate distributor plate thickness with 0.0775 to 0.202 free area
A.A. Youssef et al.: Bubble Columns with Internals

fraction of the plate.

Aksel’rod et al. Air–water and Air–aqueous Same as Vorotnikova and Aksel’rod. [53]/ Single-, two-, and three-row tube bundle of different Heat transfer and
[52] glycerin (cross-flow) sieve tray (6.7% open area) pitch (1.13–1.6 cm) hydrodynamics
Air–water and Air- Plexiglas: 20 cm
20 cm/grid plate (30% Single tube (12/1.5 mm), in-line tube bundle with
transformer oil (counter- open area) spacing 2 cm
2 cm, in-line tube bundle with
flow) spacing 4 cm
2 cm, and staggered tube bundle

Authenticated | 10.248.254.158
with spacing 4 cm
4 cm

Download Date | 9/1/14 2:33 PM

Blass and Air–water, Air–water–sand Acrylic glass: 14 cm diameter-325 cm Perforated plates of 2 and 4 mm diameter holes Liquid and solid
Cornelius [54] (diameters 60, 78, and height/perforated plate (lowest one of leading to free cross-sectional areas of 0.011–0.36 phase mixing
200 µm), Air–water–glass sectionalizing plates) relative to column cross-section
spheres (diameters 48, 78,

Brought to you by | Aston University Library & Information

800 µm)
Chen et al. [55] Air–water Plexiglas: 7.5 cm diameter – 3 m height/ 37 plates made of 6 mesh/in stainless steel wire Gas holdup and
gas distributor consisting of four 0.15 cm screen were mounted 5 cm apart pressure drop
diameter hyperdermic needles
Glass: 5 cm Karr column – 4 m height/ 84 perforated plates placed at a 2.54 cm interval
same sparger as above were attached to the column’s central axis made
from Teflon sheet with 1.27 cm diameter holes
leaving a free area of 53%.
Chen and Yang Air – Water and 0.2% CMC Plexiglas: 0.05, 0.075, and 0.15 m 37 circular plates made from 6 mesh stainless steel Gas holdup, bubble
[57] aqueous solution diameter – 3 m height/four nozzles of wire screen sheets with a fractional free area of size, interfacial area
1.5 mm diameter for the first two columns 0.64 mounted 0.05 m apart on a central shaft of and mass transfer
and eight nozzles of the same diameter 5 mm diameter – plates were 1.5 mm smaller in characteristics
for the last column diameter than the column.
 Kawasaki et al. Air–water 0.15 m diameter-2 m height/single hole Multiple draft tubes with perforated plates: draft Gas holdup, mass
[58] sparger tubes of 0.1 m diameter and length each, perforated transfer
plates (0.003 m thick): 1 with 161 holes of 0.003 m
diameter each and 0.007 m square pitch, 3 with 161
holes of 0.006 m diameter each and 0.007 pitch,
and 5 with 78 holes of 0.009 m each and 0.01 m
pitch.
Al Taweel et al. Air–water (with addition of 0.1 m diameter-2.5 m height/two-phase Honeycomb partitioning plates (2 placed 0.5 apart, Axial mixing
[59] small quantities of venture sparger with 0.86, 1.17, and and 8 placed 0.2 m apart) each of 50 mm thick and
polypropylene glycol 1.6 mm throat diameters made of extruded aluminum with 0.4 mm wall
methyl ether) thickness – 7 mm hexagonal cell structure
Palaskar et al. Air–water Acrylic: 6.2 cm diameter-77 cm height, Acrylic perforated sieve plates with percentage free Liquid phase mixing
[60] and 20 cm diameter-90 cm height/ areas of 0.5%, 1.48, 10.8, and 100%
Kemoun et al. Air–water 0.2 m diameter-2.4 m height/10 laterals 3 sieve trays constructed of 6.35 mm thick acrylic Gas holdup
[62] welded to the main gas manifold with 40 sheet and contained 42 holes of 6.35 diameter each
or 200 holes of 350 µm holes in each. in a triangular pitch
Colmenares et al. Air–water–sieved sand of Acrylic: 38 cm square-2.2 m height/ 49 tubes in an aligned arrangement of seven rows Gas and solid
[63] diameter 130, 510, and and seven columns (pitch to diameter ratio of 1.75), concentration
2,300 each tube is of 2.54 cm diameter distributions
Dreher and Air–water Polyacrylic: 10, 15, 38 cm diameter – 6, 4, Two perforated brass plates of 1 mm thickness and Liquid phase back
Krishna [66] 4 m height/Brass perforated plate with 10 mm diameter holes. Open area of 18.6 and mixing
0.5 mm diameter holes and triangular 30.7%
pitch of 7 mm
Nosier [67] Nitrogen-acidified Prespex: 12
12
80 cm/G4 sintered- Five arrays (6 tubes each) of copper tubes each of Mass transfer
dichromate solution glass distributor (pores of 5–10 microns) 8 mm diameter and 12 cm length with equal
longitudinal and transverse pitch of 16 mm
Pandit and Doshi Air–water 0.41 m diameter – 2.87 m height/sparger 1, 2, or 3 sectionalizing plates with 269, 604, 607, Mixing time study

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
[72] with 0.136% free area (25 holes of 3 mm or 617 holes of 5, 8, 6 and 7 mm diameter
diameter each) respectively. The % free area range is 4–23%
Doshi and Pandit Air–water 0.41 m diameter – 2.87 m height/ 3 or 4 sectionalizing plates with 61, 269, 537, 604, Mixing behavior
[71] spargers with 0.136% (25 holes of 3 mm 607, or 617 holes of 10, 5, 5, 8, 6, and 7 mm characteristics.

Brought to you by | Aston University Library & Information

diameter each) and 0.6% (251 holes of diameter respectively. The % Free area ranged
2 mm diameter each) free area between 4–23%.
Alvare and Al- Air–water 19 cm diameter – 241 cm height/9.5 mm 4 trays of three different configurations: a) 12 holes, Gas holdup
Dahhan [73] diameter single point nozzle 1.74 cm hole diameter, 10.2% open area; b) 52
holes, 0.6 cm hole diameter, 5.2% open area; and c)
105 holes, 0.6 cm hole diameter, 10.2% open area
Alvare and Al- Air–water Same as above Same as above Liquid phase mixing
Dahhan [15]
Mecaial and Air–water and air–NaCl PVC: 10 cm diameter-207 cm height/ Two trays with five different configurations each Hydrodynamics and
A.A. Youssef et al.: Bubble Columns with Internals

Sadik [74] solution single point nozzle (10 mm diameter) and (see Figure 21 for details) liquid mixing
perforated plate (55 holes of 1 mm
diameter each giving 0.6% open area)
187
188 A.A. Youssef et al.: Bubble Columns with Internals

Figure 25 Schematic diagram of the reactor with inserted shafts

and cooling tubes. (1) cylindrical reactor, (2) gas inlet, (3) gas
product outlet, (4) the shafts, (5) regulator, (6) guide plates,
(7) pipe system for heat exchanging, (8) catalyst reactor through
Figure 24 Pilot scale FT synthesis reactor [75]. pipes, (9) supports, and (10) catalyst filling and emptying pipe [76].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 189

Figure 26 Various designs of slurry phase bubble column compartmentalized by means of shafts and with the presence of internal heat
exchanging tubes [76].

design was meant to decrease the disadvantageous recir- a useful increase of the temperature in that section. For
culating effect well-known to occur in large, commercial- that, two designs were suggested, as shown in Figure 27 (a)
scale columns. Kölbel and Ackermann proposed a way to and (b). In the first, an insulated layer of steam is
subdivide the reactor space using similar vertical shafts maintained in the annular space 9, while in the second
which were open at both the top and the bottom, as design, a gas buffer is formed by extracting gas into the
shown in Figure 25. The new design was meant to sup- annular region 4 from the liquid medium.
press the strong liquid recirculation that results from the The ultimate result of the Kölbel and Ackermann
large difference in buoyancy forces between the center of work was the elimination of the large vertical recircula-
the column and the walls. The gas supply to the shafts tion loops shown in Figure 28 and the formation of a
was equalized, and the reaction results were similar to stable liquid (slurry)–gas suspension system in each
what one would obtain in a reactor of the same size as shaft with uniformly sized gas bubbles and a uniform
one shaft with its own gas distributor. rate of rise, as shown in Figure 29.
The configurations of the shafts presented by the The latter study is significant since it not only
authors are shown in Figure 26 with “a” denoting the shows a thorough understanding of the needs in practice
vertical shafts and “b” indicating the heat exchanger but also offers a methodical solution to one of the
tubes. Figure 26(a), (b), (c), and (d) illustrates the heat major drawbacks of bubble columns (back mixing). It
exchange pipes positioned inside the shafts, while also provides insight on scale-up related issues.
Figure 26(e) and (f) shows the heat exchange pipes Compartmentalizing the bubble column might be the
located in part or entirely through the intermediate basis of a novel scale-up technique for bubble column
spaces between the shafts, respectively. To eliminate reactors in parallel. In other words, we seek to scale out
the dead spaces between the shafts caused by circular the reactor as opposed to the traditional scale up. As
geometry (Figure 26 (a, e, and f)), hexagonal geometry mentioned above, the latter study proved useful in gra-
was suggested (Figure 26 (b, c and d)). dually increasing the temperature farther from the gas
It was recommended that a vapor (steam) buffer be distributor located at the bottom of the column. To permit
maintained in the upper part of the cooling tube to allow this temperature increase, Kölbel and Langheim [77]

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
190 A.A. Youssef et al.: Bubble Columns with Internals

Figure 27 Close designs of the cooling tubes (Shafts not shown) [76].

Figure 28 Liquid recirculation in bubble columns [76].

proposed reducing the cooling surface of the heat Figure 29 Uniform flow distribution due to shaft’s presence [76].
exchanger tubes in a step-wise manner from the bottom
to the top of the reactor. This is done by having two or synthesis, less methane formation, and an increase in
more of the tubes combine into a single tube or into a the catalyst’s performance.
smaller number of tubes, as shown in Figure 30 (b), The axial distribution of the heat transfer surface
compared to the conventional design shown in area in the reactor is an important design constraint of
Figure 30 (a). The main results of the modifications industrial significance that was recognized about 40
were a higher CO conversion when performing FT years later by Maretto and Krishna [61] when modeling

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 191

Figure 31 42ʺ commercial-sparged contactor [45].

conversion. The other extreme, assuming plug flow beha-

vior for both phases, would produce the highest conver-
sion, but would also drastically affect the isothermicity. A
Figure 30 Conventional (a) and improved (b) cooling systems – temperature profile would then develop even with the
(1) reaction chamber, (2) gas inlet, (3) gas sparger, (4) gas outlet, presence of heat exchanging bundles. More details will
(5) vertical cooling pipes, (6) cooling liquid outlet manifold (header), be presented in coming pages.
(7) cooling liquid inlet manifold (header), (8) conduits, and (9) vapor The use of packing material in bubble columns has
collecting chamber [77].
been practiced for long time to render favorable flow
conditions. These include reducing the liquid phase
the FT synthesis in a slurry bubble column, as will be back mixing and improving mass transfer rates by
discussed later in this section. increasing the interfacial area between the gas and the
Fair et al. [45] investigated the heat transfer in a 42ʺ- liquid phases. Reviews written by Hofmann [78] and
diameter commercial-scale bubble column, shown in Gestrich and Harth [79], as well as the summary provided
Figure 31. The column contained 421.5ʺ OD aluminum by Deckwer [8], are good references for interested read-
tubes arranged in two concentric circles of 38.5 and ers. Following are samples of the studies performed in
34.5ʺ diameters, one of which was replaced by a copper packed bubble columns.
tube in which nine thermocouples were imbedded at One of the earliest works on packing bubble columns
three levels to accomplish the heat transfer study. was performed by Carleton et al. [80]. Their study
Measured temperatures of the gas–liquid mixture included a wide range of column diameters (Figure 32)
were completely uniform, both vertically and radially, to as well as packing materials and sizes, all of which are
within 1ʺ of the heating surface. The authors claim that summarized in Tables 5 and 6.
this uniformity indicated the very large degree of liquid Based on their extensive experimental results, they
back mixing. This same finding will also be implied in the reached the following conclusions comparing a packed
work by Maretto and Krishna [61]. bubble column at the same interstitial gas velocity
However, it is naïve to assume an ideal isothermal (defined as the velocity through space not occupied by
condition in an industrial reactor due to complete back the packing) with a column with no packing:
mixing with no further considerations. For example, a) The gas holdup increased, with an increase in the gas
Maretto and Piccolo [10] have found that, for a FT synth- mean residence time.
esis, the assumption of completely mixed CSTR for both b) The pressure drop was, surprisingly, not higher than
the gas and the liquid phase would yield a very modest for unpacked columns.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
192 A.A. Youssef et al.: Bubble Columns with Internals

c) The spread (variance) of gas and liquid residence times Table 6 Range of packings [80].
was considerably reduced.
d) The interfacial area available for mass transfer was Type Material Size Wall thickness Free space
(cm) (cm) (%)
increased.
e) The rate of mass transfer in a non-reacting system was Raschig Stoneware 3.8 0.43 75
increased. Raschig Stoneware 1.9 0.23 77
f) These findings are in accordance with later investiga- Raschig Stoneware 1.3 0.18 71
Raschig Stoneware 1.0 0.13 73
tions by Hofmann [78].
Raschig Porcelain 0.6 0.08 72
Lessing Stoneware 1.0 0.15 60
Packing bubble columns as means of improving Dixon Stainless steel 0.6 0.02 95
the interfacial area for gas–liquid contacting was Spheres P.V.C. 1.3 — 41
attempted by Voyer and Miller [81]. They used a pack-
ing type that had proven to be successful earlier, in

fluidized beds, which offer little resistance to the flow

and which occupy only 5% of the column’s volume.
Six-mesh cylindrical screen packing as well as six-mesh
corrugated screen packing (both of ½ʺ) were investi-
gated and compared to the case of the unpacked col-
umn. The liquid phase (NaOH solution and water) was
continuously fed through ¼ʺ holes on the column’s
wall, while the gas phase (N2 and CO2) entered through
a sieve plate. The researchers applied photographic,
chemical absorption, and physical desorption meth-
odologies for their measurements. Figure 33 highlights

Figure 32 Details of column [80].

