Gas Holdup and Overall Volumetric
Oxygen Transfer Coefficient in Airlift
Contactors
R. Ade Bello,’ Campbell W. Robinson, and Murray Moo-Young
Biochemical Engineering Group, Department of Chemical Engineering,
University of Waterloo, Waterloo, Ontario, Canada N2L 3G 1
Accepted for publication August 10, 1984
The two major types of airlift contactors, concentric-tube
and external-loop, were investigated for their gas holdup
(riser and downcomer) and overall mass transfer char-
acteristics. Results obtained in batch charges of tap water
and 0.15 kmol/m3 NaCl solution are reported for external-
loop airlift contactors having downcomer-to-riser cross-
sectional area ratios, A d / A , , ranging from 0.11 < A d / A ,
< 0.69 and for concentric-tube types in the range 0.13
< Ad/A, 4 0.56.
The results are compared and contrasted for the two
types of airlift contactors and also with corresponding
results obtained in a bubble column ( A d / A r = 0). Both
the riser gas holdup, ( E ~ ) , , and the overall volumetric
mass transfer coefficient, (K,aDIi,were found to increase
with increasing gas superficial velocity and the corre-
sponding specific gassing power input, ( P G / V D ) T . How-
ever, at a specified power input both and (KLaDIT
decreasedwith increasingAd/A,due to a resultant increase
in circulating liquid velocity. Unifying generalized cor-
relations for both types of airlift contactors are given for
) ~ (KLaD)pThese correlations are expected
both ( E ~ and
to be applicable as well to continuous-liquid flow bubble
columns at comparable liquid velocities.
INTRODUCTION
Airlift contactors are a special class of pneumatic
contactors which currently are receiving much attention
for potential application to various fermentation sys-
tems. Their self-generated liquid circulation has been
shown to give them added advantages, for example,
improved heat transfer and mixing, compared to the
bubble columns.’x2 However, most of the earlier
work^^-^ normally were carried out in only one size
o r type of airlift contactor and also with only one or
two of the several operating or design parameters being
investigated. As a result, only an incomplete under-
standing of their operative characteristics has emerged
to date. Some reported results in various types of
airlift contactors for the riser gas holdup, ( E ~ ) and
~ ,
the overall volumetric mass transfer coefficient,
are shown in Figures 1 and 2, respectively.
* Present Address: Department of Chemical Engineering, Uni-
versity of Lagos, Lagos, Nigeria.
Biotechnology and Bioengineering, Vol. XXVII, Pp. 369-381 (1985)
0 1985 John Wiley & Sons, Inc.
These figures also feature some literature results ob-
tained in bubble columns. Some of the discrepancies
in Figures 1 and 2 may arise from differences between
the liquid phase physicochemical properties and the
types of gas spargers used in the various test systems.
However, from the wide spread of these results, it is
obvious that more detailed information on the intrinsic
properties of airlift contactors is required to account
for the Iarge variation in the reported results and for
the differences in their characteristics in relation to
the conventional bubble column.
In this work the volume-average gas holdup in the
riser and the downcomer and the overall volumetric
mass transfer coefficient were measured in four external-
loop airlift contactors (fixed riser and three inter-
changeable downcomers) with the downcomer-to-riser
cross-sectional area ratio, A d / & , ranging from 0.11
to 0.69 (see Table 111) and in three concentric-tube
airlift contactors with A d / A , of 0.13, 0.35, and 0.56
(fixed outer column and three interchangeable internal
concentric cylinders). These results are compared and
correlated with the reported results in the literature
and with results obtained in a bubble column (A,/A,
= 0) for both tap water (“coalescing” system) and
0.15 kmol/m3 NaCl solutions (bubble-coalescence-in-
hibiting system).
EXPERIMENTAL
Apparatus and Materials
The airlift contactors (AL) used in this work are as
shown in Figures 3(a) and 3(b) with dimensions given
in Table XII. These were constructed of Plexiglas (acrylic
resin). Experimentation was carried out in batch
changes of tap water and an electrolyte solution (0.I5
kmol/m3 NaCl). Waterloo tap water is drawn from
deep wells and has quite consistent properties. Typical
concentrations are: total dissolved solids, 325 mg/L
(range 240-400 mg/L); total hardness, 320 mg/L; “M”
alkalinity, 260 mg/L; chlorides, 32 mg/L; pH 7.4. Air
CCC 0006-3592/85/030369-
13$04.00
 0.50

I
1 1 I 1 I l i l l I I I I l l 1 1 1
0.5 1 5 10 50 100

UG or (uG)r (m/s x 1 0 2 )
Figure 1. Gas holdup in bubble columns ( and in the nser of airlift contactors ( E ~ ) , .See Table I for legend.
E ~ )

or nitrogen was sparged through a perforated plate
with fifty-two 1.02-mm holes (triangular-pitch) in the
case of the external-loop AL contactors and the bubble
column [Fig. 3(a)], while a ring sparger, containing
fifteen 1.02-mm holes, was used with the concentric-
.""
tube A L contactors [Fig. 3(b)]. In the latter contactors
air was spared in the annulus which then was the riser
section.
For the external loop AL contactor studies, with
the exception of the A , / A , = 0.69 case, only one of
the two attached downcomers was in service at any
one time. The internal cylinders of the various con-

50f / /
centric-tube airlifts were mounted 10 cm above the
bottom head plate in all cases.
All runs were made at local barometric pressure
and at room temperature (15-18°C).

