You are on page 1of 13

Chemical Engineering Science. Vol. 48, No. 23. pp.

40234035,
Printed in Gnnc Britain.

1993.

000-2509,93
S6.W + 0.M
0 1993 Pergamon Press Ltd

AN AVERAGE GAS HOLD-UP AND LIQUID CIRCULATION


VELOCITY IN AIRLIFT REACTORS WITH EXTERNAL LOOP
Z. KEMBLOWSKI.
Department

J. PRZYWARSKI

and A. DIAB

of Chemical Engineering, Faculty of Process and Environmental


Engineering,
University of L6di Wblczahska 175.90-924 L&IS% Poland

Technical

(First received 7 July 1992; accepted in revised form 12 February1993)


Abstract-Experimental
investigations were carried out in model airlift reactors with external loops. Two
reactors of laboratory scale (height 1.95 m, internal diameter 0.1 m, volume 0.08 m3) and pilot-plant scale
(height 7.1% m, internal diameter 0.2 m, volume 0.7 m3) were used. The influence of reactor geometry, gas
sparger design, liquid properties (both Newtonian and non-Newtonian)
and the amount of introduced air
was investigated. The influence of gas sparger design on gas hold-up and liquid velocity was found to be
negligible. A modified method for the prediction of liquid circulation velocity-based
on the energy balance
of the reactor-was
proposed. An original dimensionless correlation for gas hold-up prediction involving
superficial velocities of gas and liquid, cross-sectional areas, as well as Froude and Morton numbers, was
obtained.

1. INTRODUCTION
During

the last two decades a large number of papers

concerning airlift loop reactors have appeared. The


principles
of operation
as well as the advantages and
disadvantages of this type of reactors are discussed in
the review papers (Chisti and Moo-Young, 1987;
Blenke, 1979, 1983; Schiigerl et al., 1977).
The basic hydrodynamic design parameters of an
airlift loop reactor are the average gas hold-up and
liquid circulation velocity.
1.1. Literuture survey
Severai important papers dealing with the two interrelated parameters of average gas hold-up and
liquid circulation velocity have been published (Bell0
et aI., 1984, 1985;Calve, 1989,Chisti et a& 1988;Chisti
and Moo-Young,
1988; Hills, 1976; Hsu and
DudukoviC, 1980; Jones, 1985; van der Lans, 1985;
Merchuk and Stein, 1981; Nicol and Davidson,
1988a,b; Philip et al., 1990; Popovii: and Robinson,
1984, 1989; Roberts, 1979; Vatai and Tekib, 1986;
Verlaan et al., 1986a). Most of them describe experimental investigations carried out for water-like media
in various model reactors of different geometries and
mostly laboratory scale [except the data presented by
Hills (1976), van der Lans (1985) and Nicol and
Davidson (1988a, b)]. Only a few papers concern
highly viscous and non-Newtonian liquids (Chisti and
Moo-Young,
1988; Popovil: and Robinson, 1984;
1989; Philip et al., 1990; Vatai and TekiC, 1986).
Some correlations for gas hold-up prediction have
been proposed in the literature (Bell0 et al., 1984,
1985; Hills, 1976; Hsu and Dudukovie, 1980;
Merchuk, 1986; Nicol and Davidson, 1988a,b;
Popovii: and Robinson, 1984, 1989; Roberts, 1979;
Vatai and Teki& 1986;Weiland and Onken, 1981).The
greatest disadvantage of most of the reported correlations is their dimensional form, which means that the
constants of the equations depend on the geometry