Table 5 Dimensions of columns [80].

Column 1 Column 2 Column 3

Column diameter, cm 7.6 15.3 30.5

Funnel diameter, cm 2.5 5.1 7.6
a, cm 1.9 3.8 7.6
b, cm 1.9 3.8 7.6
c, cm 1.3 2.5 3.8
No. of holes in plate 6 27 102
Figure 33 Liquid holdup measurements [81].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 193

the effects of packing types as compared to an empty much higher interfacial areas in the column with packing
column, and it also shows the impact of utilizing NaOH leads to much higher volumetric mass transfer coeffi-
solution instead of water on the liquid holdup. cients, indicating a higher volumetric productivity for
Furthermore, the advantages of packing are manifested the packed column.
as fourfold increase in interfacial area at low gas velo- Following Voyer and Miller’s study, Mashelkar and
cities (0.5 ft/s) to about double the area at 2.5 ft/s (see Sharma [49] investigated the mass transfer in bubble
Figure 34). and packed bubble columns. The details of the packed
bubble columns used in their study are presented in
Table 7. With packing, higher gas holdup values were
obtained, which the authors explained in two ways:
The packing insertion created (1) non-coalescing condi-
tions and (2) a more tortuous path for the bubbles to go
through, resulting in slower rise velocities of the bub-
bles. The values of the effective interfacial area were
always improved in packed bubble columns as com-
pared to unpacked ones. An interesting observation
was made by Mashelkar and Sharma. It was shown
that if too much packing is used, occupying a lot of
space in the column, then the interfacial area based on
the column volume could be inferior to that of the
column with no packing, thus negating some of the
advantages of the packed column.
An additional advantage to the observed improve-
Figure 34 Interfacial area from visual observations [81]. ment in mass transfer is the reduced back mixing in
both gas and liquid phases, which results in the smal-
ler column’s volume. Finally, the authors found it jus-
Hence, it was concluded that discrete cylindrical screen tifiable to assume plug flow behavior for the liquid in
packing is preferred on mass transfer grounds. The over- the packed bubble column. Caution must be taken,
all liquid side mass transfer coefficient, Kl, for cylindrical however, when dealing with large diameter columns,
screen packing does not differ much from that of the as Magnussen et al. [82] reports; and Deckwer [8] con-
empty column. Only a small reduction at low liquid cludes that packing large-scale units may lead to an
velocities and a slight increase at higher liquid velocities increase in the dispersion of the liquid phase. An opti-
was observed in the packed column for the mass transfer mization step is thus needed to find the best packing
coefficient compared to the column with no packing. The material and size, to minimize pressure drop, to

Table 7 Details of packed bubble columns used in the study [49].

No. Column diameter Column height Material of Type of packing Fractional voidage of Type of run
(cm) (cm) construction used packing

1 6.6 120 Glass 3/8ʺ ceramic 0.615 (1) Continuous counter

Raschig rings current
2 10 66 Perspex 3/8ʺ ceramic 0.615 (1) Semi-batch
Raschig rings
1ʺ ceramic 0.774 (2) Continuous Co-current
Raschig rings
3/4ʺ glass spheres 0.42
3 20 90 Perspex 1ʺ ceramic 0.7 (1) Semi-batch
Raschig rings
1ʺ Intalox saddles 0.605
1ʺ Pall rings 0.93

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
194 A.A. Youssef et al.: Bubble Columns with Internals

maximize the holdup, and to decrease dispersion with- diameter) were used in a 0.0635 m diameter glass column
out pulsation. operated in a semi-batch mode, with nitrogen continu-
Aksel’rod et al. [52], in addition to their study of ously fed into water. The authors observed that gas
horizontal internals discussed earlier, also performed holdup remained essentially unchanged by the vertical
experiments using internal bundles of rods (tubes) internals as long as they occupy a small volume of the
along the flow direction. They claimed that placing the column. Only when 36% of the CSA of the column was
internal heat exchangers in the direction of the flow covered with internals was a pronounced increase in
decreases “the gradient of the static level of clear liquid” holdup observed. This is a significant observation since
(i.e., the bed expansion) as compared to the horizontal the gas holdup is a very important parameter affecting
structure. They preferred such configurations due to the the performance of bubble columns. The axial dispersion
decrease in the angle of attack provided for the two- coefficient increased because of the presence of vertical
phase flow (from 90° in the case of the transverse tube internals in the bubbly flow regime in which the coeffi-
bundle to 0° in the case of tubes bundled along the flow) cient is a strong function of the gas velocity distribution
leading to a better heat transfer with moderate gas velo- at the bottom of the column. The gas velocity distribu-
cities. They concluded that, irrespective of the type of tion, in turn, is largely affected by the structure of the
tube bundle used, the heat transfer coefficient is higher internals. However, in the slug flow regime at higher gas
than for the single tube case. They also investigated grid throughputs, the axial dispersion coefficient appeared to
plates as distributors as well, in a Plexiglas column with be insignificantly affected by the internal rods and bas-
a 20
20 cm2 cross-section arranged with four internals: kets. It is surprising that slugs could form since the
(a) a single tube, 12/1.5 mm; (b) an in-line tube bundle internals occupied a fairly large section of the CSA (up
with spacing of 20
20 mm; (c) an in-line tube bundle to 36%). One would have expected the bubbles’ sizes to
with spacing of 40
20 mm; and (d) a staggered tube be confined to the free spaces (pitch). The slug flow
bundle with spacing of 40
20 mm using water and regime appears mostly in small diameter columns (such
transformer oil as the liquid phase. In this case, they as the one used in this study), and that this flow regime is
obtained higher heat transfer coefficients on the axis of unlikely in industrial units.
the column at the center. The smallest values of the heat In a study of FT synthesis using bubble columns,
transfer coefficients were obtained at the column walls Kölbel and Ralek [86] provide “notes on the development
and in the lower parts of the bed. This was later con- of large-scale reactors.” They assert that the phenomena
firmed by Korte [83] as well as at our laboratory through increasingly encountered with the increase in the col-
Wu’s [84] extensive study on heat transfer in bubble umn’s diameter, such as a non-uniform gas distribution
columns with a wide range of operating conditions. and vigorous liquid back mixing caused by the tendency
Aksel’rod et al. [52] correlated their data for the latter of the gas to move upward closer to the column center,
study with the following: which causes a back flow of liquid along the column
walls, can be limited by inserting a vertical bundle of
Nu ¼ Cðlch =dt Þ0:12
ðGaÞ0:2=3
Reliq 0:3
Pr0:44 cooling tubes. Kölbel and Ralek suggested installing open
vertical shafts at the top and bottom and having a hon-
They emphasized that the tube bundle arrangement and eycomb cross-section. These internals were located above
the location of the tubes have a significant effect on the the distributor at such a height that the gas could be
rate of heat transfer and flow resistance, and that the introduced either directly into each shaft or uniformly
underestimation of these factors may lead to errors in into the reactor space via a gas injector centrally located
the design of bubble column reactors. in the column’s bottom. The internals constituting the
The back mixing characteristics of bubble columns heat exchanger (cooling tubes) can be arranged centrally
with vertically suspended tubes were investigated by around the shafts or in corners, as shown in Figure 35.
Shah et al. [85]. They examined various combinations The cooling surfaces of these shaft reactors can be
and diameters of cylindrical glass rods and screen bas- reduced from bottom to top to allow an increase of the
kets that were suspended symmetrically about the axis of suspension temperature with increased height. However,
the column to study their impact on the gas holdup and the authors did not provide any data for comparison
the dispersion coefficient. Six, 16 and 23 glass rods between their case-study and other systems where no
(0.004 m diameter); 2 and 4 glass rods (0.011 m dia- internals were used.
meter); 1 glass rod (0.032 m diameter); 1 and 2 screen Unfortunately, no cost estimation for the above
baskets (0.019 m diameter) and 1 screen basket (0.032 m design was provided by the authors, although increased

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 195

of the other adjacent plate (Figure 46). The authors claim

that the presence of the plates, which preferably range
from 4 to 16, arranged at an apparent circle of 0.65 to 0.75
of the column diameter, favors a higher extent of bubble
breakup due to a powerful entrainment of bubbles by the
liquid in the column. Hence an effective agitation is
provided. However, we expect the design and installation
of such a patented draft device to be cumbersome due to
its obvious complexity.
Figure 36 below shows the column (1), the plates (2),
and an aeration pipe (3).
Yamashita [88] studied the effect of vertical tubes
and rods on gas holdup in three different column dia-
meters, namely: 8, 16, and 31 cm inner diameter. In a tap-
water/air system, he attributed the observed increase in
gas holdup in columns with internals to the fact that
tubes and rods decreased the open CSA of the bubble
Figure 35 Honeycombed inserts with cooling pipes [86].
column where bubbles can rise, thus generating a higher
gas velocity. Yamashita [88] also found that gas holdup
installation cost must be considered in the bubble col- does not depend on the arrangement of vertical tubes;
umn’s design. however, it increased with both the number of tubes and
A draft device designed and patented by Hagino et al. rods (N) as well as with their outer diameter.
[87] is shown in Figure 36. It consists of a number of Interestingly, he also found that at a very small separa-
vertical plates coaxially arranged within the bubble col- tion distance between the internals (6 mm), the gas
umn such that each plate has one vertical edge in out- holdup decreased, showing a minimum at N = 70. In
wardly spaced overlapping relation to a vertical edge of this case, the retardation of radial bubble motions due
an adjacent plate and has the other vertical edge in to close internals was the reason Yamashita states for the
inwardly spaced overlapping relation to a vertical edge decrease. These conclusions, as provided by Yamashita,

Figure 36 Draft device for bubble columns-a) perspective view of the draft device and b) transverse sectional view [87].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
196 A.A. Youssef et al.: Bubble Columns with Internals

indicate that the percentage covered by tubes, the tube At Universität Dortmund, in the 1980s and early
pitch and the dimensions, are the most important factors 1990s, the researchers and coworkers of P.-M. Weinspach,
to be investigated in bubble columns with internals. investigated a variety of bubble column systems including
However, the gas holdup decrease attributed to small internals. The following two examples from this group
distances between the tubes is counter-intuitive. are the most comprehensive and well structured studies
Further examination of this effect is needed to ensure involving bubble columns with internals so far.
reproducible conclusions. Korte [83] investigated the heat transfer in bubble
O’Dowd et al. [89] used a slurry bubble column with columns with and without internals. He utilized both
and without internal baffles to generate data for the vertical and horizontal (cross) tube bundles in addition
hydrodynamic parameters that describe the behavior of to the empty column (equipped only with the heat trans-
both gas and solids under the influence of the additional fer probe) (see Figure 38). The latter arrangement is
internals. The baffles arrangement, as shown in favored when low pressure is applied inside the reactor,
Figure 37, consisted of five vertically arranged Plexiglas while the former is preferred when medium or high
rods of 0.019 m diameter, each 1.88 m high (in a 0.108 m pressure is needed or when steam is employed in the
diameter acrylic column). They used a dual conductivity heat exchanging process. Qualitative and quantitative
probe for the gas phase analysis, whereas the 1D sedi- evaluations of the heat transfer coefficients were
mentation dispersion model was employed for the solids’ performed.
behavior. The system studied consisted of glass spheres-
water-Nitrogen as the three phases. They claimed that gas
holdup, bubble size, and interfacial area depend upon
the presence of baffles. They disagreed with Shah et al.
[85], who asserted that no appreciable effect is present,
with baffles occupying up to 25% of the column’s total
CSA. Their results, report a larger bubble size at high gas
velocities in the case of baffled columns as compared to
empty ones. These results, however, are questionable
because the position where the probe was placed in the
relatively large free area between the baffles (Figure 37)
does not properly capture the internals’ effect. Such a
setting may result in the elimination of breakup phenom-
ena by the baffles and may favor the coalescence phe-
nomena occurring in the heterogeneous flow regime.
Their findings are summarized as follows: gas holdup
increases in the baffled column as compared to the un-
baffled one, and bubble size “increases” in the baffled
column at high gas velocities.
Figure 38 Schematic diagrams of the heat transfer study arrange-
ments [83].