Methods
Gas Holdup
The average volumetric gas holdup was measured
separately in the riser and downcomer of each contactor.
This was done by means of hydrostatic differential
pressure measurement along the vertical axis of each
section of the contactor. Two static pressure probes
were located 1.0 rn apart in the middle two-thirds of
each column; these were connected to a U-tube ma-
nometer filled with Meridian Indicating fluid (D-8325)
( p , = 1750 kg/m3).
0.3
0.05 01 0.5 1 5 10 Calculation of the gas holdup was done using equation
( p0/vL lT ( kW/ ma)
(l), which also has been used by previous workers.'
Prn - PL h z
Figure 2. Overall volumetric oxygen mass transfer coefficients in EG = ~-
airlift contactors and bubble columns. See Table I1 for legend. PL Ah

370 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 27, MARCH 1985

Table I. Legend to the literature gas hold-up data in Figure 1.

Contactor D , D, Superficial velocities Sparger

No. Reference type" (m) AdIA (mls) Liquid tY Pe

1 Yoshida and BC 0.152 - U , < 0.30 Water Single hole

Akita' 0.301 u, = 0
2 Kastanek et al.' BC 0.15 - U , < 0.30 Water Perforated
0.30 u, = 0 plate
3 Deckwer et al.* BC 0.20 - u, < 0.15 Water Cross of
u, < 0.10 nozzles
BC 0.15 - u, < 0.15 Water Porous plate
u, < 0.10
4 Hills9 BC 0.15 - 0.4 c U , s 3.0 Water Single hole
U , = 1.78
5 Chakravarty CT 0.045 3.94 (U,), < 0.20 Water Perforated
et a1.I' 0.059 1.87 plate
0.074 0.83
6 El-Gabbani" CT 0.084 0.29 0.021 c (UG),s 0.092 Water Ring sparger
0.066 s (U,), c 0.097
7 Hatch' CT 0.206 1.12 0.05 < (U,), < 0.30 Fermentation Ring sparger
0.30 < (UL),< 0.80 Medium
8 Onken and EL 0.10 0.25 0.005 s (UG)r< 0.12 Water Porous plate
Weiland' 0.10 < (U,), < 0.40
9 Orazem et al.', sc 0.152 1.o 0.07 s (U,), c 0.32 Na,SO, Single hole
0.16 < (UL)r< 0.22 Solution

a BC, bubble column; CT, concentric-tube airlift contactor; EL, external-loop airlift contactor; SC, split cylindrical airlift contactor.

Table 11. Legend to the literature (K,a,), data in Figure 2.

Contactor DT or D, HD(Ho)
No. Reference typea &/A (m) (4 Liquid Sparger type

1 El-Gabbani" CT 0.29 0.095 1.65 Water Ring sparger

(1.47)
2 Fukuda CT 0.38- 1.67 0.20 1.82 Na,SO, Perforated
et a1.I' (1 .00) Solution plates
3 GasnerI4 RS 1.o 0.20 x 0.30 1.22 Na2S0, U-shaped
(0.71) Solution
4 Botton CT 1.14 0.19 1.30 Na,SO, Concentric-ring
et al." Solution
CT 0.09 0.48 1.50 Na2S03 3 Concentric
Solution rings
5 Hatch' CT 1.12 0.30 2Xb Fermentation Ring
(2.6) Medium
6 Lin et al.' EL 0.11 0.15 3.0 Fermentation Perforated pipe
(2.87) Medium
7 Orazem and sc 1.o 0.15 1.22 Na,SO, Single hole
Erickson" (1.07) Solution
sc 1.o 0.15 0.66 Na2S03 Single hole
(two (0.51) Solution
stage) each stage
8 Sinclair and CT 3.0 0.15 0.23 Water Concentric-ring
Ryder'' (0.18)
9 Schiigerl BC - 0.14 4 Water Perforated
et a1.l' plate
BC - 0.14 4 Water Porous plate
10 Deckwer BC - 0.20 7.23 Water Cross of
et a1.' nozzles
BC - 0.15 4.4 Water Porous plate
11 Kastanek l9 BC - 0.10 0.6-1.2 Water Perforated
0.30 plate
12 Yoshida and BC - 0.15 0.9-1.82 Water Single hole
Akita'
13 Onken and EL 0.25 0.10 8.5 Water Porous plate
Weiland' (8.5)

a BC, bubble column; EL, external-loop airlift contactor; CT, concentric-tube airlift contactor; RS, rectangular split airlift contactor;

SC, split cylindrical airlift contactor.

Clear liquid height.

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 371

Table 111. Contactor geometric specifications." both absorption and desorption modes at various lo-
cations in the risers and downcomers of the airlift
contactors showed negligible axial variations and,
hence, indicated an essentially well mixed liquid phase.
External-Loop Airlift and Bubble Column Contactors Therefore, the volumetric mass transfer coefficient
0.152 0 0 BC
0.152 0.051 0.11 EL
based on the total contactor liquid volume, (&aL),,
0.152 0.076 0.25 EL was calculated using a model that assumes a well-
0.152 0.102 0.44 EL mixed liquid phase and a plug flow gas phase."
0.152 0.076 + 0.102 0.69 EL
Concentric-Tube Airliff Contactors
0.152 0.051 0.13
0.152 0.076 0.35
The volumetric mass transfer coefficient, based on
0.152 0.089 0.56
the total dispersion volume (KLaD),,was then cal-
* HD = 1.80 m; Ho = 1.55 m. culated using
BC, bubble column; EL, external-loop.
(3)

Overall Volumetric Mass Transfer Coefficient, The pneumatic power of the gas input was taken as
KLaD that due to isothermal expansion of the gas, according
The transient dissolved oxygen tension profiles were to
measured along the central axes of the columns, using
a polarographic dissolved oxygen probe (Yellow Springs
Inst., Yellow Springs, OH, model YSI 5739), which
was held in place by means of a pulley and wire mech-
anism.20The probe was connected to a Yellow Springs
OT
dissolved oxygen meter (model YSI 57) with the output
response simultaneously recorded on a Brinkman re- GAS OUT
corder and via a digital voltmeter on a digital printer.
Measurements of the transient oxygen tension in t t t

r
HO
BALL-
VALVE

o o0o ~ o0:

0
0

0
0

j
D

0
0

O
0

0: 1 O 0

0
0

0
0

.
,"
'0

airlift contactor.