and properties of the system. Only the correlations


proposed by Hsu and DudukoviC (1980) and Vatai
and TekiC (1986) seem to be more general than the
others because they contain dimensionless parameters
such as Froude, Reynolds and Weber numbers, and
superficial velocities ratio of gas and liquid. However,
there is a lack of evidence showing that they have been
applied for other geometries. According to some researchers there is an obvious need for studies of largescale reactors (Blenke, 1979; Bello et al., 1984).
Various equations for liquid velocity prediction
have also been reported in the literature, generally for
water or similar liquids. Most of them are simple
correlations involving two or three parameters (Be110
et al., 1984; Nicol and Davidson, 1988a,b, Popovii:
and Robinson, 1988; Roberts, 1979; Vatai and TekiC,
1986). Some of them are semitheoretical model equations derived from energy or pressure balance equations. They take into account most of the known
factors that influence liquid circulation but not all of
the terms involved are known or may be easily estimated for calculations (Chisti et al., 1988; Chisti and
Moo-Young, 1988;Hsu and DudukoviE, 1980). There
are also some very simple correlations available derived from drift flux model but they take into account
neither the properties of gas and liquid nor the
geometry of the reactor (Calvo, 1989; Nicol and
Davidson, 1988a,b). Therefore, reported results are
mostly limited to particular cases for which they were
obtained.
It follows from the above-mentioned literature data
that there are two main approaches to the modelling
of liquid circulation velocity in an airlift loop reactor.
One of them begins with a pressure balance over the
circulation loop (Blenke, 1979, Hsu and Dudukovik,
1980; Kubota et al., 1978; Merchuk and Stein, 1981;
Verlaan et al., 1986b). The other one begins with
a total energy balance of the reactor (Chisti et al.,
1988; Chisti and Moo-Young, 1988; Calvo, 1989;
Young et al., 1991).

4023

2. K~MBLOWSKI et al.

4024

Only the model proposed by Chisti and MooYoung (1988), concerning both Newtonian and nonNewtonian liquids, will be discussed further because
this approach seems to be more convincing. The
authors wrote the energy balance of the reactor in the
following form:
Ei=E,+ED+E,+ET+EF

(1)

ments were carried out in two external-loop model


airlift reactors of distinctly different geometries. The
purpose of the work was to develop
-a

dimensionless correlation for the prediction of


average gas hold-up, and
-a
simple method for the prediction of liquid
circulation velocity.

where Ei is the energy input due to isothermal gas


expansion
2. A MODEL
HOLD-UP

E, is the energy dissipation due to wakes behind


bubbles in the riser,

E. = Ei -_p&,VLx&e~

(3)

I?,, is the energy dissipation due to stagnant gas in the


downwmer,
.% = PLgh, VU&&,

(4)

& is the energy dissipation due to fluid turn around


at the bottom of the reactor,

zP&(1 -ed
2

VLD

-AD
(1

-eD)

(5)

E, is the energy dissipation due to fluid turn around

at the top of reactor,

zPU

--El!)
2

VLR
(1 -.5x) AR

(@

and finally EP is the energy loss due to friction in the


riser and the downcomer.
EF = APFa VLx AR + APp,

VLD AD.

(7)

Applying complicated experimental correlations


for wall shear stress in tube flow of gas-liquid mixture,
given by Sokolov and Metkin (1981) and Metkin and
Sokolov (1982), Chisti and Moo-Young (1988) developed an equation for the prediction of liquid circulation velocity (see Table 1). For the prediction of
gas hold-up-which
is necessary for the calculations-they
recommended the correlation proposed
by Popovii: and Robinson (1984). This correlation
was obtained for power-law liquids in a laboratoryscale model reactor.
Summing up, it follows from the literature review
that
-the
reported investigations were concerned
mainly with Newtonian liquids of low viscosity,
-the majority of the published data were obtained
in laboratory-scale model reactors, and
-the results of measurements were usually presented in the form of dimensional dependencies.
1.2. Scope of the work
Taking into account the above statements our investigations were concerned
with highly viscous
Newtonian and non-Newtonian fluids. The experi-

FOR
AND

PREDICTION

LIQUID

OF AVERAGE

CIRCULATION

GAS

VELOCITY

2.1. Gas hold-up


It was decided to develop a classical dimensionless
correlation for gas hold-up prediction in order to
avoid the modelling of gas-liquid interaction in a twophase flow. Reported results concerning the influence
of various physical factors on the hydrodynamics of
an airlift reactor are often contradictory and strongly
affected by system properties. Therefore, they have
limited general importance [e.g. Nicol and Davidson
(1988a,b) and contradictory conclusions on surface
tension influence reported by Wachi et al. (1991)]. Let
US consider a generalized Newtonian liquid (i.e. purely
viscous fluid with shear-rate-dependent viscosity)
whose rheological properties can be approximated by
the power-law model of Ostwald-de Waele. We assume, on the basis of literature data and our own
considerations, that the average gas hold-up in the
riser, &R,is a function of superficiti velocities of gas,
V,,, and liquid, VLR, densities of gas, Pc, and liquid,
pL, liquid rheological parameters, n and k, surface
tension, crL, height of the liquid head, H, cross-sectional areas of the riser, AR, and downcomer, Ar,,
total cross-sectional area of holes or nozzles of the gas
sparger, A,, and acceleration due to gravity, g:
ER