The study mainly involved two bubble columns of 19.6

and 45 cm in diameter with heights of and 6.81 and 6.2 m,
respectively.
Several systems were used, with air as the gas phase
and with either water or a fluid of higher viscosity as the
liquid phase. Thirteen different configurations for the
tubes and for the position of the heat transfer probe
were utilized, as shown in Figure 39.
First, Korte investigated the heat transfer in bubble
columns without internals and found that an increase in
the superficial gas velocity (in the homogeneous regime)
Figure 37 Baffled internals design [89]. results in a steep increase in the heat transfer coefficient

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 197

induced liquid recirculation in the larger diameter

column.

4.1 Study of heat transfer with longitudinal

internals

As observed in the case of the empty bubble columns,

Korte reported that the heat transfer coefficient steeply
increases up to a gas velocity of 20 cm/s. With a further
increase in the gas flow rate, the curve levels off. Liquid
viscosity was found to have a significant effect on heat
transfer. However, neither the superficial liquid velocity
Figure 39 Arrangements of the longitudinal tube bundles and the nor the gas-induced local liquid velocity affected the heat
heat transfer probe [83].
transfer, as shown in Figure 41.
It is to be noted that in a system of low liquid
values. On the other hand, in the heterogeneous regime, viscosity and at low gas velocity, only a marginal
the gas velocity has a limited effect on the heat transfer increase in heat transfer was observed with the reduction
coefficient. Such profiles show agreement with those in the open CSA, while with systems involving liquids of
reported by researchers from the groups of Saxena and high viscosities, a drop in the open CSA results in a boost
Dudukovic, as will be presented below. The liquid phase of the heat transfer coefficient values. For example,
properties (viscosity, concentration, thermal conductiv- arrangement (11), shown in Figure 39 above, causes gas
ity, etc.) largely affect the heat transfer coefficient values bubbles to rise in the annular region between the tubes
as well. For example, an increase in the liquid viscosity and the column’s wall due to flow dispersion resulting
causes a significant decrease (>50%) in the heat transfer from the geometry of the bundle blocking the central
coefficient values (Figure 40). region. Exceptions to the latter case may also apply,
With regard to the effect of geometric parameters on depending on the configuration and density of the tube
heat transfer, it was found that the values of the heat bundle. Even with high viscosity liquids, the heat transfer
transfer coefficients are higher at the column’s core com- coefficient values were found to be almost unaffected
pared to the wall region. In addition, measurements in when arrangement (8) is utilized. One possible reason,
the 45 cm diameter column showed higher heat transfer as described by Korte, is that the small pitch prevents
coefficient values than in the 19.6 cm diameter column, bubbles from agglomeration. Hence, only small bubbles
which was explained by the enhancement of the gas- exist in the system.

Figure 40 Effect of superficial gas velocity and liquid viscosity on the heat transfer coefficient [83].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
198 A.A. Youssef et al.: Bubble Columns with Internals

Figure 41 Effect of the liquid velocity on the heat transfer coefficient [83].

Korte noticed that once the flow in the vicinity of the internals that becomes more prominent in systems of
probe is hindered by other tubes or by the column’s wall high liquid viscosity. At low liquid flow rates, in the
(arrangement (13)), lower heat transfer coefficients result. homogeneous flow regime, the gas-induced turbulence
This observation can be related to Wasan and Ahliwalia’s inside the horizontal bundle is responsible for the heat
[90] model based on the consecutive film and surface transfer. However, in the heterogeneous flow regime,
renewal theory, where the continuous passage of bubbles when large bubbles are expected to form but are instead
next to the heat source would result in the heat transfer destroyed by interaction with the horizontal tube bundle,
phenomenon. It is clear that the presence of an obstacle no additional turbulence occurs and no increase in the
(tubes or walls) in this surrounding would decrease the heat transfer was noted. The heat transfer increased with
heat transfer coefficients. an increase in the number of horizontal tube rows (only
Figure 42 (a and b) shows that, similar to the case of at liquid velocities vLo > 12 cm/s). However, no further
empty columns, there is a decrease in the heat transfer increase in the heat transfer was observed beyond the
coefficients with an increase in the radial position away utilization of three rows.
from the column’s axis. Doubling the spacing between two consecutive rows
Korte correlated, for the first time, his experimental leads to better heat transfer when low liquid but high
results for longitudinal internals using the following gas velocities are applied. At high liquid velocity, such
relation: an effect was reversed and doubling the spacing
resulted in lower heat transfer; whereas, at low gas
31 i0:84 0:2 tR 0:14 ηL 0:3
   
and low liquid velocities, the effect of spaces between
h
St ¼ 0:139 ReG FrG Pr2:26
L :Af : : ;
dR ηLW rows depends on the viscosity, which varies greatly
between superficial liquid velocity of 1 cm/s and that
which implies that α,ðνGo =Af Þ0:16 :A0:04
f ffi νG;eff 0:16 of 38 cm/s used in these experiments. In addition, no
Therefore, the effect of the free CSA with respect to effect of rows spacing was observed in air/water
the total column’s CSA can be accounted for using an systems; and as viscosity increases, only a marginal
effective gas velocity vG;eff . effect is reported.
As distances between the rows of horizontal tubes
increase, gas-induced liquid flow becomes weaker, and
4.2 Study of heat transfer with horizontal only a fraction of the large eddies, characteristic of churn
internals turbulent flows, are maintained. Korte reports that when
using 11 rows of horizontal tubes, lower heat transfer was
The investigations done by Korte with horizontal tubes obtained in the larger diameter column (45 cm diameter).
concluded that the superficial liquid velocity has, clearly, However, when using 5 rows of horizontal tubes, lower
a significant effect on heat transfer in case of horizontal heat transfer coefficients were obtained in the smaller

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 199

Figure 42 Effect of radial position on heat transfer coefficient [83].

diameter column (19.6 cm) (Figure 43), which means that Bernemann [91] studied qualitatively and quantita-
the column’s diameter is an effective parameter in deter- vely the fluid dynamics and mixing behavior (character-
mining the heat transfer in bubble columns with horizon- istics) of the liquid phase in bubble columns with vertical
tal internals. internals.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
200 A.A. Youssef et al.: Bubble Columns with Internals

Figure 43 Effect of column’s diameter and number of rows on the heat transfer coefficients [83].

Figure 44 Internals arrangements [91].

He used two experimental columns of 19 and 45 cm dia- ● The transition from the homogeneous to the hetero-
meter. Ten configurations of internals were used in these geneous flow regime was found to be independent of
columns, which are shown in Figure 44 below, in addition both the column’s diameter and the internals’ config-
to using the empty columns. The characteristic geometric uration and occurred at around an effective gas velo-
dimensions of the internals are given in Table 8. city of 3.5 cm/s where the effective gas velocity was
It is to be noted that the internals were formed of defined as:
25 mm diameter PVC tubes, which were fixed to the col-
umn’s top and extended to just above the gas distributor. VG;eff ¼ VG;o =Af
The main observations and findings from
Bernemanns’s work can be summarized as follows: With Af = (DR2–nR.dR2)/DR2

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 201

Table 8 Geometrical details of the tube bundles (after Bernemann [91]).

Bundle configuration (No.) 1 2 3 4 5 6 7 8 9 10

No of tubes 6 7 12 13 18 19 6 15 37 63
Relative free CSA 0.90 0.88 0.79 0.77 0.69 0.67 0.98 0.95 0.89 0.81
Relative wall surface area 1.79 1.52 2.18 2.32 2.97 3.11 1.33 1.83 3.06 4.50
Pitch (mm) 40 70 40 40 40 40 - 120 70 40
Tube diameter (mm) 25

Since the demarcation of the flow regime transition is a ● The addition of internals increases the liquid recircu-
very difficult task, the above conclusion is interesting lation as well as the central liquid velocity (Figures 45,
since it implies that studies on flow regime transition in 46, and 47). Later, Forret et al. [92] has demonstrated
empty columns (e.g., [6]) are still valid when internals are the same result using a liquid tracer technique in a
inserted. We do believe, however, that further investiga- 1 m diameter column.
tions are needed to test this finding. ● The enhancement in the central liquid velocity upon
● The placement of the vertical internals caused an the insertion of internals was found not to be due to
increase in the effective gas velocity because of the the decrease in the open CSA yielding the increase in
CSA covered by the tubes. A similar reasoning was vG,eff, but to the stabilization that the vertical internals
given by Yamashita [88]. impose on the flow.
● The liquid phase flow profile was maintained after the ● This flow stabilization was related to several phenom-
introduction of internals: upward in the center region ena associated with the insertion of internals, as the
and downward in the wall region. longitudinal internals tend to damp the undisturbed
● The liquid velocity was found to be higher in the tumbling motion of the rising large bubbles. These
column with internals as compared to the column internals impose a more linear rising motion and
without internals regardless of the gas velocity used. hence, a higher rising velocity for these large bubbles
● The inversion point (zero liquid flow) between the results in line of a corresponding increase in the liquid
positive and negative liquid velocities in the column phase axial velocity. In addition, the addition of inter-
is maintained at about the same dimensionless radius nals decreases the free relative CSA. The final result is
of ~0.7. a higher liquid velocity.

Figure 45 Effect of internals on the profile of the axial component of the liquid velocity.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
202 A.A. Youssef et al.: Bubble Columns with Internals

Figure 46 Effect of internals on the central liquid velocity profile [91].

Figure 47 Effect of column diameter and internals on the profile of the axial component of the liquid velocity [91].

Bernemann [91] implies that only the gas holdup increase Bernemann [91] agreed with Shah et al. [85] and
was due to the diminution of the open CSA, while the Kafarov [93] that the increase in the dispersion coefficient
central liquid velocity boost was due, mainly, to the was not due to the increase in the actual gas velocity
physical existence of the tubes in the column. These resulting from the decrease in the free (open) CSA but
conclusions need to be supported with fundamental was caused by the physical effect that the tubes them-
models. selves imposed on the flow and mixing behavior through
An increase in the effective dispersion coefficient was their effect on the bubbles rising motion, as described
observed as a result of the addition of vertical internals. above. For the internals configuration of the highest
This was found to apply to both cases, the small and the packing density (i.e., the smallest open CSA) in the
large diameter columns (Figure 48 (a and b)). small diameter column, the effective dispersion

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 203

Figure 48 Effect of the gas velocity and internals on the effective dispersion coefficient in (a) the small diameter column and (b) the large
diameter column [91].

coefficient was found to reach a plateau at lower gas The effect of a static mixer placed vertically in a draft
velocities because the small bubbles were moving fast tube of a slurry bubble column was investigated by
in twirls and the large bubbles were prevented from Gaspillo and Goto [94]. They reported its effect on pres-
forming (Figure 49). sure drop, on minimum gas velocities for complete sus-
It is unfortunate that the work accomplished by the pension of particles, and on rates of mass transfer from
Germans did not include extensive modeling studies gas to liquid and from liquid to solids. The presence of a
and was limited to experimental investigations that, draft tube was found to decrease the minimum gas velo-
although comprehensive and well structured, cannot city, while the static mixer was found to increase the
be generalized until fundamental models are minimum gas velocity needed for complete solid suspen-
developed. sion. The static mixer in the draft tube, shown in

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
204 A.A. Youssef et al.: Bubble Columns with Internals

Figure 49 Effect of the gas velocity and internals on the effective dispersion coefficient [91].