372
I
GAS IN

Figure 3. (a) External-loop airlift contactor. (b) Concentric-tube

b

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 27, MARCH 1985

f AGAS IN
RESULTS AND DISCUSSION
Riser Gas Holdup
External-Loop AL and Bubble Column
The typical pattern of the riser gas holdup in the
external-loop type of AL and the bubble column is
shown in Figure 4. The value of ( E ~ ) in, each contactor
increased with increasing specific gassing power input,
( P G / V D ) ?However,
. a significant influence of the geo-
metrical parameter, A d / A r , on ( E ~ also
) ~ is evident.
As the downcomer diameter increased relative to the
riser diameter, that is, as &/Ar was increased, the
riser gas holdup decreased by up to 25-3Wo at constant
(PG/VD)T for A , / A , increasing from 0 to 0.69.
The effect of A , / A , on ( E ~ ) ,in the external-loop
AL contactors can be attributed mainly to the observed
effect of &/A, on the circulating liquid velocity. The
term A,/A, has been shown to be the physical parameter
which most affects the circulating liquid velocity in
airlift contactors.’”’ There is a substantial increase in
liquid velocity as A,/A, is increased. For example,’
with reference to Figure 4, at ( P G / V D ) T = 188 W/m3
[corresponding to (U,),= 2.16 cm/s and 3.46 cm/s

0.3
0.2
t-
for &/A, of 0.11 and 0.69, respectively], as A,/A,
was increased from 0.11 to 0.69, the circulating liquid
linear velocity, (VL), = ( U L ) , / ( l - E ~ ) , , increased
from 8.5 to 38.8 cm/s.
As the liquid velocity increased for the same gas
flowrate, the residence time of the bubbles in the riser
0.1 - decreased and, hence, the gas holdup decreased. This
-
ww
0.07-
direct influence of liquid velocity on the riser gas holdup
is evident from Figure 5. As the gas holdup approaches
Ar the volumetric feed ratio of the gas and the liquid,
0.05- 0 0 69 that is, as the liquid velocity increases, at a constant
0.04 - 0
13
044
0.25 ( U G ) r ,the gas holdup tends to approach a constant
a 0.11 value and the airlift contactor seems then to exhibit
0.03 - 8 0.00
the properties of a “transport reactor.” The decrease
in ( E ~ ) caused
, by increasing ( UL)roccurs irrespective
I I I I I I I I
of regulation of the liquid velocity. As shown in Figure
5 , data obtained by reducing the liquid velocity by
throttling the liquid flow in the contactors agree well
Figure 4. Riser gas holdup in bubble column and external-loop with the trend, as also was observed by Onken and
airlift contactors (water). Weiland.’

1 -
Ad
Ar
0.69
0.20 h P 0.44
0.25
0.11
0.00
0.44 (T) *
0 . 2 5 (TI *

0.10 c 0

0 10 20 30 40

RISER SUPERFICIAL LIQUID VELOCITY [ ( u ~ ) ~ (]x lo2 m/s)

Figure 5. Variation of riser gas holdup in external-loop airlift contactors with riser superficial liquid
velocity. 0.15 kmol/m’ NaCl solution. *Valve in lower horizontal connecting section throttled to 45”
angle.

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 373

 It may be deduced from the (UL),-(zG),relationship
that a batch-liquid bubble column [(V,), = 0, A , / A ,
= 01, operated at specified (UG),,would give a larger
limiting gas holdup than the AL. This is verified by
the BC data shown in Figures 4 and 5. However, the
gas holdup achieved in continuous-liquid flow BC op-
erations may be greater than in an AL contactor be-
cause, as shown by Hill9 and Bottom et a1.,21high
liquid flow reduces the gas holdup in bubble columns.
In general, somewhat larger gas holdups (10-15%)
were obtained with the 0. 15-kmol/m3 NaCl solution
than with water, in agreement with previous works87'8s22
on the effect of ionic strength of the liquid phase on
gas holdup.
Concentric-Tube AL Contactors
The riser gas holdup results obtained in the three
concentric-tube AL contactors ( A d / A , = 0.13, 0.35,
and 0.56) are shown in Figure 6. Unlike the case with
the external-loop AL contactors (Fig. 4), the riser gas
holdup does not show any significant variation with
A d / & at a constant specific gassing power input.
However, this does not negate the observed effect of
A,/A, on circulating liquid velocity and its concurrent
effect on ( E ~ ) as
, discussed for the external-loop case.
Due to the geometric setup of the concentric-tube AL
contactors used in this work (Table I), for a constant
gas volumetric flowrate [i.e., constant ( P G / V D ) T ] , as
A , / A , increased, the cross-sectional area of the riser,
A , , decreased. Therefore, the riser superficial gas ve-
locity increased, as opposed to ( UG),being constant
for the external-loop case. This creates an opposing
influence on ( E ~ ) ~Simultaneous
. increases in A,/A,
and (VG),result in a higher liquid circulation rate2
which, as shown in Figure 5 , in turn causes a net
decrease in ( E ~ ) ~At. the same time, however, the
increase in (UG),effects an increase in ( E ~ ) , . " Thus,
at a constant gassing power input the tendency for an
A 0.13
0.35
increase in (zG), due to the increase in (VG)rseems
to be offset by a decrease in ( E ~ ) , caused
. by the increase
in the liquid circulation velocity. Thus, there is no
apparent effect of varying A,/A, in concentric-tube
AL contactors, at least for the range of parameters
tested in this work.
Riser Gas Holdup Correlation
The main parameters that influence the riser gas
holdup in airlift contactors must be taken into account
in order to achieve a meaningful correlation of the
results obtained in the two types of AL contactors.
These parameters include the riser superficial gas ve-
locity, (UG),,and the riser superficial liquid velocity,
( U L ) , ,as well as A d / & . Other possible parameters
are the liquid viscosity, the surface tension, and the
ionic concentrations. These latter parameters have not
been studied in this work and hence are not considered.
A dimensionless form of correlation, involving the
slip ratio, (VG/U,),.and (1 + A,/A,), is used to describe
the variations of ( E ~ )in~ these contactors. The use
of the slip ratio to predict gas holdup is prominent in
two-phase flow The results are shown plotted
in Figure 7. The results obtained in water and 0.15
kmol/m3 NaCl solution are well represented by equa-
tions (5) and (6), respectively, with respective cor-
relation coefficients of 0.95 and 0.96.
Both the water and salt solution data can be represented
by equation (7) with a correlation coefficient of 0.96,
0 4 -
I - KAWAGOE AND ROBINSON (231 E L
2 - ONKEN AND WEILAND ( 5 ) E L
03- 3-HATCH(3lCT
4 -
DRIFT - F L U X MODEL
5 -
EL -GABBANI (Ill C T
6 -
EO ( 4 1
7 -
EO ( 5 1
- 8 -
HILLS ( 9 1
UL'0 4 m / r