I(v,R,

VLR,

PC,

PL,

n,

OL,,

H,

AD,

AR,

Ag).
(8)

A dimensional analysis leads to the following relation between gas hold-up and dimensionless moduli:

where the Froude number is defined as


Fr

(VLR

vGR)2

(10)

gdn
and the Morton number is defined as
4(n-1)

3n

4n

1 14

I *
(11)

A classical relation of Morton number to other


well-known numbers is as follows:
MO=-

Eo Wet
R&R

(12)

Chisti and Moo-Young, 1989

Re,.<23aI~,

For laminar flow:

7,

ah
7L=y-&.

_+Lir
(I -es)*

I ++s)($>

L
where Ap = 4 D

I
+~P&&

VLR=

Chisti et al., 1988

=,.I,(

ER

Bell0 et al., 1985

Belle et al., 1984

Author

vt

-0

>I

!!
s
(1 -~a)* ( A,,

Proposedcorrelation

= 0.1 l-O.69

$=O.ll-I

Medium: CMC solutions


H=3.2lm
Reactor volume 1.46m3
Ds = 0.142m

+s=O.ll-I
D

Medium: water
H = 3.21 m
Reactor volume I .46 m3
Dn = 0.142 m

holes of 1.02mm

Da = 0.152 m, H = I.8 m
V,, = 0.0137-0.086 m/s
Sparger:perforated stainlessplate with 52

Media: water and 0. I5 kmoljm NaCl

holes of 1.02mm

2 = 0.1 l-O.69
AR
Da = 0.152 m, H = I.8 m
V,, = 0.0137-0.086 m/s
Sparger:perforated stainlessplate with 52

Media: water and 0. I5 kmol/m3 NaCl

Exuerimental range

Table I. Selectedliterature correlationsfor the prediction of gas hold-up and liquid circulation velocity in airlift reactors

Merchuk and Stein, 1981

Hills, 1976

Author

k(
n
)

6n + 2

8v;-D4
, x =

o.7n-0~591, c = 411+n~,n(~)~~~1_~~2(~-~

fD = 0.0791Re-0.25

fR = 0.046Re-0~2

ss=~[(t+~)~~ts(lt~)(~~]

= 0.24 t 1.35fV: t VL,)o.q3

Re,, =

where

For turbulent flow:

Proposed correlation

Table 1. (Coned.)

0.14x0.14m
H = 4.05 m
Single-orifice and multinozzle sparger

Rectangular columns

Medium: water

V,, t 0.3 m/s

Medium: water

Experimental range

. ..

____

Vatai and Tekic, 19B6

.-

PopoviC and Robinson, 1989

Popnvic and Robinson. 1988

Popovic and Robinson, I984

Philip er al., 1990

..___..~

0.416

.__

v~o.82p~i~l0s

01= 0.052 for bubbly flow,


0.0204 for slug flow

p=5OOOV,,

pen= kjr-

for Newtonian liquid


0.35 for non-Newtonian liquid

_3n+l
1 -nS1

C2 = 0.24-0.33

= 0.1 l-0.44

= 0.1 l-0.44

Media: water, CMC solutions


(n = 0.866, k = 0.304-0.745 Pas)

AR
perr = 0.02-0.5 Pas
pL = 1003-1240 kg/m3
uL = 0.059-0.079 Nfm

2 =0.1l-0.44

Media: CMC sohlions, 5 I .8%sucrose


V,, = 0.03-0.26 m/s

Da=O.lSm,H=l.88m

Media: CMC sotutions


y = 0.015-0.5 Pa s

Da=O.iSm,H=1.88m

Media: CMC solutions


keff = 0.015-0.5 Pas

Media: mineral oils, sugar


syrup, CMC, xanthan
Internal-loop reactor

z. KEMBLoWsKr
where Eo is the Eotvos number, WeL the Weber
number for the liquid and ReLR the Reynolds number
for the liquid in the riser column.
If one applies the following classical definitions of
Eotviis number,
E.