Saxena’s group at the University of Illinois – Chicago

carried out a series of studies for the heat transfer inves-
tigations in bubble columns in the late 1980s and early
1990s. In the course of their studies, which will be men-
tioned later in more detail, they used heat transfer
probes, the design of which is shown in Figure 51 (a, b,
and c) [95]. It is of interest to present these probes not
only as heat transfer coefficient measuring devices but
also as vertical internals immersed inside the column in
an annulus and also to point out their effect on the
various hydrodynamic parameters. These internals had
diameters up to 5.08 cm in a column as small as
10.8 cm in diameter.
It was concluded that the gas holdup does not
depend on the diameter of the probe, while the heat
transfer coefficient does depend on the hydraulic dia-
meter (decreases with probe diameter increase). The rea-
son is that liquid mixing decreases with annulus size
reduction but bubble coalescence remains almost unin-
fluenced, resulting in a constant gas holdup with probe
diameter change. The constant gas holdup observed by
the authors contradicts the findings of Yamashita [88],
who attributed the variations in gas holdup to the percent
open area and to the tube diameter, which are both
varied in this work. Especially when one examines the
Figure 50 Static mixer in a draft tube inside a slurry bubble column
largest probe (50.8 mm) which occupies ~22% of the total
[94].
CSA of the column, it is hard to understand the physical
basis behind a constant gas holdup. It is also noteworthy
Figure 50, was also effective in increasing the rate that in the same system, Saxena et al. [96] and Saxena
of mass transfer (when a single nozzle was used as and Chen [97] found an effect of these probes on the heat
gas distributor to decrease the pressure drop in the transfer coefficient (prominent with the 50.8 mm dia-
column). meter probe), which comes as a surprise since the heat

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 205

Figure 51 Design details of heat transfer probes used internally in the column [95].

transfer phenomena are closely correlated with the gas For three-phase systems (air–water–glass beads), the
holdup and the bubble characteristics [98, 99]. gas holdup was found to be higher for 37 tubes than for 7
Saxena et al. [100] investigated the effect of internal tubes. This finding indicated that gas holdup depends
tubes simulating the heat exchanger configurations with more upon the geometrical configuration of tube bundle
increasing sizes and configurations of 5, 7, and 37 stain- as the temperature increases than upon changes in the
less steel tubes of 19 mm diameter each in a 0.305 m system’s thermophysical properties that control the bub-
Pyrex glass column. These tube bundles blocked 1.9, 2.7, ble size. Unfortunately, the authors do not specify the
and 14.3% of the total column’s CSA, respectively. As reason for the gas holdup increase in the case of 37 tubes,
shown in Figure 52, the tubes were located in three con- whether because more CSA is blocked, which enhances
centric hexagonal rows around the column axis at radial the actual gas velocity or because different physical con-
distances of 36.5, 73, and 109.5 mm for the 37 tubes siderations result from bundle presence. It is, however,
arrangement. important that the tubes arrangement affects the gas
The authors reported an inhibition of the bubble holdup in a more prominent way than the system’s phy-
growth (coalescence) as the tube bundle becomes tighter sical properties (at high temperature).
for the internals configuration shown in Figure 52c as Saxena [101] investigated the characteristics of the
compared to the sparsely located tubes in Figure 52b. above configuration of tube bundle with 37 tubes in
Such inhibition is further confirmed with their results conjunction with systems of different physical properties
from the arrangement of 37 tubes shown in Figure 52a. proposing a novel heat exchanger design. The heat
The extent of decrease in bubble growth was even more exchanging bundle is shown in Figure 53. He concluded
pronounced as either the temperature or the superficial that a tube bundle heat exchanger configuration occupy-
gas velocity increases. ing the entire column cross-section is preferable based on

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
206 A.A. Youssef et al.: Bubble Columns with Internals

Figure 52 Internals geometrical configurations and the heat transfer probe locations (dark circles) [100].

an increase in the radial and vertical distance above the

gas distributor plate. In addition, the 37 tubes bundle
configuration (occupying only 14.3% of the total CSA)
does not unfavorably influence the hydrodynamic char-
acteristic of the slurry bubble column. In consequence,
the mass transfer rates and hence the chemical produc-
tion rates in the column remain uninfluenced by the
insertion of the internals inside the column. This latter
claim is, obviously, unclear and confusing since one
would expect the column’s performance to be affected
by the presence of the tubes. In earlier studies from the
Saxena group, the gas holdup and the bubble dynamics
are both altered as a result of introducing such internals
in the system which directly implies a corresponding
Figure 53 37 tube bundle [101]. change in the mass transfer rate and the productivity.
Thimmapuram et al. [102] studied the heat transfer
mass transfer and residence time considerations. Reliable from immersed tubes in a baffled slurry bubble column.
scale-up of heat exchanger size is claimed to be possible Their experimental data revealed that with tube bundles
by retaining the geometrical configuration and pitch and of different sizes (similar to the configurations shown in
is illustrated through measurements in both 0.108 and Figure 52), the heat transfer coefficient was sensitive to
0.305 m columns. Saxena summarized the study with the the bundle size and configuration. The reason for the
findings that the heat transfer coefficient decreases with variations lie in the bubble size and the mixing pattern

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 207

it generates. For the less dense tube bundles, the large recirculation were observed; as the superficial gas veloci-
bubbles rise through the central region of the column, ties increased further, a well-mixed and churned liquid
while for the 37 tubes case, bubbles rise through the with no cellular liquid circulation scheme, and with bub-
entire column cross-section more uniformly. bles always moving upward were the phenomena prevail-
They proposed, via a careful examination of their ing in the system. This was followed by the formation of
data, the following dimensionless relation for the estima- slugs in the column’s upper section as gas velocities
tion of the heat transfer coefficient: increased further. These slugs sizes depend upon the inter-
nals arrangement inside the column.
St ¼ CðReFrPr2:54 Þ0:25 The authors investigated hw’s dependence on the size
of the internals at different superficial N2 velocity (for
They improved on the correlation given earlier by
Nitrogen-Therminol and Nitrogen-50% Therminol–Red
Deckwer et al. [103], based on Higbie’s surface renewal
iron oxide systems). The results are summarized for
theory of interphase mass transfer and on Kolmogoroff’s
both cases in Figure 54.
theory of isotropic turbulence, that claimed
For a 7-tube bundle, as shown in Figure 55, which
St ¼ 0:1ðReFrPr2 Þ0:25 blocks about 22% of the column cross-section, the heat
transfer was found to be about the same as in a column
Deckwer’s correlation was proposed for use with super-
with a single tube of 50.8 mm blocking almost the same
ficial gas velocities up to 0.1 m/s, but Thimmapuram et al.
CSA. In other words, the measured valued of the heat
[102] examined it with a large number of data points,
transfer coefficients were almost the same when the same
comparing experimental and calculated values of St., to
relative occupied CSA was maintained regardless of the
find absolute average deviations of up to 28.9% when
configuration of the blocking internals being a single
using the latter correlation. Therefore, they proposed the
tube or a bundle.
former correlation to improve the data reproduction.
Saxena and Chen [105] analyzed the experimental
It is to be noted, however, that both correlations have
data from Saxena and Chen [97] along with the data
the same starting point, which is Kast’s [104] earlier
obtained from a 0.305 m bubble column equipped with
proposed correlation:
37 tubes of 19 mm diameter each, arranged in an equi-
St ¼ f ðReFrPr2 Þ: lateral triangular configuration, as shown in Figure 52a.
They reported that Deckwer’s (1980) theory for heat trans-
Saxena et al. [96] and Saxena and Chen [97] examined the fer coefficient estimation worked well for un-baffled col-
effect of a single tube diameter on the heat transfer coeffi- umns up to a gas velocity of 0.16 m/s, while for baffled
cient in a 0.108 m diameter column. The heat transfer columns, some modifications were needed. They assert
coefficient is smaller for a 50.8 mm tube than for 19 and that the thermal boundary layer theory has limited
31.8 mm tubes because in the latter cases, the bubble size applicability because of the lack of gas holdup values
is smaller than the annulus size, providing a relatively near the heat transfer surface.
superior liquid mixing. In the former study, small bubbles The effect of the volume fraction covered by internals
were found to coalesce only at the column’s top at low gas was examined by Pradhan et al. [41] who tested by two
velocities, while at higher velocities, more coalescence and types of internals (helical coils and vertical straight tube

Figure 54 Dependence of hw on the nature of internals for the Nitrogen-Therminol: (a) without and (b) with solids [97].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
208 A.A. Youssef et al.: Bubble Columns with Internals

In a partial agreement with Yamashita [88], Pradhan

et al. [41] highlighted the effect of internals (tubes and
helical coils) as means to enhance the gas velocity lead-
ing to a higher gas holdup. They, contrary to Yamashita,
claimed that bubbles can move rapidly in the radial
direction despite the vertical obstacles, without mention-
ing any change in behavior when the gap between the
internals becomes too small. It should be mentioned that
their system was an air–aqueous CMC solution instead of
an air–water, as in the former study by Yamashita. In a
later study by the same research group, De et al. [106]
carried out the investigations for gas holdup in a different
system with (a) aqueous sodium sulfate, (b) butanol, and
(c) glycerin as the liquid phase and with air as the gas
phase. Their conclusions do not add much to those
already published by Pradhan et al. [41] as far as the
Figure 55 7 tube-bundle arrangement. (1) heat transfer probe,
effect of internals is concerned.
(2) ring clamp, (3) spacer plates, (4) locating stud, (5) telescopic
Schlüter et al. [2] reported different configurations for
locating stud, (6) column wall, (7) Teflon rounded cap, (8) stainless
steel spring, (9) locking pin, (10) calrod heater, and (11) brass indirect heat transfer, including the ones involving inter-
tube [96]. nal structures (vertical and horizontal), which are shown
in Figure 56.
The authors correlated and investigated experimen-
bundles). Six different Xv values (covered volume frac- tally the heat transfer in two- and three-phase bubble
tions) ranging from 0.014 to 0.193 were studied, and their columns with internals. The effects of superficial gas
results showed that gas holdup increased with an velocity, tube pitch and reactor diameter on the heat
increase in Xv. Moreover, helical coils provided higher transfer coefficient is shown in Figures 57a and 57b in
gas holdup than vertical tubes, a finding they attributed columns of 0.19, 0.29 and 0.45 m in diameter for several
to the fact that inter-tube gaps allowed larger bubbles to tube arrangements. The effect of liquid phase viscosity
escape, with vertical tubes decreasing the gas holdup, was examined by using both water and propylene glycol.
while with helical coils, smaller gaps were available. They found the heat transfer coefficient to be higher
This work is one of a few studies where coils were utilized with of liquids of lower viscosity as compared to highly
as internals, and a comparison was made between their viscous liquids. The latter was found to be truly indepen-
effect and the effect of regular tube internals. dent of the tube arrangement utilized. They also con-
cluded that the heat transfer increases with column

Figure 56 Indirect heat transfer means in bubble column reactors [2].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 209

Figure 57a Influence of Ug, tube pitch, and Dr on the heat transfer coefficient [2].

Figure 57b Influence of Ug, tube pitch, and Dr on the heat transfer coefficient [2].

diameter only in case of low viscosity liquids. The heat simulated a methanol synthesis unit in La Porte, Texas.
transfer coefficient was found to be independent of the Their study comprised an air–water and air–drakeoil
dimensions of the column, which is equipped with ver- (Drakeoil® 10) systems in an 18ʺ diameter column. The
tical internals except in low viscosity liquids. geometry of the internals as shown in Figure 58 is com-
Chen et al. [107] took advantage of the CT and the posed of two concentric circular bundles of 1ʺ aluminum
Computer Automated Radioactive Particle Tracking tubes, each containing 8 tubes equally spaced.
(CARPT) facilities in CREL (Chemical Reaction Chen et al. [107] concluded that such internals, with
Engineering Laboratory) at Washington University in St. ~5% filled area based on the total column surface area,
Louis to investigate the effect of internals on gas holdup, have no significant effect on liquid recirculation velocity,
liquid recirculation, and turbulent parameters. They while gas holdup increased slightly due to the heat

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
210 A.A. Youssef et al.: Bubble Columns with Internals

Figure 58 Heat exchange tubes internals (5% of total CSA is cov-

ered) [106]. Figure 59 Schematic of multi-stage bubble column slurry reactor.

exchanger tubes. The turbulent stresses and eddy diffu- behavior for both the “dense” and the “dilute” phases.
sivities were lower in the presence of internals. These The conditions of an industrial unit were considered,
results were expected since the percentage covered CSA including 3 MPa pressure and 513 K maximum tempera-
by heat exchanging internals typically needed for the ture. The researchers simulated a system consisting of
methanol synthesis process (5%) cannot cause significant N2-turpentine-5 with a solid holdup of 35% operating in
changes in the flow. the churn-turbulent regime. They used a multi-stage
Our review of the pertinent research in chronological reactor model that assumed the dilute phase to be in
order reveals that fundamental modeling of bubble col- plug flow and the dense phase (pseudo homogeneous
umns was never tackled until the turn of the millennium liquid and catalyst) to be well-mixed. The reactions
(year 2000). A few empirical correlations were proposed were chosen by following the standard Langmuir–
by Aksel’rod et al. [52], by Korte [83] and by Hinshelwood type developed by Yates and Satterfield
Thimmapuram et al. [102], mostly for the heat transfer [109]. Their results indicated an increase in syngas con-
coefficient prediction, but these were, at best, applicable version with an increase in the number of stages. They
for the systems in which they were developed and exclu- also concluded that to obtain high conversions (>95%),
sive in nature. One has to admit that, with the various superficial gas velocities must not be lower than
arrangements of internals proposed by the researchers 20 cm/s. An important addition here is that as the num-
and with the absence of a unified methodology for ber of stages increases, the required area of the cooling
designing the tube bundle companies guard the informa- tubes increases as well. In their simulation, the area for
tion as a trade secret, it is difficult to introduce inclusive heat transfer was varying (decreasing) from the lower
correlations. In addition, if researchers have faced, and stage to the upper one. This is reminiscent of the patent
still are facing, uncertainties with empty column models of Kölbel and Langheim [77] and relates to the work by
and CFD simulation closures, the quest is more elusive Maretto and Piccolo [10]. The latter study investigated
for bubble columns equipped with internals. Much work the FT process in a bubble column without staging and
is needed in this regard. found that, even for the ideal case of assuming plug
Modeling bubble columns with internals, however, flow behavior for both phases, which would yield the
was attempted by Maretto and Krishna [61], by Forret highest conversion, a temperature profile forms inside
et al. [92] and by Larachi et al. [108]. the reactor, even though a heat exchanging tube bundle
Maretto and Krishna [61] developed a model for a was inserted. The solution to this non-isothermal unsafe
multi-stage bubble column slurry reactor for the FT pro- operating mode could not be reached by increasing the
cess. Staging in the column was achieved by introducing area of heat transfer and keeping a limiting temperature
sieve plates that also represented spacers for the cooling level inside the reactor since the conversion dropped
tubes, as shown in Figure 59 below. dramatically. Hence, Maretto and Piccolo developed a
The main idea of such design is to decrease the back multistage bubble column reactor model that assumed a
mixing of the liquid and to approach a plug flow more realistic completely mixed, liquid phase and a