Q 0.00
-E O . ( 4 . 1 0 ) ('G)r
0 1

Ad
0 07
0 069 EL
0 044 EL
D 025 EL
0 05
A 0 11 EL
A 0 13 CT
8 0 35 CT
0 56 CT
0.05 0 03 I
70 loo 200 300 500 700 loo0 0 2 03 0 4 0 5 07 10

+2 )
(3°.56(
(PJV,,)~ (w/m3)
1
Figure 6. Riser gas holdup in bubble column and concentric-tube
airlift contactors (water). Figure 7. Correlation of riser gas holdup in airlift contactors (water).

374 BIOTECHNOLOGY A N D BIOENGINEERING, VOL. 27, MARCH 1985

since there is no significant difference between the Figure 9. While the downcomer gas holdup in the
results at this level of ionic concentration. external-loop AL contactors was much smaller

(EG)r = 0.16($):( 1+ ?) (7)

(0-50%) than the riser gas holdup, in the concentric-
tube type, at the same gas flowrate, (&G)d was 80-95%
of the riser gas holdup. This observation is seen to
There is good agreement between the data of this work be consistent with all the available literature data, as
and those of other worker^^.'^ particularly for gas holdup shown in Figure 9.
values higher than 6%. There is up to 30% disparity This big difference in the riser versus downcomer
at lower gas holdups. In Figure 7 (line 4) the drift flux gas holdup characteristics of the two types of AL
model of Zuber and Finlay,’6 was used for 5 mm di- contactors can be explained in terms of their geometry
ameter bubbles in air-water systems, which is the and other parameters that may influence the downcomer
typical mean bubble diameter reported by Kawagoe gas holdup - i.e., the riser gas holdup; the liquid
and Robinson” for an external-loop airlift contactor velocity in the downcomer, ( u L ) d ; the size(s) of the
operated at equivalent conditions. The data of Hills’ bubbles in the downcomer, d B ;and the geometry of
obtained in a continuous-liquid flow bubble column the section just preceding the downcomer.
operated at relatively high liquid velocities are rep- External-loop AL contactors have interconnecting
resented by line 8 in Figure 7. sections (normally horizontal) between the riser and
the downcomer. Thus, even if bubbly flow conditions
exist in the riser, very much higher gas and liquid
DOWNCOMER GAS HOLDUP IN AL velocities are required to keep the flow bubbly in the
CONTACTORS horizontal section.z4727 For the conditions at the scale
In Figure 8 the downcomer gas holdup is presented of operation used in this work, stratified and plug flow
for both the concentric-tube and the external-loop AL conditions existed, whereby the bubbles coalesced to
contactors. Comparing the absolute values of (&G)d to form plugs of gas and liquid along the upper part of
the values of the riser gas holdup, ( E ~ ) , , shown in the connecting section. Therefore, only a small per-
Figures 4 and 6, distinct differences in the patterns of centage of the discrete bubbles from the riser were
variation of ( E ~ with
) ~ ( E ~ are
) ~ seen to exist in the available for transportation into the downcomer.
two types of airlift contactors. These are even more Nonetheless, a large fraction of these discrete bubbles
clearly evident in the plot of ( E ~ against
) ~ ( E ~ in
) ~ were carried down through the downcomer, due to
the higher liquid velocity there than in the riser (constant
circulating liquid volumetric rate with downcomer di-
ameter smaller than the riser’s). The downcomer su-
-
Ad perficial liquid velocity in the external-loop airlift con-
04 - Ar tactors was at least 0.50 m/s.” This is equivalent to
03-
0 069 EL the terminal rise velocity of 0.025 m diameter bubbles,
which is much bigger than that of the 0.005-0.006 m
diameter bubbles reported by Kawagoe and Robinsonz5
to be prevalent in this type of contactor at equivalent
operating conditions. Hence, for the conditions of this
work, ( E ~ was) ~ found to be a function only of (+),
(the latter controls that potential supply of bubbles to
the downcomer). Figure 9 shows that this function is
linear; it can be represented by
=
( E G ) ~ 0.79(~,), - 0.057 (8)
In the concentric-tube airlift contactor, however,
there is no interconnecting section between the riser
and the downcomer and as such, more of the discrete
bubbles at the top of the riser are available for trans-
portation into the downcomer. Furthermore, the
downcomer liquid velocity (20-30 cm/s) was only about
one-half or less than that in the external-loop airlift
contactor. The residence time of the bubbles and hence
the downcomer gas holdup then were relatively higher
70 lo0 200 300 400500 7oQ lo00 in the concentric-tube type L0.07 d ( E ~ d) 0.231
~ than
( PG/VD)T ( W / m 3 )
in the external-loop contactors EO.01 d ( E ~ 6) 0.111.
~
Chakravarty et al.” reported coalescence and redis-
Figure 8. Downcorner gas holdup in airlift contactors (water). persion of the bubbles in the downcomer of concentric-