~PL --PoWi
OL

Weber number

(13)

for the liquid,


(14)

and takes the Reynolds number for the liquid as


a generalized Reynolds number by Metzner and Reed
in the special case of a power-law fluid,

et al.

al. (1988) was chosen. However, their suggestion to


calculate pressure drop according to Sokolov and
Metkins model was not followed on the basis of the
opinion expressed by Verlaan et al. (1986a). According to these authors, if the average gas hold-up is less
than lo-15%,
the friction losses in the riser can be
calculated as for single-phase flow of liquid. Because
usually one may not expect higher values of average
gas hold-up in the riser of airlift reactor with external
loop, we suggest determining the energy loss due to
friction in the riSer and downcomer sections by means
of the Fanning equation in the following form:
L
APF

where f is the Fanning

2.2. Liquid circulation velocity


In order to correlate the experimental data on
liquid circulation, the approach proposed by Chisti et

(16)

vh

friction factor. Hence,

EF = 2f$

then eq. (11) will be obtained. Equation (11) may,


therefore, be considered as a generalized definition of
the Morton number for two-phase Bow of gas and
power-law liquid.
A classical Morton number for Newtonian fluids
was previously used in the case of two-phase flow by
Cliff er al. (1978) and also by Miyahara er al. (1986)
[according to Chisti and Moo-Young (1987)].
It is well known that the average gas hold-up in the
riser depends strongly on liquid superficial velocity.
On the other hand, the liquid velocity is influenced
not only by the amount of gas introduced but also by
the reactor geometry. In other words, the liquid velocity is also a function of the reactor geometry. Therefore, some of the geometrical ratios proposed in eq. (9)
are represented by liquid velocity term and can probably be omitted in the final form of the correlation.

2f -

(17)

VipL( VLA).

We propose to calculate the Fanning


using the following equations:

friction factors

for laminar flow (ReL < 1000)

/=E
for turbulent

(18)

flow (ReL > 2000)

f=-

0.0792
Ret

(19)

where the Reynolds number of the liquid is defined in


the case of the riser by eq. (15) and in the case of the
downcomer by the following equation:
ReL1, =

VtDdh
8m-,k

(20)

The above ranges of laminar and turbulent flow were


proposed by Hsu and DudukoviC (1980).
Combining eqs (l)-(6) and (17) for riser and downcomer columns, a general equation for the prediction
of the liquid superficial velocity in the riser is obtained:

If the phases are almost completely separated at the


top of the reactor, gas hold-up in the downcomer is
often less than 1%. In this case eq. (21) can be rewritten in the form

An average gas hold-up and liquid circulationvelocity


Equation (22) incorporates the geometrical dimensions of the reactor, the Fanning friction factors in the
riser and downcomer, frictional loss coefficients at the
top and bottom of the reactor, and the gas hold-up in
the riser. The geometry of the reactor is known, the
Fanning friction factors may be assumed roughly and
checked after calculations, and the frictional loss coefficients at the top and bottom of the reactor may be
taken from the literature. The only unknown parameter is the average gas hold-up, which should be
predicted separately.

3. EXPERIMENTAL

3.1. Media
Both Newtonian (tap water without and with
added surfactants, glycol, sugar syrup) and nonNewtonian media (power-law CMC solutions) were
used for investigations concerning verification of
energy balance approach. The following ranges of
media properties were obtained:
-liquid
density
- surface tension
-power-law
parameter
-power-law
parameter