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 211

plug flow behavior, for the gas phase and with model, noteworthy that the configuration of Forret’s internals
reached high conversions and productivity. To achieve was as follows: 56 tubes, each 63 mm in diameter
isothermal conditions along the column’s axis, varying arranged in a square pitch of 108 mm occluding ~22%
heat transfer areas are to be maintained, as shown in of the total area.
Table 9 below. Nosier and Mohamed [110] investigated the mass
transfer rates at a copper helical coil in a bubble column.
Table 9 The division of the specific heat exchange surface area per They used an air–acidified chromate solution system in a
unit volume among the various stages, aR, with a variation in the glass cylindrical experimental column of 7.9 cm diameter
number of stages [10]. and 50 cm in height. One longitudinal helical coil of
4.7 cm diameter, with the tube diameter of 0.3 cm and
Number of stages aR
the pitch of 0.84 cm, was used as the internal structure.
Ntot = 1 Ntot = 2 Ntot = 3 Ntot = 4 Seventeen rings per coil were used. Their setup is shown
in Figure 61. They concluded that the mass transfer coef-
I 1 0.642 0.437 0.328
ficient was found to decrease with an increase in the
II 0.358 0.378 0.31
III 0.185 0.249 number of active rings per coil. They correlated their
IV 0.113 experimental mass transfer data with the following
Total 1 1 1 1 equation:
245
J ¼ 0:778 ðFr:ReÞ0:263 ðdt =lw Þ0:
Forret et al. [92] used a basic tracer technique in a 1 m
diameter column for the experimental part of the work dt and lw being the coil tube diameter and the working
and observed a decrease in the liquid fluctuating velo- section length in cm, respectively, and J being the dimen-
city and an enhancement of large-scale liquid recircu- sionless mass transfer factor.
lation with the presence of internals, as shown in They also concluded that helical coils are more effi-
Figure 60. They successfully used the 1D ADM to deter- cient transfer surfaces than arrays of straight tubes. They
mine the dispersion coefficient in case of the column asserted that the presence of coils in the bubble column
without internals. However, as the large-scale recircu- diminishes the back mixing effect. Their results comple-
lation (which increased with internals) contributes to ment those reported by Pradhan et al. [41] as well as
liquid back mixing, the 1D ADM could not be used in those presented later by Balamurugan and Subbarao
case of columns with internals. A two-dimensional (2D) [111].
model developed by Forret et al. predicted well the Soraker et al. [112] patented a design, close to that
effect of internals on liquid mixing by accounting for proposed by Hall and Taylor [75], for a heat exchange
a radial dispersion coefficient Drad,2D, a radially depen- system for bubble columns consisting of a bundle of
dent axial velocity, and an axial dispersion coefficient tubes of a square or triangular pitch. A header at one
Dax,2D. The best fit was obtained when Dax,2D/Drad,2D end supplies (removes) the heat exchanging fluid and
was equal to 400. This contradicts Pradhan’s et al. two or more sub-headers are connected to the header
[41] claim that vigorous bubble motion is present in via distribution conduits. These sub-headers, connected
the radial direction despite the existence of internals, as well to the tubes, are located in at least two different
but agrees with Yamashita’s [88] findings. It is radial positions and in at least two different transverse

Figure 60 Enhancement of large recirculation scale and decrease of radial dispersion due to internals [92].

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
212 A.A. Youssef et al.: Bubble Columns with Internals

internals, as shown in Figure 63, in a 15 cm diameter

column using the Dynamic Gas Disengagement techni-
que. Twenty-one and forty-one stainless steel helical
springs of 0.5 mm wire formed in a 1.9 cm diameter coil
were found to yield an increase in the total bubble
holdup. The increase in the volume fraction of internals
resulted in a further increase in the gas holdup. A max-
imum 140% increase in gas holdup was obtained upon
inserting the internals, compared to the case of a column
with no internals at all. Also, the enhancement of the gas-
liquid interfacial area more than doubled for the bubble
column with helical springs internals.
A recent patent by Hawthorne et al. [113] discusses an
arrangement of cooling tubes for a FT slurry reactor that
the authors claim will simplify both the removal and the
reinstallation of the tubes for maintenance. In their work,
Hawthorne et al. [113] give an example of the configura-
tion shown in Figure 64 that consists of 576 tubes in a
reactor of 21.5 ft in diameter. The tubes are arranged in
bundles of four or twelve, as shown in Figure 63(a). Such
a setting is sought to ease both the movement of the
bundles in the horizontal direction via support rails and
Figure 61 Experimental setup [109]. also the removal of the damaged bundle upward through
a top removal nozzle.
Recently, to execute a numerical study of bubble
columns with internals, Larachi et al. [108] carried out
CFD simulations for the flow behavior in five different
cases, four of which contained internals arrangements
(star with wall clearance, star with core clearance,
dense, and sparse) that use a single superficial gas velo-
city of 12 cm/s. The occluded surface areas ranged
between 2 and 16.2%, and the tubes of 1ʺ diameter were
arranged in a triangular pitch configuration. They
assumed a constant bubble size (neglecting coales-
cence/dispersion effects) and a steady drag force as the
sole force (neglecting all other forces such as lift, wall,
and turbulent diffusion). In agreement with Forret et al.
Figure 62 Design of patented heat exchange tubes (headres and
[92], Larachi et al. [108] confirmed that internals both
subheaders) [111].
enhance the presence of small pockets rather than
large-scale gas bubbles and also suppress liquid circula-
tion within the inter-tube gaps.
directions (Figure 62). The investigators also showed tie The well-known core-annulus flow resulted in the
rods of different configurations stiffening the tubes simulation of the uniform internals configurations,
together at multiple axial positions. They claimed that which confirms the results highlighted by Bernemann
such staggered design minimizes the risk of horizontal [91]. However, for the non-uniform internals implanta-
CSA restriction (referring to back mixing, etc.). It is clear tion, a complex flow behavior was detected (Figure 65).
from the complexity of the drawing below that the design The simulation of Larachi et al. [108] disagrees with the
of the internal heat exchanger to be used in bubble experimental results of Chen et al. [107] regarding the
columns is not a straightforward job. location of the liquid circulation velocity inversion
Recently, Balamurugan and Subbarao [111] investi- point. On the one hand, Chen et al. [107] claimed that
gated the effect of vibrating helical springs’ type the location of the flow inversion point is closer to the

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 213

Figure 63 (a) Experimental setup used and (b) helical springs holder [110].

Figure 64 (a) Elevation view of a tube bundle; (b) Cross-sectional view of a tube bundle; and (c) Cross-sectional view of reactor [112].

column’s central axis when internals covering 5% of the discrepancy might be explained by the different
total CSA were inserted. On the other hand, Larachi’s arrangement of internals and system used in both
simulation findings do not support such a claim. This studies.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
214 A.A. Youssef et al.: Bubble Columns with Internals

Figure 65 CFD simulations for different internals configurations (red-upflow, blue-downflow) [107].

Forret et al. [114] studied the effect of internals on bubble in the understanding of the fluid dynamics phenomena in
column hydrodynamics at different scales. Their experi- bubble columns Today, turbulence parameters can be
ments were run in columns of diameters of 0.15, 0.4, and measured and the flow regime transition can be pin-
1 m, with and without internals. The internals had the pointed for various systems [6, 118].
exact same arrangement as in Forret et al. [92]. They Moreover, the extent of back-mixing of the phases
concluded that the root mean square (r.m.s.) fluctuation has been the subject of several investigations (e.g., [119],
velocity, although decreasing, has a maximum at the Han, 2007 [120]). Three phase systems have been utilized
inversion point (~r/R = 0.7). They also found that the by researchers to mimic applications where a solid phase
liquid recirculation was enhanced through the addition catalyst is typically employed (e.g., [16, 84]).
of internals due to an increase in the convective effect In addition to the above, the faster computers stimu-
and a decrease in the dispersion effect. lated further CFD research of the flow in bubble columns
Most recently, Youssef and Al-Dahhan [115] investi- (e.g., [18, 24, 28, 29]).
gated the effect of vertical internals covering 5 and 22% of However, all the above work has been, for the most
the total column’s CSA on the bubble dynamics and gas part, limited to columns with no internals.
holdup in an 8ʺ bubble column. They reported an Studies of bubble columns with internals focused on
increase in the gas holdup with the presence of internals. global hydrodynamics were reported by Yamashita [88],
They also found that, as 25% internals are inserted, Kawasaki [58], and Alvaré and Al-Dahhan [73]. The radial
smaller bubbles are present in the system yielding an distributions of gas holdup were obtained in the work of
increase in the gas-liquid interfacial area and a decrease Chen et al. [107] and Youssef and Al-Dahhan [115].
in the bubble chord length. On the other hand, slower The overall back-mixing data in bubble columns with
velocities were obtained for the bubbles. internals was reported by Mashelkar and Sharma [49],
Table 10 provides a summary of the above-mentioned Westerterp et al. [121], and Alvaré and Al-Dahhan [15].
studies on vertical internals However, in order to draw any general conclusions all
these studies need to be augmented by data sets based on
a variety of internals’ designs and configurations.
Moreover, the reported studies excluded 3 phase systems
5 Concluding remarks and
leaving the impact of internals on the solid phase mixing
recommendations an open question. Heat and mass transfer are essential
phenomena to understanding the behavior of bubble
As indicated in the first section of this review, enhance- columns with internals. They have been scrutinized by
ments in the measuring capabilities led to advancements Saxena and Patel [95] and Kawasaki [58], respectively. A

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 Table 10 Summary of research work on bubble columns with vertical internals: (a) literature and (b) patents.

Author(s) System used Column and sparger features Internals specifications Investigated subject and/or
parameters

a) Literature
Hall and Taylor [75] Water gas–hydrogen–mill- Stainless steel: 7 7/8ʺ diameter- 7′ 6ʺ height/ Cooling tube bundle: central tube of 2 3/8ʺ Design and operation of FT Synthesis
scale Fe2O3 sintered bronze (Porosint) disk 1/8ʺ thick, OD surrounded by a circular array of 8 tubes of pilot plant
12.5 μm pore size (effective area = 0.15 ft2) 1 1/16ʺ OD each. The array has a diameter 5 5/8ʺ
Fair et al. [45] Air–water Commercial scale 42ʺ diameter – 10′ height; 9ʺ 42 Aluminum tubes of 1.5ʺ diameter in 2 concentric Heat transfer
ring with 0.03ʺ orifices circles
Carleton et al. [80] Nitrogen–water and Details are given in Tables 4 and 5 in text Gas holdup, pressure drop, gas and
oxygen-cobalt catalyzed liquid RTD, and interfacial area
sodium sulfite
Voyer and Miller [81] Nitrogen–water and 5.5ʺ diameter-0.67 to 7.8 ft/ sieve plate 1/2ʺ 6 mesh cylindrical screen packing and 1/2ʺ 6 Interfacial area
Nitrogen–NaOH solution (5% free area, 0.15ʺ diameter holes at 1/2" mesh corrugated screen packing (corrugated and each
triangular pitch) layer perpendicular to the next one).
Mashelkar and Sharma [49] Nitrogen and CO2 (latter is Glass: 6.6 cm diameter and 120 cm height; Details are given in Table 6 Mass transfer
solute gas) sodium Perspex: 10 cm, and 20 cm diameter with 66
carbonate–bicarbonate cm, and 90 cm height/single half centimeter
buffer diameter tube
Aksel’rod et al. [52] Air–water and air– Plexiglas: 20
20 cm square column-height a) single tube, 12/1.5 mm; b) in-line tube bundle with Heat transfer
transformer oil not specified/grid plate distributor spacing of 20
20 mm; c) in-line tube bundle with
spacing of 40
20 mm; and d) staggered tube bundle
with spacing of 40
20 mm
Shah et al. [85] Nitrogen–water Glass: 6.35 cm diameter/sparger 6, 16 and 23 glass rods (0.004 m diameter); Gas holdup and back mixing
details not specified 2 and 4 glass rods (0.011 m diameter);