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 375

 O.I6 - #I EL-GAEEANI ( 111 WATER CT
EL-GAEBANI ( Ill SALT CT
f KAWAGOE AND ROBINSON (23) EL
+ ORAZEM ET AL (12)
0.'4- @ HATCH ( 3)
e CHAKRAVARTY ET AL (10) CT

0.12 -
0.10-

( %)d
0.08 -

0.06 -
WATER SALT $
0 0.69 E-L
0 0.44 E-L
0.25 E-L
A 0.31 E-L
A A 0.13 CT
8 m 0.35 CT
0 0.56 CT

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24

(&h
Figure 9. Relationship between riser and downcomer gas holdups.

tube airlift contactors, leading to the formation of flowrates to bubbly-slug and slug flows at higher gas
smaller sized bubbles (than in the riser) with lower flowrates.
terminal rise velocity. Thus, despite the relatively lower In both types of A L contactors there is a consistent
liquid velocity, more bubbles are transported in the pattern of variation of (K,u,), with changing A d / A , .
downcomer. A direct linear relationship was also ob- As the downcomer diameter was increased relative to
tained between ( & G ) d and ( E ~ in) concentric-tube
~ airlift that of the riser, that is, as &/A, was increased,
contactors of this and similar works in the literature (K,u,), decreased. This relationship previously has
(Fig. 9), as expressed by
(EG)d = O-WEG)~ (9)
Experimental results in larger sizes of airlift contactors 0.07 -
-
would be needed to confirm the applicability of equa- 0.05 -
tions (8) and (9) to an industrial scale contactor.
-
--
'n
0.04

0.03-
VOLUMETRIC MASS TRANSFER COEFFICIENT 1

The results for the overall volumetric oxygen mass -+

P 0.02-
transfer coefficient, ( K L ~ D ) inT , water are given in
-
-f
Y
Figures 10 and 11 for the external-loop and the con- -
A,
centric-tube AL contactors, respectively. The value 0.01 -- 0 069
0 044
of (KLu,), was found to increase with increasing gassing - 0 0 0 25
power input per unit total volume, ( P G / V D ) Tin, each 0.007 - A 0.11
0.00
of the AL contactors, as well as in the bubble column
( A d / A , = 0). At relatively high values of (PG/V,)j-, 00 0 5
0.5 0.7 1.0 2 3 4 5 6 7 8 9
the rate of increase in (K,u,), with increasing
( P G / V D ) =is less and the curves tend to flatten. This
( pC /v, bT ( ~ / m 3x 1 6 ' )

is indicative of the change in the two-phase flow pattern Figure 10. Overall volumetric mass transfer coefficient, ( & , u ~ ) ~ ,
in these contactors from bubbly flow at the lower gas in the bubble column and the external-loop airlift contactors (water).

376 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 27, MARCH 1985

 8 The fundamental reason for the observed variation
7 - of &a,), with A , / A , can be identified by (I) the
6 - effect of the variation of the liquid velocity with
A,/A, and (2) the relative amount of mass transfer in
5-
the downcomer (i.e., the effect of the downcomer
4 -
volume).
The effect of circulating liquid velocity on (K,a,),
can best be appreciated by separately examining its
effects on KL (=kL) and for oxygen transfer from
air. Blenke29found K L to be only slightly dependent
on fluid dynamics; the well-known correlations of
~ ~ KLto be determined
Calderbank and M o o - Y o ~ n gshow
-
Ad
mainly by the bubble diameter, classified as large
Ar (>2 mm) or small (c0.5 mm). Hence, the liquid velocity
*A
iJ does not appreciably affect KL. For spherical bubbles,
000
043
8 0 35
0 56 a, depends on the gas holdup and the bubble diameter,
I I I 1 I 1 I I 1 " l l
according to
1
t 2 3 4 5 6 7 8
a, = -6%
(10)
(pG/v0 lT i w / m 3 r TO-') dl3
Figure 11. Overall volumetric mass transfer coefficient, For an external-loop AL contactor, Kawagoe and
in bubble column and concentric-tube airlift contactors (water). Robinsonz observed no significant change in the Sauter
mean bubble diameter with changing liquid velocity
not been obvious in the literature, mainly because for 0.02 s (V,), S 0.07 m/s. As shown in Figure 4,
most previous workers evaluated (KLa,), in an AL the riser gas holdup, however, decreases as the liquid
of fixed A,/A,. Only Hatch3 showed any experimental velocity increases. Therefore, with decreasing holdup
variation of (K,a,), with geometry. The results of this but with relatively constant bubble size, the specific
study, therefore, provides a means of optimizing the interfacial area and, hence, KLaDwill decrease with
relative diameters of the downcomer and the riser. increasing liquid velocity. As shown in Figure 12, up

0.001 I I I I I I I
0 10 20 30 I

RISER SUPERFICIAL LIOUID VELOCITY [(uL ( x lo2 m / s

(0. IS kmol/rn'
Figure 12. Effect of circulating liquid velocity on volumetric mass transfer coefficient, (KLuD)T
NaCI). *Data obtained with valve in the lower horizontal pipe throttled half-closed.