pL
oL
n
k

=
=
=
=

4029

the valves applied in the experiments-one


totally
open and the other completely closed. The schematic
views of the reactors and the photograph of the upper
part of the pilot-plant scale apparatus are shown in
Figs l-3.
Compressed air was introduced only through the
bottoms of the reactors by means of various gas
spargers. There were gas spargers containing porous
plates made of glass beads of three different size
ranges: 100-160, la-250
and 250-500~.
Each of
them was applied with three diameters: 30, 50 and
70 mm. Single-nozzle gas spargers of six diameters-l,
1.4, 2, 2.8, 3 and 4 mm-were also used.
For the pilot-plant scale reactor only the 3 mm gas
sparger was applied.
Air flow rate, average gas hold-up in the riser and
downcomer, liquid velocity in the downcomer, pressure of compressed air, temperature of the system and
atmospheric pressure were measured during the experiments. For each run properties of the liquid phase
were carefully determined. Rheological properties
were measured by means of the rotational rheometer
Rheotest 2. Surface tension was determined by the
classical stalagmometric method. Average gas holdup was determined by means of the manometric tcchnique described by Hills (1976). The following equation was applied for the estimation of the average gas
hold-up:

998-1286 kg/m,
0.0420-0.08 16 N/m,
0.758-1,
0.001-0.261 Pa s.

The properties of the experimental media for which


the data points presented in Figs 4-12 were obtained
are summarized in Table 2.

Ah,

s=-

AZ

3.2. Apparatus and procedure


Two mode1 reactors were used for the experiments.
One, of laboratory-scale, was 1.95 m high, of 0.1 m
internal diameter of the riser and the downcomer, and
80 dm3 volume. The other one, of pilot-plant scale,
was 7.18 m high and had a riser of 0.2 m internal
diameter. It had two possible geometrical arrangements, with the downcomer internal diameter equal to
0.15 or 0.2 m. The total volume of the second reactor
was 700 dm.
Liquid circulation was directed through one of the
two downcomers of the pilot-plant scale reactor by
butterRy valves placed in the bottom connections of
the reactor. There were only two possible positions of

where Ah,,, is the difference of manometer readings


and AZ is the vertical distance between the connections of manometers to the column.
Liquid circulation velocity was measured in the
downcomer according to the flow follower method,
described previously by Jones (1985) and Philip et al.
(1990), and modified by Diab (1991). Separation of
phases was complete at the tops of the model reactors.
Therefore, only a single-phase flow existed in the
downcomers. In this way, it was possible to determine
the liquid superficial velocity in the riser applying the
continuity equation for liquid flow. The procedure for
evaluation of the mean linear velocity of the liquid in
the downwmer was as follows:

Table 2. Properties of experimentalmedia used for data presentationin Figs 4-12

Figure

Density
Medium

b/m)

Surfacetension

Power-law parameters
k

V/m)

(Pa s)

Glycol solution

1011

0.0699

0.0015

Water
Water with surfactant
Glycol
Glycol solution
Sugar syrup
CMC solution

1000
997
1011
1284
1010

0.0735
0.0690
0.0420
0.0699
0.0814
0.0769

1
1
1
:
0.758

0.001
0.001
0.0176
0.0015
0.141
0.261

6, 9

Water
CMC solution

998
1018

0.0726
0.0802

1
0.855

0.001
0.212

7, 10

Water

998

0,0726

0.001

5, 8

1115

Z. KEMBLOWSKI et a/.

4030

Fig. 1. Schematic diagram of the laboratory scale airlift


model reactor with externalloop.

The velocity of small thin aluminium flakes


floating in the liquid was measured in the downcomer 10 to 15 times in order to obtain the
maximum velocity in the axis of the column,
According to the velocity profile determined
previously as a function of Reynolds number of
the liquid in the downcomer, the mean linear
velocity was calculated. All details were given by
Diab (1991).
4. RESULTS

AND DISCUSSION

4.1. Gas spargers


The influence of gas sparger geometry on the average gas hold-up in our model reactors, despite the
changes offlow regime from uniform bubbly to chumturbulent flow, has proved to be negligible during the
experiments. Figure 4 presents some experimental
data obtained in the model reactor of laboratory scale
for ethylene glycol solution and for various gas
spargers. For other media the results are similar.
Therefore, the correlation obtained for gas hold-up
prediction, proposed below, can be regarded-as
a first approximation-as
being independent of gas
sparger geometry.
4.2. Gas hold-up
The gas hold-up data obtained in our two model
reactors-as
well as other complete data available in
the literature (Bell0 et al., 1985; Merchuk and Stein,
1981; Roberts, 1980) concerning airlift external-loop
reactors with complete phase separation at the
top-were
applied in order to obtain a dimensionless
correlation for gas hold-up prediction. The final form