Authenticated | 10.248.254.158
1 glass rod (0.032 m diameter);

Download Date | 9/1/14 2:33 PM

1, 2, and 4 screen baskets (0.019 m diameter) and
1 screen basket (0.038 m diameter). The smaller
baskets were a) empty, b) filled with 0.0032 m
polyethylene packing and c) filled with 0.0159 m

Brought to you by | Aston University Library & Information

diameter glass rods.
The larger baskets were a) and b) same as above and
c) filled with 0.032 m diameter rod. (Rods were all
1.143 m tall)
Kölbel and Ralek [86] Syngas-catalyst slurry Not Specified Vertical honeycombed shafts with cooling pipes Liquid back mixing and catalyst
arranged centrally around or in corners efficiency
O’Dowd et al. [89] Nitrogen–water–glass 10.8 cm diameter column – 1.94 m height/ Internal baffles: 5 vertical rods (1 central and 4 around Solids dispersion coefficient, local gas
spheres perforated plate with 72 holes of 0.001 m at 90 deg. each) of 0.019 m diameter and 1.88 m holdup and, bubble size and interfacial
diameter height area
A.A. Youssef et al.: Bubble Columns with Internals

(continued )
215
 216

Table 10: (continued)

Author(s) System used Column and sparger features Internals specifications Investigated subject and/or
parameters

Yamashita [88] Air–water 31 cm diameter/single nozzle of 60 mm Single rod of 6 mm diameter (hanging 36.5 cm above Overall gas holdup
diameter downward on central axis (10 cm bottom)
above bottom) Multi rods and pipes:
Small separation *(6 mm):
18, 44, 70 and 85 internals of 14 mm diameter.
Large separation ( 8mm):
37, 28, 21 and 10 internals of 22 mm diameter.
9 internals of 60 mm diameter.
9 internals of 48 mm diameter.
16 cm diameter/single nozzle of 27.6 mm Single pipe and rod (sitting on bottom)
diameter horizontally on side wall (10 cm Multi rods and pipes:
above bottom) Large separation *(8mm):
2 internals of 14, 22, 38 and 20 mm diameter
5, 6, and 11 internals of 22 mm diameter
8 cm diameter/single nozzle of 10 mm Single pipe and rod
diameter horizontally on side wall (4.2 cm
A.A. Youssef et al.: Bubble Columns with Internals

above bottom)
Saxena et al. [116] Air–water–red iron oxide 10.8 cm diameter/perforated plate Solid cylinder of 19 mm diameter Heat transfer modeling
powder
Gaspillo and Goto [94] Nitrogen–water–resin 9.7 cm diameter-37 cm height/single nozzle of A static mixer in a draft tube. The mixer was Mass transfer
(Amberlyst 15) 1 mm diameter and a plastic ball with many successively composed of right-hand 90° angle
fine pres corrugated stainless steel sheets
Saxena and Patel [95] Air–water–glass beads Same as Saxena et al. [116] Axial cylindrical probes of 19, 31.8, and 50.8 mm Heat transfer and gas holdup

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
diameter along the central axis of the column
Saxena et al. [96] Nitrogen-Therminol 66–red Same as Saxena et al. [116] Single cylindrical probe (19, 31.8, and 50.8 mm Heat transfer, gas holdup
iron oxide powder diameter) and bundle of 7 tubes of 19 mm diameter
each in a triangular pitch of 36.5 mm

Brought to you by | Aston University Library & Information

Saxena et al. [100] Air–water 30.5 cm diameter – 3.3 m height/perforated 5, 7, and 37 Stainless Steel tubes (the latter in 3 Overall gas holdup
Air–water–glass beads plate of 0.8 mm diameter orifices in square bundles of 3 concentric hexagonal rows) of 19 mm
arrangement of 9.5 mm pitch diameter each and the pitch is 36.5 mm
Saxena and Rao [117] Nitrogen-Therminol– Same as Saxena et al. [100] 37 Stainless Steel tubes in a bundle of 19 mm diameter Overall gas holdup
magnetite each in equilateral pitch of 36.5 mm
Saxena [101] Air–water Same as Saxena et al. [100] 1, 5, 7 and 37 tube bundle (3.25 m in height), pitch as Heat transfer
Air–water–glass beads Saxena et al. [100]
Nitrogen-Therminol
Nitrogen-Therminol–
magnetite
Saxena and Chen [97] Air–water Same as Saxena et al. [116] 1 and 7 tubes bundle Heat transfer
Same as Saxena et al. [100]
Same as Saxena et al. [116]
 Air–water-solids (glass
beads (50 μm), magnetite
(37.5, 49, 58, 69, 90.5 μm),
red iron oxide powder
(1.02, 2.38 μm))
Nitrogen-Therminol Same as Saxena et al. [100] 37 tubes bundle, pitch as Saxena et al. [100]
Nitrogen-Therminol-Solids
(red iron oxide, 1.7 μm),
magnetite (28, 36.6, 37 μm)
Nitrogen-Therminol
Nitrogen-Therminol-Solids
(magnetite (65μm))
Pradhan et al. [41] Air–aqueous CMC solution 0.102 m diameter column – 2.5 m height/64 Helical coils (made of 6 mm Co tube) of 3.5 cm and/or Overall gas holdup and pressure drop
holes of 1.5 mm diameter each in a 1.2 cm 6.8 cm diameter in 2.5 cm pitch and bundles of vertical
triangular pitch straight tubes (Stainless Steel with 1.2, 1.5, and 2.0 cm
outer diameters)
Thimmapuram et al. [102] Air–water Same as Saxena et al. [100] 1, 5, 7, and 37 tubes bundles, pitch as Saxena et al. Heat transfer
Air–water–glass beads [100]
Saxena and Chen [105] Air–water–magnetite Same as Saxena et al. [116] 7 tubes bundle, pitch as Saxena et al. [100] Heat transfer
Air–water–red iron oxide
powder
Nitrogen-Therminol
magnetite
Nitrogen-Therminol–red
iron oxide powder
Air–water–glass beads Same as Saxena et al. [100] 37 tubes bundle, pitch as Saxena et al. [100]
Nitrogen-Therminol-glass
beads
Nitrogen-Therminol

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
Magnetite
Schlüter et al. [2] Air–water 0.29 m diameter – 4.27 m height/sieve tray Longitudinal tube bundle (each of 25 mm diameter) of Heat transfer
Air–Propylene glycol 40, 70, and 120 mm pitch
Chen et al. [107] Air–water 18ʺ (44 cm) diameter/301 holes of 0.77 mm 16 Aluminum tubes of 1ʺ diameter each in two bundles Gas holdup and its radial profile, liquid

Brought to you by | Aston University Library & Information

Air–Drakeoil diameter each on 14 concentric circular rings at at r/R = 0.39 and 0.61 recirculation velocity, turbulent
1.5 cm apart stresses and eddy diffusivities
De et al. [106] Air–sodium sulfate 0.05 m diameter column –2.5 m height/plate Helical coils of 3.5 cm diameter and straight tubes of Overall gas holdup
Air–butanol sparger 1.2 cm and 1.5 cm diameter
Air–glycerine
Maretto and Krishna [61] Syngas-paraffin C16H34–Co 7 m diameter –30 m dispersion height Vertical cooling tubes and spacer sieve trays Reactor productivity and reaction
based catalayst kinetics modeling
Forret et al. [92] Air–water 1 m diameter/perforated plate: 312 holes of 56 tubes of 63 mm diameter each and a 10.8 cm square Liquid mixing-axial dispersion
2 mm diameter and 50 mm pitch pitch coefficient
Nosier and Mohamed [110] Air–K2Cr207/H2So4 Glass: 7.9 cm diameter-50 cm height/G4 One helical coil of copper of ring, diameter 4.7 cm, Mass transfer
A.A. Youssef et al.: Bubble Columns with Internals

sintered glass distributor of 7 cm diameter and tube diameter of 0.3 cm, and of pitch 0.84 cm. 17 rings
0.5 cm thick. The average diameter of the pores in the coils.
of sintered distributor was 5–10 microns.
217

(continued )
 218

Table 10: (continued)

Author(s) System used Column and sparger features Internals specifications Investigated subject and/or
parameters

Larachi et al. [108] Air–water Simulated lab scale 19 cm diameter and pilot Tubes of 1ʺ diameter and triangular pitch in 4 CFD simulations (gas holdup, liquid
scale 100 cm diameter arrangements: dense (253 tubes), sparse (31 tubes), axial velocity and turbulent kinetic
star/wall clearance (121 tubes), star/core clearance energy)
(132 tubes)
Balamurugan and Gas (Air) – liquid (NA) 15 cm diameter/perforated plate with 126 holes 21 and 41 stainless steel helical springs of 1.9 cm coil Bubble size and holdup
Subbarao [100] of 0.2 cm diameter each in 1 cm square pitch diameter made of 0.5 mm wire
Youssef Air–water 0.19 m diameter –2 m height/perforated plate: 12 and 48 Plexiglas rods of 1/2ʺ diameter each located Gas holdup and its radial profile, and
and Al-Dahhan [115] xxx holes of yyy mm diameter each in zzz cm in two concentric circles and in triangular pitch, bubble dynamics
triangular pitch respectively

(b) Patents
Kölbel and Langheim [77] CO+H2-watery solution of 1.4 m diameter – 12 m height/NA 360 pipes that reduces to 270 at 3 m above the gas Improvement of cooling system design
(US 2,852,350) Fe(NO3)3 and Cu(NO3)2 inlet, to 180 after 3 further meters and finally to 90 by
moving 3 more meters higher
A.A. Youssef et al.: Bubble Columns with Internals

Kölbel and Ackermann [96] Gas-slurry (no details Applicable to any column with diameter 30 cm Vertical shafts (circular or hexagonal) with cooling Overcoming the liquid recirculation
(US 2,853,369) given) up to 3m and above, and more than 1.5 m in tubes within or in between the circular shafts or “rolling move ment” and back mixing
height various arrangements within the hexagonal ones
Hagino et al. [87] Seed culture of starting 80 cm diameter – 4m height Draft device (56 cm diameter – 3 m height) consisting Liquid agitation enhancement
(US 4,327,042) Corynebacterium of 4 vertical plates (50
3 m) with 1 cm spacing, each
glutamicum ATCC 21543 – of which has one edge in outwardly spaced
fermentation medium overlapping relation to a vertical edge of an adjacent

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
plate and has the other vertical edge in inwardly
spaced overlapping relation to a vertical edge of the
other adjacent plate
Koros 1995 [122] CO+H2– hydrocarbon wax 5.76ʺ ID – 50′ height/0.5ʺ nozzle Elongated, vertical, parallel tubes (here, reaction Reaction kinetics (rate and conversion)

Brought to you by | Aston University Library & Information

(US 5,384,336) (Co catalyst) zones) in a shell housing surrounded by outer cooling
jacket
Soraker et al. [112] CO+H2– slurry (Co catalyst) Not specified Parallel tubes bundle with a header at one end to Internals design (minimizing cross
(WO 2005/065813) which 2 or more subheaders are connected and also to sectional blockage)
the tubes (of triangular, square or rotated square pitch)
Hawthorne et al. [113] FT system 21.5′ diameter column 576 cooling tubes of 4ʺ diameter each in bundles of 12 Cooling tube arrangement facilitating
(US 7,108,835) or 4 tubes each in 9 rows the removal of plurality of cooling
tubes