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 377

to a 60% decrease in ( K , U ~was ) ~ found to occur when
the riser superficial liquid velocity increased from 0
to 0.40 m/s at a constant gas flowrate. The decrease
in ( K L u D ) , as A d / A ,increases therefore can be attributed
to the increase of liquid velocity with increasing
Ad / A , .
The effect of the downcomer volume on the mass
transfer rate can be accounted for by considering the
effective mass transfer rate for the entire contactor as
being the sum of the rates in the component sections,
as expressed by
( K L a D ) T ( v D > T = (K.LaD)r(v D ) r
(1 1)
+ (KLaD ) d ( V D ) d
If it is assumed that mass transfer in the downcomer
is negligible, i.e., there is negligible bubble slip velocity,
then equation (1 1) reduces to
For an AL having virtually the same riser and down-
comer dispersion heights, equation (12) can be written
as
(13)
The values of (K,aD),calculated according to equation
(12) from the measured ( K , u ~ )are ~ plotted against
(PG/VDITin Figure 13 for the same data in Figure 10
and 11. The nondependency of ( K L u D ) , with A d / A ,
shows that the assumption of negligible mass transfer
in the downcomer is valid. This implies that as A,/A,
increases, thereby relatively increasing the downcomer
volume, the effective volume in which mass transfer
occurs decreases, and hence the overall volumetric
mass transfer coefficient based on the total dispersion
volume decreases. Negligible mass transfer in the
downcomer has also been reported by El-Gabbani."
This observation of negligible mass transfer in the
1
to
3 cations. The high liquid circulation rate, which is re-
0.00 EL
0.11 EL
2 0 0 . 2 5 EL
0 0.44 EL
0 0 . 6 9 EL
A 0.13 C T
-
2
Y
0 . 3 5 CT
0.56 CT
( PC/VOlT ( w/m3x 10-21
Figure 13. Volumetric mass transfer coefficient in the riser
(&ad, (water).
378 BIOTECHNOLOGY A N D BIOENGINEERING, VOL. 27, MARCH 1985
downcomer might seem surprising, particularly for the
concentric-tube airlift contactors which had an (&G)d
as much as 80-95% of ( E ~ ) However,
~ . since most of
the bubbles in the downcomer are being carried down
by the liquid, they experience only a negligibly small
slip velocity; therefore, promotion of mass transfer
also would be very small. Hence, it is conceivable
that most of the mass transfer would take place in the
riser of the airlift contactor, particularly in the case
of the relatively short contactors used in this work.
Bubble Column Versus Airlift Contactors
Examination of Figures 10-12 reveals that the
(K,u,), values obtained in the bubble column of this
work ( A d / A , = 0) are higher than the corresponding
values obtained in the AL contactors. The previous
discussions on the effects of the circulating liquid ve-
locity and the downcomer volume on (K,u,), support
this finding and further suggests that batch-liquid bubble
columns would give the highest &a,), for a specified
(VG>,.This contradicts the conclusions of some pre-
vious investigator^^,"^'^ of AL columns in which the
results were compared to those obtained in bubble
columns with different operating conditions, or sparger
type and/or mode of operation. The results of Onken
and Weiland' and the interfacial area measurements
of Kawagoe and R o b i n ~ o n , 'however,
~ agree with the
results of this work.
The effect of continuous liquid flow, steady-state
operation on the characteristics of both bubble column
and AL contactors was not examined in this study,
and the above conclusion with respect to a batch-
liquid BC may not be valid for the former case. Most
previous BC workers, who used low superficial liquid
velocities (V, S 0.04 m/s), have showed no effect of
liquid throughout velocity on K,aD. The recent works
of Botton et al.''.'' and Hills' done at high gas flows
and earlier works by Shulman and Molstad3' and Vinter
et aL3' found a pronounced influence of liquid flowrate
on both the gas holdup and K L u D , for relatively short
columns ( ~ 1 . m).
0 Further work is required in order
to be able to generalize or to limit this finding for all
modes of operation.
The higher (K,u,), in the batch-operated bubble
column does not necessarily give it an overriding ad-
vantage compared to the AL contactor for all appli-
ported to promote better mixing* and heat transfer, if
accompanied by a (K,a,), which meets the transfer
rate requirement should give the airlift contactor an
economic advantage, particularly for operations where
mixing is important.
Correlation of {&a& Results
A dimensional analysis of the parameters that may
affect ( K , U , ) ~in the airlift contactor was carried out.
 (KLaD), = f 9 ( u L ) r 9
[(uG>r Dd Dr , H D
9 9
resented by an equation with a correlation coefficient
of 0.97 and constants CI, C,, and C, of 2.28, 0.90,
PL? P , r , DLI 81 (14) and - I , respectively.
The effect of (UL)ron (K,a,), has been discussed In Equation (16) the Stanton number gives a measure
earlier, and it is considered as an independent parameter of the mass transfer rate relative to the convective
in equation (14) because it has been shown to be variable flow; the latter can be varied as desired in airlift con-
with geometry or design of the AL contactor. Per- tactors. Comparison of some literature data in airlift
forming a dimensional analysis on equation (14) gives contactors and continuous-liquid-operated bubble col-
the dimensionless groups in equation (15). umns is shown in Figure 14. There is quite a good
agreement between the results of this work and those
of El-Gabbani,” Lin et al.? and Kawagoe and
Robinson” in airlift contactors and Alvarez-Cuenza
and Nerenberg33in a bubble column. Large deviations
occurred for the data of Onken and Weiland’ obtained
in a 8.5-m-tall contactor and Schugerl et a1.I8 in a
4.0-m bubble column. It thereby appears that the devia-
HD
-,
Dr
Re,, We,, Pe, Fr