Fig. 2. Schematic diagram of the pilot-plant scale airlift


model reactor with external loop.
of the equation is
(24)
The range of experimental
eq. (24) was as follows:

data

used to develop

Ea = 0.002-0.21
V,,

= 0.001-0.50 m/s

V,,

= 0.07-1.3 m/s

V
-YE = l-153
VCR
Fr = 0.00-14.1
MO = 2.47 x lo- -0.390
AD
= 0.11-l.
AR
Additionally, geometrical parameters not included in
the correlation were of the following ranges:
H
- = 10.2-22s
DR
A
AR = (5.60-360)

x lo- 5

ReLR = 40-130,000.

An average gttshold-up and liquid circulation

velocity

Fig. 3. View of the upper part of the pilot-plant scale airlift reactor with external loop.

The exponent of the Morton number in eq. (24) is


actually very small. One may, therefore, expect that
the Morton number is negligible. However, the range
of numerical values of this number for the experimental data is so wide that even taking the power to
be 0.012 it changes from 0.746 to 0.989.
Plots presenting the dependencies of the experimental values of the average gas hold-up as a function
of gas superficial velocity, obtained in our model
reactors, are shown in Fig. 5. They illustrate a strong
influence of liquid properties and reactor geometry on

the results obtained. The values of the experimental


average gas hold-up in the riser are plotted against
those calculated using eq. (24) in Figs 6-8. Their comparison shows a satisfactory validity of eq. (24) with
up to f20% deviation.
4.3. Liquid circulation velocity
Hots presenting the dependencies of the experimental values of the liquid superficial velocity in the
riser as a function of gas superficial velocity, obtained
in our model reactors, are shown in Fig. 9. They

2. KEMIILOWSKIet al.

4032
0.07
n

O.Ci5
0.05

0.06

A 4A
0

0.04

0.04

Ox)3

I
-=0.03

. 0.02
S

0.02

0.0 1
O.oi
0.00
0.00

L
0.00

0.00

0.06

0.0

0.02 0.03 0.04 0.06

0.06

O.i5

0.10

OR

en. predicted
1-l

Imlsl

Fig. 4. Experimental values of the average gas hold-up in the


riser column of the laboratory model airlift reactor obtained
for various gas spargers(experimental medium-glycol
solution); single-nozzle spargers: (0) 4 mm; (A) 2.8 mm; (I)
2 mm; (A) 1.4 mm; (Cl) 1 mm; porous-plate spargers: (0)
3Omm;(+)50mm;(V)70mm.

Fig. 6. Comparison of the experimental values of the average gas hold-up in the riser column of the laboratory model
airliftreactor with those.predicted by eq. (20); single-hole gas
sparger 2.8 mm (symbol; asin Fig. 5).

J
0.00

J
V,

Lm/sl

Fig. 5. Experimental values of the average gas hold-up in the


riser column of model reactors; laboratory scale model reactor (single-hole gas sparger 2.8 mm), experimental media:
(0) water; (0) water with added surfactant; (A) glycol; (A)
glyml solution; ( W) sugar syrup; (Cl) CMC solution; pilotplant scale model reactor (single-hole gas sparger 3 mm),
water: (0) AD/AR = 1; (V) AD/AR = 0.56;
CMC solution:
(+) AD/AR = l;(V) AD/AR =0.56.
illustrate a strong influence of liquid properties
and
reactor
geometry
on the results
obtained.
Figures lo-12
present a comparison
of the proposed

model with our experimental data and the data given


by Roberts (1979). One may see that over a large
range of reactor size and liquid properties, the proposed model predicts liquid superficial velocity in the
riser with up to + 20% error. The higher deviation of
the calculated values observed in the range of small
velocities (i.e. in the laminar region) may be caused by
the assumption that friction loss coefficients at the top
and the bottom of the reactor do not depend on the
superfkial Reynolds number of the liquid. Therefore,