Notes: *Separation = distance between holes on the support plate – the outer diameter of pips and rods.
 A.A. Youssef et al.: Bubble Columns with Internals 219

single widely accepted theory or correlation did not The gap size between internals is important in the
emerge and further work on both fronts is still a neces- longitudinal funneling of liquid flow; it is also responsi-
sity. Finally, the modeling of these systems and making ble for the decrease of the liquid kinetic turbulent energy
use of CFD enhanced tools remains important in seeking [107]. The large-scale liquid recirculation increases in the
“a priori” prediction of the transport parameters of presence of vertical internals (tubes) covering large CSA
interest. (~22%) [113], while less covered CSA (5%) does not affect
Below we summarize some of the notable reports and liquid recirculation much [106]. A 2D model fitted well
conclusions reached by researchers dealing with bubble the large columns without and with vertical (tubes) inter-
columns with internals. nals [92].
Sectionalizing bubble columns via perforated trays The heat transfer increases with addition of vertical
was reported to lead to an increase in overall gas holdup tubes [52] but so does the axial dispersion coefficient [85]
[45, 58, 62, 73]. However, a similar increase in gas holdup making the temperature more uniform throughout the
was observed when utilizing horizontal tubes instead column.
[63]. Blass and Cornelius [54] reported a decrease in This review summarized the studies performed in
bubble coalescence upon addition of horizontal sectiona- bubble columns equipped with internal structures, and
lizing plates, which could explain increased gas holdup it is clear that much remains to be done. A thorough
but Kemoun et al. [62] claimed that the horizontal trays investigation of the causes for the discrepancy in many
induce bubble coalescence, may create gas pockets reported results was not undertaken as part of this review
below trays, and present an obstacle to the uniform and may be reported at a later time when more detailed
bubbles. A more fundamental analysis of the conditions information becomes available. The data on columns
that yield these seemingly opposite results needs to be with internals are few in comparison to the columns
performed. with no internals. Thus the phenomenological models
Regarding overall liquid back mixing, there are many for columns with internals are still in their infancy com-
reports [15, 49, 60, 67, 120] that horizontal internals in pared to those developed for (empty) bubble columns
bubble columns tend to reduce the overall back mixing with no internals.
so that the benefits of reactor operation as an ideal Further development is needed to bring rigor and
cascade of perfectly mixed tanks in series can be understanding to the performance of bubble columns
approached. However, some researchers [51, 54] found with internals. The studies involving bubble columns
that with an increase in the plate open free area, the must include the effect of internal structures on the var-
liquid backflow ratio increases so that the whole bubble ious parameters affecting the reactor’s performance. More
column with trays approaches a single mixed stirred collaboration between industry and academic institutions
vessel behavior. While a decrease in the open free area should be developed to generate the rational basis for
reduces the liquid recirculation this comes at the expense design of columns with internals which are dominant in
of an increase in the pressure drop [60, 71]. commercial practice.
The overall mass transfer coefficient was found to
increase upon sectionalization with perforated trays [58]
and to decrease upon an increase in the number of hor-
izontal rows of tubes [67]. Unfortunately, the observed
6 Nomenclature
pressure drops were no reported in the two systems and
Af Relative free cross-sectional area
more fundamental analysis for this behavior has not been
C Coefficient (avg. 0.16 over the bubble-bed
performed. volume)
On immersion of vertical internals, the bubbles were d Probe diameter
found to break into smaller ones [41, 100]. However, D Column diameter (cm)
Dl Liquid phase axial dispersion coefficient
O‘Dowd et al. [89] reported an increase in the bubble
(m2/s)
size upon insertion of the baffles. Longitudinal tube dR Diameter of the rod/internal (m)
internals tend to decrease the area available for flow DR Diameter of the column (m)
resulting in an increase in the gas holdup values when dt Tube diameter (m)
compared to columns with no internals [88, 106, 111]. The FP New flow parameter (=VG*%Free area/Vl)
Fr Froude number
same applies when helical coils were used by Pradhan FrG Froude number of the gas phase
et al. [41]. Ga Gallileo number

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
220 A.A. Youssef et al.: Bubble Columns with Internals

Hc Column height (cm) Sc Schmidt number

Hn Height of the overflow above the weir (m) Sh Sherwood number
hw Heat transfer coefficient (W/m2.K) StD Stanton diffusion number
J Mass transfer factor (St.Sc0.66) tR Tube pitch
Kl Liquid phase mass transfer VG Gas phase superficial velocity
coefficient (ft/s) vG eff Gas velocity based on the available free
kLa Volumetric mass transfer coefficient (s–1) cross-sectional area
L Liquid height in the column (m) vGo Gas velocity based on the total cross-
lch Characteristic linear dimension of the sectional area of the column
bubbling bed (= ratio of the volume of clear Vl Liquid phase superficial velocity (m/s)
liquid held on the tray by a counterflowing vLo Liquid phase velocity based on the total
air to the area of wetted surface in the cross-sectional area of the column
bubbling zone) Wg Gas (air) velocity (m/s)
lw Working section length (cm) WGD Mean superficial gas velocity
N Number of equal, homogeneously mixed WLd Orifice liquid velocity of the plates
stages Wliq Liquid rate of flow over weir (m3/m.h)
n number of data points ΔPC Frictional pressure drop due to liquid
Neff Number of effective stages circulation (N/m2)
nR number of rods/internals εg Overall gas holdup
Nu Nusselt number ηL Dynamic viscosity of the liquid (Pa.)
Pe Peclet number ηLw Dynamic viscosity of the liquid in the wall
PrD Prandtl diffusion number region
PrL Prandtl number of the liquid phase κ Backflow ratio defined as u'L/uL where
Re Reynolds number the nominator is the superficial velocity of
ReL Reynolds number based on liquid velocity the liquid which flows through plate and
Reliq Reynolds number based on liquid velocity the denominator is the net value of
ReS Reynolds number based on slip velocity superficial liquid velocity
S Tube pitch (m) φ Cross-sectional area of holes

References
1. Duduković MP. Opaque multiphase reactors: experimenta- 12. Behkish A, Lemoine R, Sehabiague L, Oukaci R, Morsi BI.
tion, modeling and troubleshooting. Oil Gas Sci Technol Gas holdup and bubble size behavior in a large-scale slurry
2000;55:135–58. bubble column reactor operating with an organic liquid
2. Schlüter S, Steiff A, Weinspach, P-M. Heat transfer in two- under elevated pressures and temperatures. Chem Eng J
and three-phase bubble column reactors with internals. 2007;128:69–84.
Chem Eng Process 1995;34:157–72. 13. Duduković MP. Relevance of multiphase reaction engineer-
3. Duduković MP, Larachi F, Mills PL. Multiphase reactors – ing to modern technological challenges. Ind Eng Chem Res
revised. Chem Eng Sci 1999;54:1975–95. 2007;46:8674–86.
4. Shetty SA, Kantak MV, Kelkar BG. Gas-phase backmixing in 14. Krishna R, Ellenberger J. Improving gas–liquid contacting in
bubble – column reactors. AIChE J 1992;38:1013–26. bubble columns by vibration excitement. Int J Multiphase
5. Krishna R, Ellenberger J. Gas hold-up in bubble column Flow 2002;28:1223–34.
reactors operating in the churn-turbulent flow regime. Am 15. Alvaré J, Al-Dahhan MH. Liquid phase mixing in trayed bub-
Inst Chem Eng J 1996;42:2627–34. ble column reactors. Chem Eng Sci 2006;61:1819–35.
6. Shaikh A, Al-Dahhan MH. A review on flow regime transition 16. Rados N. Slurry bubble column hydrodynamics: experimen-
in bubble columns. Int J Chem Reac Eng 2007;5. tation and modeling. DSc Thesis, Washington University,
7. Amstel JP, Rietema K. Wet air oxidation of sewage sludge: Saint Louis, MO, 2003.
part II-the oxidation of real sludges. Chem Ing Tech 17. Svendsen HF, Jakobsen HA, Torvik R. Local flow structures in
1973;45:1205–11. internal loop and bubble column reactors. Chem Eng Sci
8. Deckwer WD. Bubble column reactors. Chichester: Wiley, 1992. 1992;47:3297–304.
9. Steynberg A, Dry M. Fischer-Tropsch technology. 18. Ranade VV. Flow in bubble columns: some numerical experi-
Amsterdam: Elsevier, 2004. ments. Chem Eng Sci 1992;4:1857–9.
10. Maretto C, Piccolo V. Fischer-Tropsch process with a multi- 19. Grienberger J, Hofmann H. Investigations and modelling of
stage bubble column reactor, US Patent 5,827,902, 1998. bubble columns. Chem Eng Sci 1992;47:2215–20.
11. Krishna R, Sie ST. Design and scale-up of the Fischer– 20. Sokolichin A, Eigenberger G. Gas–liquid flow in bubble
Tropsch bubble column slurry reactor. Fuel Process columns and loop reactors: part I detailed modelling and
Technol 2000;64:73–105. numerical simulation. Chem Eng Sci 1994;49:5735–46.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 221

21. Lapin A, Lübbert A. Numerical simulation of the dynamics of 39. Berg S, Schlüter S, Weinspach P-M. Ruckvermischung in
two-phase gas–liquid flows in bubble columns. Chem Eng Blasensaulen mit Einbauten. Chem Ing Tech 1995;67:289.
Sci 1994;49:3661–74. 40. Millies M, Mewes D. Calculation of circulating flows in bub-
22. Delnoij E, Kuipers JAM, van Swaaij WPM. Computational ble columns. Chem Eng Sci 1995;50:2093–106.
fluid dynamics applied to gas-liquid contactors. Chem Eng 41. Pradhan AK, Parichia RK, De P. Gas hold-up in non
Sci, 1997;52:3623–38. Newtonian solutions in a bubble column with internals.
23. Pan Y, Dudukovic MP, Chang M. Dynamic simulation of Can J Chem Eng 1993;71:468–71.
bubbly flow in bubble columns. Chem Eng Sci 42. Steiff A, Weinspach P-M. Heat transfer in stirred and non-
1999;54:2481–89. stirred gas liquid reactors. Ger Chem Eng 1978;150:150–161.
24. Sanyal J, Vásquez S, Roy S, Dudukovic MP. Numerical simu- 43. Dyer P. Catalyst and reactor development for a liquid phase
lation of gas–liquid dynamics in cylindrical bubble column Fischer-Tropsch process (Air Products and Chemicals Report
reactors. Chem Eng Sci 1999;54:5071–83. to DOE-Final report Task 4), 1989.
25. Krishna R, van Baten JM. Rise characteristics of gas bubbles 44. Davis B. Fischer–Tropsch synthesis: overview of reactor
in a 2D rectangular column: VOF simulations vs. experi- development and future potentialities. Top Cat
ments. Int Commun Heat Mass Transf 1999;26:965–74. 2005;32:143–68.
26. Olmos E. Etude expérimentale et numérique des 45. Fair JR, Lambright AJ, Anderson JW. Heat transfer and gas
écoulements gaz-liquide en colonnes à bulles. Thèse de holdup in a sparged contactor. Ind Eng Chem Process Des
doctorat, Institut National Polytechnique de Lorraine, Dev 1962;1:33–36.
Nancy, France, 2002. 46. Schügerl K. Development of bioreaction engineering. Adv
27. Lapin A, Paaschen T, Junghans K, Lübbert A. Bubble column Biochem Eng Biotechnol 2000;70:41–76.
fluid dynamics, flow structures in slender columns with 47. Fair JR. Trends in distillation technology. Ind Eng Chem
large-diameter ring-spargers. Chem Eng Sci 1962;54:53–57.
2002;57:1419–24. 48. Fair JR. Developments in distillation technology. Ind Eng
28. Chen P, Sanyal J, Dudukovic MP. Numerical simulation of Chem 1964;56:61–64.
bubble columns flows: effect of different breakup and coa- 49. Mashelkar RA, Sharma, MM. Mass transfer in bubble and
lescence closures. Chem Eng Sci 2005;60:1085–101. packed bubble columns. Trans Inst Chem Eng1970;48:162–72.
29. Chen P, Gupta P, Dudukovic MP, Toseland BA. 50. Khoze AN, Burdukov AP, Nakoryakov VE, Pokusaev BG,
Hydrodynamics of slurry bubble column during dimethyl Kuz’min, VA. Convective heat transfer in a dynamic two-
ether (DME) synthesis: gas–liquid recirculation model phase bed. J Eng Phys 1971;20:759–76.
and radioactive tracer studies. Chem Eng Sci 51. Sekizawa T, Kubota H. Liquid mixing in multistage bubble
2006;61:6553–70. columns. J Chem Eng Jpn 1974;7:441–6.
30. Gupta P, Ong B, Al-Dahhan MH, Dudukovic MP, Toseland 52. Aksel’rod LS. Vorotnikova NI, Kozlov AA. Heat transfer and
BA. Hydrodynamics of churn turbulent bubble columns: gas- several aspects of hydrodynamics of bubble beds on sieve
liquid recirculation and mechanistic modeling. Cat Today trays equipped with tube bundles. Heat Transf – Sov Res
2001;64:253–69. 1976;8:25–33.
31. Degaleesan S, Dudukovic MP, Pan Y. Experimental study of 53. Vorotnikova NI, Aksel’rod LS. Heat transfer during the trans-
gas induced liquid-flow structures in bubble columns. AIChE verse flow of a bubbling stream around pipes and pipe
J 2001;47:1913–31. bundles. Trudy Moskovskogo Instituta Khimicheskogo
32. Wild G, Poncin S, Li H, Olmos E. Some aspects of the Mashinostroeniya 1975;57:101–108.
hydrodynamics of bubble columns. Int J Chem Reac Eng 54. Blass E, Cornelius W. The residence time distribution of
2003;1. solid and liquid in multistage bubble columns in the
33. Yang GQ, Du B, Fan LS. Bubble formation and dynamics in cocurrent flow of gas, liquid and suspended solid. Int J
gas–liquid–solid fluidization – a review. Chem Eng Sci Multiphase Flow 1977;3:459–69.
2007;62:2–27. 55. Chen BH, Yang NS, McMillan AF. Gas holdup and pressure
34. Mudde RF. Gravity-driven bubbly flows. Ann Rev Fluid Mech drop for air-water flow through plate bubble columns. Can J
2003;37:393–423. Chem Eng 1986;64:387–92.
35. Diaz ME, Iranzo A, Cuadra D, Barbero R, Montes FJ, Galan 56. Karr AE. Performance of a reciprocating-plate extraction
MA. Numerical simulation of the gas-liquid flow in a column. AIChE J 1959;5:446–52.
laboratory scale bubble column: influence of bubble size 57. Chen BH, Yang NS. Characteristics of a cocurrent multistage
distribution and non-drag forces. Chem Eng J bubble column. Ind Eng Chem Res 1989;28:1405–10.
2008;139:363–79. 58. Kawasaki H, Hirano H, Tanaka H. Effects of multiple draft
36. Jakobsen HA, Lindborg H, Dorao CA. Modeling of bubble tubes with perforated plates on gas holdup and volumetric
column reactors: progress and limitations. Ind Eng Chem mass transfer coefficient in a bubble column. J Chem Eng
Res 2005;44:5107–151. Jpn 1994;27:669–70.
37. Devanathan N, Moslemian D, Dudukovic MP. Flow mapping 59. Al Taweel AM, Ramadan AM, Moharam MR, El Mofty SM,
in bubble columns using CARPT. Chem Eng Sci Ityokumbul MT. Effect of honeycomb inserts on axial mixing
1990;45:2285–91. in bubble columns. Chem Eng Res Des 1996;74a:456–62.
38. Kumar SB, Devanathan N, Moslemian D, Dudukovic MP. 60. Palaskar SN, De JK, Pandit AB. Liquid phase RTD studies in
Effect of scale on liquid recirculation in bubble columns. sectionalized bubble column. Chem Eng Technol
Chem Eng Sci 1994;49:5637–52. 2000;23:61–9.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
222 A.A. Youssef et al.: Bubble Columns with Internals