The effects of viscosity, surface tension, diffusivity,

1 tions result from the height effect not yet incorporated
in equation (16). Further work on the influence of
the height and diameter of the airlift contactor on
(KLaD),is therefore needed to obtain a complete di-
ionic concentration, and gravitational force were not mensionless correlation in the form of equation (16).
studied in this work. Also the effect of the important
parameter, H D / D r ,cannot be determined by the results
of this work since only one height of dispersion was CONCLUSIONS
used. Hence, for the scope of this work, correlation The riser gas holdup, ( E ~ ) and ~ , the volumetric mass
of the form of equation (16) was obtained: transfer coefficient, (K,a,), , in airlift contactors were
found to be mainly dependent not only on the superficial
gas velocity or the equivalent gassing power input, as
in the case of bubble columns, but also on the circulating
The results obtained in water gave values of C1, C,, liquid velocity in the riser. For specified liquid-phase
and C3 of 1.99, 0.87, and - 1, respectively, and 2.57, physicochemical properties, column configuration, and,
0.92, and - 1 for 0.15 kmol/m3 NaCl solutions as plotted likely, sparger type (to some degree), (U& uniquely
in Figure 14. However, there is no significant difference is estabfshed by the downcomer-to-riser cross-sectional
in the water and salt solution correlations at this level area ratio, A,/A,, at a given gas rate. The values of
of ionic concentrations. Both sets of data can be rep- ) ~ &aD), both increase with increasing gassing
( E ~ and
power input, but an increase in the circulating liquid
velocity, i.e., increased A d / A , , causes both to decrease.
Ad
A~
CONTACTOR
TVPE
W//7O
1,
,”/’ /’
,. A batch-liquid bubble column (Ad/Ar = 0) thus was
found to give higher riser gas holdup and (K,a,), than
0 044 EL the A L contactors used in this study. AL contactor
025 EL
0
A 011 EL
mass transfer performance was poorer than that of
1 A 0.13 CT Y $
A
‘
.
the BC because of the effect of the high circulating
1 m 0.35
056
CT
CT
/‘
,?..’/;d,LGABBANllll1 CT liquid velocity in the riser as well as the negligible
7b/
/
/‘
7%. 2 -LIN ET AL (41
-
3 KAWAGOE + ROBIN%
231
EL
EL mass transfer capabilities of the downcomer section.
1
4 -0NKEN + WEILAND (51 While the concentric-tube-type of airlift contactors
5 - ALVAREZ - CUENCA characteristically gave downcomer gas holdups which
+NERENBERG(321
UL * 7 5 c m l s
were 80-95% of that in the riser, the downcomer gas
-
6-SCHUGERL ET AL(IB1
UL I 6 7 - 2 21 c m h
70- DECKWER €7 AL fa)
holdups in the external-loop AL contactors were much
UL- 0 . 6 3 c m 0
lower (%SO% of the riser holdup) at the same operating

-
u
8 ~A+E R
-
7b-DECKWER ET AL ( E l
9 43 cmh
conditions. The results of this study and all available
literature data indicates a linear relationship between
9-0.15rmollm’ NoCI

003 0 0 5 0 1
I I I I l l I l 1
0 5 1
I I 1 1 U
5 10
the riser and downcomer gas holdups.
The results of this study have led to a better fun-
damental understanding of the interactions between
the various parameters that govern the operational
Figure 14. Dimensionless correIation of the volumetric mass transfer
coefficient data in airlift contactors (water and salt solution). characteristics of airlift contactors, regardless of type.
*(&aD), calculated from (aD)T measurements and kL values of Cald- Correlations applicable to the scale of contactors tested
erbank and Moo-Young.w have been presented. As a result, some guidance toward

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 379

the goal of establishing optimum design criteria for D dispersion
various potential industrial applications also has been G gas phase or pneumatic
L liquid phase
achieved. However, before this goal ultimately is
m manometric
reached, further studies using a wider range of liquid- r riser
phase physicochemical properties and done at larger T total or top
scales will be necessary in order to provide a sound
basis for the design and scaleup of airlift contactors.