61. Maretto C, Krishna R. Design and optimisation of a multi- columns with large diameter. Chemie Ingenieur Technik
stage bubble column slurry reactor for Fischer-Tropsch 1978;50:811.
synthesis. Cat Today 2001;66:241–48. 83. Korte H. Heat transfer in bubble columns with and without
62. Kemoun A, Rados N, Li F, Al-Dahhan MH, Dudukovic MP, internals. PhD Thesis, University of Dortmund, 1987.
Mills PL, et al. Gas holdup in a trayed cold-flow bubble 84. Wu C. Heat transfer and bubble dynamics in slurry bubble
column. Chem Eng Sci 2001;56:1197–205. columns for Fischer-Tropsch clean alternative energy. DSc
63. Colmenares A, Sevilla M, Goncalves JJ, Gonzalez- Thesis, Washington University, Saint Louis, MO, 2007.
Mendizabal D. Fluid-dynamic experimental study in a bub- 85. Shah YT, Ratway CA, Mcilvried HG. Back-mixing character-
ble column with internals. Int Commun Heat Mass Transf istics of a bubble column with vertically suspended tubes.
2001;28:389–98. Trans Ins Chem Eng 1978;56:107–12.
64. Akita K, Yoshida F. Gas holdup and volumetric mass transfer 86. Kölbel H, Ralek M. The Fischer-Tropsch synthesis in the
coefficient in bubble columns. Ind Eng Chem Process Des liquid phase. Cat Rev, Sci Eng 1980;21:225–74.
Dev 1973;12:76–80. 87. Hagino H, Odagiri H, Okutani J. US Patent 4,327,042 (Dec.
65. Schumpe A, Deckwer WD. Gas holdups, specific interfacial 22, 1980).
areas, and mass transfer coefficients of aerated carboxy- 88. Yamashita F. Effects of vertical pipe and rod internals on
methyl cellulose solutions in a bubble column. Ind Eng gas holdup in bubble columns. J Chem Eng Jpn
Chem Process Des Develop [87];21:706–11. 1987;20:204–206.
66. Dreher AJ, Krishna R. Liquid-phase backmixing in bubble 89. O’Dowd W, Smith DN, Ruether JA, Saxena SC. Gas and
columns, structured by introduction of partition plates. solids behavior in a baffled and unbaffled slurry bubble
Catalysis Today 2001;69:165–70. column. AIChE J 1987;33:1959–1970.
67. Nosier SA. Solid-liquid mass transfer at gas sparged tube 90. Wasan DT, Ahliwalia MS, Consecutive film and surface
bundles. Chem Eng Technol 2003;26:1151–54. renewal mechanism for heat and mass transfer from a
68. Mochizuki S, Matsui T. Liquid-solid mass transfer rate in wall. Chem Eng Sci 1969;24:1535–42.
liquid-gas upward concurrent flow in packed beds. Chem 91. Bernemann K. On the hydrodynamics and mixing of the
Eng Sci 1974;29:1328–30. liquid phase in bubble columns with longitudinal tube
69. Nosier SA, El-Kayar A, Farag HA, Sedahmed GH. Solid-liquid bundles. PhD Thesis, University of Dortmund, 1989.
mass transfer at gas sparged fixed bed of Raschig rings. Int 92. Forret A, Schweitzer JM, Gauthier T, Krishna R, Schweich D.
Commun Heat Mass Transf 1997;24:733–40. Liquid dispersion in large diameter bubble columns, with
70. Cavatorta ON, Bohm U. Heat and mass transfer in gas spar- and without internals. Can J Chem Eng 2003;81:360–66.
ging systems: empirical correlations and theoretical mod- 93. Kafarov VV, Kruglik AE, Trofimov VI. Comparative evaluation
els. Chem Eng Res Des 1988;66a:265–74. of the effect of installation of some standard heat exchan-
71. Doshi YK, Pandit AB. Effect of internals and sparger design gers in bubble-type columns on the average gas content
on mixing behavior in sectionalized bubble column. Chem and structure of liquid-phase streams. Zhurnal Prikladnoi
Eng J (Amsterdam, Netherlands) 2005;112:117–29. Khimii 1975;48:229–32.
72. Pandit AB, Doshi YK. Mixing time studies in bubble column 94. Gaspillo PD, Goto S. Mass transfer in bubble slurry column
reactor with and without internals. Int J Chem Reac Eng with static mixer in draft tube. J Chem Eng Jpn 1991;24:680–
2005;3. 2.
73. Alvaré J, Al-Dahhan MH. Gas holdup in trayed bubble col- 95. Saxena SC, Patel BB. Heat transfer investigations in a bub-
umn reactors. Ind Eng Chem Res 2006;45:3320–26. ble column with immersed probes of different diameters. Int
74. Mecaial N, Sadik B. Hydrodynamic and RTD of Sectionalized Commun Heat and ass Transf 1991;18:467–78.
Bubble Column. Proceedings of the 12th International 96. Saxena SC, Rao NS, Yousuf M. Hydrodynamic and heat
Conference on Fluidization – New Horizons in Fluidization transfer investigations conducted in a bubble column with
Engineering, 2007. fine powders and a viscous liquid. Powder Technol
75. Hall CC, Taylor AH. Design and operation of a fluid catalyst 1991;67:265–75.
pilot plant for Fischer-Tropsch synthesis. J Inst Petrol 97. Saxena SC, Chen ZD. Heat transfer in baffled bubble col-
1955;41:101–24. umns of dilute slurries of fine powders and viscous liquids.
76. Kölbel H, Ackermann P. US Patent 2,853,369 (Sept. 23, Exp Heat Transf, Fluid Mech Thermodyn 1993. Proc. World
1958). Conf., 1993; 3rd, 2:1451–8.
77. Kölbel H, Langheim R. US Patent 2,852,350 (Sept. 16, 1958). 98. Li H, Prakash A. Heat transfer and hydrodynamics in a three-
78. Hofmann H. Packed upflow bubble columns. Chemie phase slurry bubble column. Ind Eng Chem Res
Ingenieur Technik 1982;54:865–76. 1997;36:4688–94.
79. Gestrich W, Harth H. Liquid-phase backmixing in packed 99. Luo X, Lee DJ, Lau R, Yang GQ, Fan L-S. Maximum stable
bubble columns. Chemie Ingenieur Technik 1981;53:308–14. bubble size and gas hold up in high-pressure slurry bubble
80. Carleton AJ, Flain RJ, Rennie J, Valentin HH. Some properties column. AIChE J 1999;45:665–80.
of a packed bubble column. Chem Eng Sci 1967;22:1839– 100. Saxena SC, Rao NS, Thimmapuram PR. Gas phase holdup in
45. slurry bubble columns for two- and three-phase systems.
81. Voyer RD, Miller AI. Improved gas-liquid contacting in co- Chem Eng J (Amsterdam, Netherlands) 1992;49:151–9.
current flow. Can J Chem Eng 1968;46:335–41. 101. Saxena SC. A novel heat exchanger design for slurry bubble
82. Magnussen P, Shumacher V, Rotermund GW, Hafnef F. columns. Transp Phenom Therm Eng Proc Int Symp 1993;
Residence time behavior of the liquid phase in bubble 6th:896–901.

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
 A.A. Youssef et al.: Bubble Columns with Internals 223

102. Thimmapuram PR, Rao NS, Saxena SC. Heat transfer from 112. Soraker P, Lian P, Vankan S. WO Patent 2005/065813 A1
immersed tubes in a baffled slurry bubble column. Chem (July 21 2005).
Eng Commun 1993;120:27–43. 113. Hawthorne WH, Ibsen MD, Pedersen PS, Bohn MS. US
103. Deckwer, WD, Louisi, Y, Zaidi, A, and Ralek, M. Patent 7,108,835 B2 (Sep. 19, 2006).
“Hydrodynamics Properties of the Fischer-Tropsch 114. Forret A, Schweitzer JM, Gauthier T, Krishna R, Schweich D.
Slurry Process” Ind. Eng. Cem. Process Des. Dev. Scale up of slurry bubble reactors. Oil Gas Sci Technol
1980;19:699–708 . 2006;61:443–58.
104. Kast W. Analyse des wärmeübergangs in blasensäulen. Int J 115. Youssef AA, Al-Dahhan MH. Impact of internals on the gas
Heat Mass Transf 1962;5:329–36. holdup and bubble properties of a bubble column. Ind Eng
105. Saxena SC, Chen ZD. Hydrodynamics and heat transfer of Chem Res 2009;48:8007–13.
baffled and unbaffled slurry bubble columns. Rev Chem Eng 116. Saxena SC, Rao NS, Saxena AC. Estimation of heat transfer
1994;10:193–400. coefficient for immersed surfaces in bubble columns invol-
106. De SK, Ghosh S, Parichha RK, De P. Gas hold-up in two ving fine powders. Powder Technol 1990;63:197–202.
phase system with internals. Indian Chem Eng Section A 117. Saxena SC, Rao NS. Estimation of gas holdup in a slurry
1999;41:T54–8. bubble column with internals: nitrogen–therminol–magne-
107. Chen J, Li F, Degaleesan S, Gupta P, Al-Dahhan MH, tite system. Powder Technol 1993;75:153–8.
Dudukovic MP, Toseland BA. Fluid dynamic parameters in 118. Degaleesan S. Turbulence and liquid mixing in bubble col-
bubble columns with internals. Chem Eng Sci umns. PhD Thesis, Washington University, Saint Louis,
1999;54:2187–97. Missouri, USA, 1997.
108. Larachi F, Desvigne D, Donnat L, Schweich D. Simulating the 119. Ong B. Experimental investigation of bubble column hydro-
effects of liquid circulation in bubble columns with inter- dynamics – effect of elevated pressure and superficial gas
nals. Chem Eng Sci 2006;61:4195–206. velocity. PhD Thesis, Washington University, Saint Louis,
109. Yates IC, Satterfield CN. Intrinsic kinetics of the Fischer- USA, 2003.
Tropsch synthesis on a cobalt catalyst. Energy Fuels 120. Han, L. Hydrodynamics, back-mixing, and mass transfer in a
1991;5:168–73. slurry bubble column reactor for Fischer-Tropsch alternative
110. Nosier SA, Mohamed MM. Mass transfer at helical coils in fuels, D. Sc. dissertation, Washington University in
bubble columns. Chem Biochem Eng Q 2004;18:235–39. St. Louis, 2007.
111. Balamurugan V, Subbaro D. Bubble size and holdup in 121. Westerterp KR, van Swaaij WPM, Beenackers AACM. Chemical
bubble columns with vibrating internals. AIChE Spring reactor design and operation. Chichester: Wiley, 1987.
National Meeting, 2006. 122. Koros RM, Westfield NJ US Patent 5,384,336 (Jan. 24, 1995).

Brought to you by | Aston University Library & Information

Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM
Brought to you by | Aston University Library & Information
Authenticated | 10.248.254.158
Download Date | 9/1/14 2:33 PM