References
NOMENCLATURE 1. R. A. Bello, C. W. Robinson, and M. Moo-Young, in Advances
in Biotechnology, Vol. I, Scientijic and Engineering Principles,
a, gas-liquid interfacial area per unit volume of gas-
M. Moo-Young, C. W. Robinson, and C. Vezina, Eds. (Pergamon
liquid dispersion (m’/m’)
Press, Toronto, 1981), p. 547.
A cross-sectional area (m’)
2. R. A. Bello, C. W. Robinson, and M. Moo-Young, Can. J.
Co, initial dissolved oxygen concentration (mol/m’)
Chem. Eng., 62, 573 (1984).
C, dissolved oxygen concentration at any time (mol/m’)
3. R. T . Hatch, “Experimental and Theoretical Studies of Oxygen
C
, final dissolved oxygen concentration (mol/m’)
Transfer in the Airlift Fermentor,” Doctoral Thesis, MIT, Cam-
constant in equation (16) (dimensionless)
bridge, Mass. (1973).
exponent in equation (16) (dimensionless)
4. C. H . Lin, B. S. Fang, C. S. Wu, H. Y. Fang, T. F. Kuo, and
exponent in equation (16) (dimensionless)
KC. Y. Hu, Biotechnol. Bioeng., 18, 1557 (1976).
Sauter mean bubble diameter (m)
5. V. Onken and P. Weiland, in Advances in Biofechnology, Vol.
internal diameter (m)
I, Scientific and Engineering Principles, M. Moo-Young,
oxygen diffusion coefficient in liquid phase (m’/s)
C. W. Robinson, and C. Vezina, Eds. (Pergamon Press, Toronto,
Froude number, F r = U : / g D , (dimensionless)
1981), p. 559.
gravitational acceleration (m/s2) 6. F. Yoshida and K. Akita, AIChE J., 11, 9 (1965).
G gas molar flowrate (mol/s) 7. F. Kastanek, J. Zahradnik, M. Rylek, and J. Kratschvil, Chem.
H Henry’s Law constant (N m/mol) Eng. Sci., 35, 456 (1980).
Ho height between horizontal connections (m) 8. W. D. Deckwer, R. Burckhart, and G. Zoll, Chem. Eng. Sci.,
HD dispersion height (m) 29, 2177 (1974).
KL overall mass transfer coefficient, based on liquid- 9. J. H . Hills, Chem. Eng. J . , 12, 89 (1976).
phase concentration driving force (m/s) 10. M. Chakravarty, S . Begum, H. D. Singh, J. N. Bamah, and
KLaD overall volumetric mass transfer coefficient based on M. S . Lyengar, Biotechnol. Bioeng. Symp. N o . 4 , 363 (1973).
gas-liquid dispersion volume (s-‘) 11. D. E. D. H. El-Gabbani, “Hydrodynamic and Mass Transfer
KLa, overall volumetric mass transfer coefficient based on Characteristics of an Airlift Contactor,” M.A.Sc. Thesis, Uni-
liquid volume (s-l) versity of Waterloo, Ontario (1977).
Pe, riser Peclet number, Pe = (UL),D,/DL (dimensionless) 12. M. E. Orazem, L. T. Fan, and L. E. Erickson, Biorechnol.
pressure at the top of the contactor (N/m’) Bioeng., 21, 1579 (1979).
pneumatic power input per unit gas-liquid dispersion 13. H. Fukuda, T. Shiotani, W. Okada, and H. Morikawa, J. Fer-
volume (W/m’) ment. Technol., 56, 619 (1978).
volumetric gas rate (at 101.3 kPa and 0°C) (m’/s) 14. L. L. Gasner, Biotechnol. Bioeng., 16, 1179 (1974).
riser Reynolds number, Re, = pL(UL),D,/pL IS. R. D. Botton, D. Cosserat, and J. C. Charpentier, Chem. Eng.
(dimensionless) J . , 20, 87 (1980).
riser Stanton number, St, = (KLaD),HD/(uL), 16. M. E. Orazem and L. E. Erickson, Biotechnol. Bioeng., 21,
(dimensionless) 69 (1979).
superficial gas velocity, based on A (m/s) 17. C. G . Sinclair and D. N. Ryder, Biofechnol. Bioeng., 17, 375
superficial liquid velocity, based on A (m/s) (1975).
volume in contactor (m’) 18. K. Schugerl, J. Liicke, and V. Oels, in Advances in Biochemical
Weber number, We, = p , D , ( U , ) f / a (dimensionless) Engineering, Vol. 7, T. K. Ghose, A. Fiechter, and N. Blake-
Greek letters brough, Eds. (Springer Verlag, Berlin, 1977), p. 1.
slope of In (transient dissolved oxygen concentration) 19. F. Kastanek, Coll. Czech. Chem. Commun., 41, 3709 (1976).
P 20. R. A. Bello, “A Characterization Study of Airlift Contactors
vs. time, P = A{ln[(C, - Cof)/(Cor - C,,)I}/At
for Applications to Fermentations,” Ph.D. Thesis, University
yI parameter defined in equation (2), y , = GH/(VL)r(Pr
of Waterloo, Ontario (1981).
+ P&HL/~) (s-’)
21. R. Botton, D. Cosserat and J. C . Charpentier, Chem. Eng. J . ,
yz parameter in equation (2), y2 = P , / ( P , + pLgHL)
16, 107 (1978).
(dimensionless)
22. K. Akita and F. Yoshida, Ind. Eng. Chem. Process Des. Develop.,
-y3 parameter in equation (2), y, = GH/(V,),P, (s-’)
12, 76 (1973).
Ah height differential in column (m)
Az manometric height differential (m) 23. J. F. Marchatere and B. W. Hoglund, Nucleonics, 20(8), 142
.zG gas holdup (dimensionless) (1962).
I*. liquid viscosity (N s/m’) 24. J. G . Collier, Convective Boiling and Condensation (McGraw
P mass density (kg/m3) Hill, London, 1972).
0- interfacial surface tension (N/m) 25. M. Kawagoe and C. W. Robinson (unpublished data).
26. N. Zuber and J. A. Finlay, J . Heat Trans., 87, C(4), 453 (1965).
Subscripts 27. D. S. Scott, in Advances in Chemical Engineering, Vol. 4,
B bubble T. B. Drew, J. W. H o o p s Jr., and T. Vermuelen, Eds. (Academic
d downcomer Press, New York, 1963) p. 199.

380 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 27, MARCH 1985

28. M. Chakravarty, H. D. Singh, J. N. Baruah, andM. S. Lyengar,
Indian Chem. Eng., 16(3), 17 (1974).
29. H. Blenke, in Advances in Biochemical Engineering, Vol. 13,
A. Fiechter, Ed. (Springer-Verlag, Berlin, 1979), p. 121.
30. P. H. Calderbank and M. B. Moo-Young, Chem. Eng. Sci.,
16, 39 (1961).
31. H. L. Shulman and M. C. Molstad, Ind. Eng. Chem., 42, 1058
( 1950).
32. A. A. Vinter, L. N. Danozhkina, and I. Y. Gorodetskii, Inf.
Chem. Eng., 12(3), 392 (1972).
33. M. Alvarez-Cuenca and M. A. Neremberg, AIChE J . , 27(1),
66 (1981).

BELLO, ROBINSON, AND MOO-YOUNG: AIRLIFT CONTACTORS 381