0.0

0.03

0.02
6,

Cxedlcted

0.04

I-l

Fig. 7. Comparison of the experimental values of the average gas hold-up in the riser column of the pilot-plant model
airlift reactor with those predicted by eq. (20); single-hole gas
sparger 3 mm (symbols as in Fig. 5).

slightly too conservative values of these coefficients


are used in the calculations in the laminar region (the
values of the coefficients actually used are summarized in Table 3).
Two basic equations
[eqs (22) and (24)] enable the
prediction of the liquid circulation
and gas hold-up in

airlift external loop reactors with almost complete


phase separation at the top. These parameters can be
predicted provided the geometry of the reactor, the
liquid properties and the amount of gas to be introduced are known. The procedure for the prediction of
the gas hold-up and liquid circulation velocity in the
riser can be summarized as follows:
(1) Assume liquid superficial velocity in the riser.
(2) Estimate gas hold-up using eq. (24).
(3) Predict liquid velocity using eq. (22) and other
related equations.

An average gas hold-up and liquid circulation velocity

4033

-I

0.20

0.30

en. aedcted

0.40

f-1

Fig. 8. Comparison

of the experimental values of the average gas hold-up in the riser column of airlift reactor, published in the literature, with those predicted by eq. (20)
(experimental medium-water);
Be110 et al. (1985): (A)
AD/AR=0.25; (V) AD/AR = 0.11; Roberts
(1979),
AD/AR = 1: (0) ID 75 mm; (I) ID 25 mm; (0) Merchuk
(1986),AD/AR = 1.

Fig. 10. Comparison of the experimental values of the liquid


superficial velocity in the riser column of the laboratory
model airlift reactor with those predicted by the proposed
model (symbols as in Fig. 5).

_-

-1

,.*~~,o
_-,f-p

1ii/
,

r
*:,

,+

,*
<, ,I

,e*

#I

,?-

0,

,**

1.

o.ov,
0.0

0.1

02

0.3

cl.4

0.5

0.6

0.7

0.8

0.9

i.0

Fig. 9. Experimentalvaluesof the liquid superficialvelocity


in the riser column of model reactors (symbols as in Fig. 5).

Fig. 11. Comparison of the experimental values of the liquid


superficial velocity in the riser column of the pilot-plant
model airlift reactor with those predicted by the proposed
model (symbols as in Fig. 5).
1.5,
s\T,

(4) If the assumed and calculated values of the


liquid velocity do not agree satisfactorily, then
return to point (2) with the last calculated value
of the liquid velocity until sufficient agreement
is obtained.

XL,,
,

-/

5. CONCLUSIONS

The results of the experimental investigations presented lead to the following conclusions:
-The
influence of the gas sparger geometry on the
average gas hold-up proved to be negligible,
despite the observed changes of flow regime from
uniform bubbly to churn-turbulent flow.
-The liquid circulation velocity cannot be directly
related to the gas superficial velocity because it
depends also on reactor geometry and gas and
liquid properties.

0.0

v
0.0

0.5

i.0

v,.

7.6
wedicted

lm!sl

Fig. 12. Comparisonof the experimentalvaluesof the liquid


superficialvelocity in the riser column of the airlift reactor
presented by Roberts (1979) with those predicted by the
proposed model (symbols as in Fig. 8).

Z. KEMFJLOWSKI et al.

4.034

Table 3. Values of the local friction factors applied in the calculations, according to Maksimov and Orlov (1949)
Experimental data obtained in/by
Laboratory scale model reactor
Pilot-plant scale model reactor, 200/200mm
Pilot-plant scale model reactor, 200/150 mm
Roberts, 1979

-The
same conclusion is valid for the average gas
hold-up.
-The
simple model [see eqs (H-(22)]
proposed
for the prediction of the liquid circulation velocity in an airlift reactor with external loop gives
satisfactory results in a relatively wide range of
physical and geometrical parameters for both
Newtonian and nun-Newtonian liquids.
-The
proposed dimensionless correlation for the
prediction of the average gas hold-up in the riser
column of an airlift reactor with external loop
[see eq. (24)] also gives satisfactory results for
a relatively wide range of physical and geometrical parameters for both Newtonian and nonNewtonian liquids.
Acknowledgement~The authors acknowledge the financial
support provided by the Polish State Committee of Scientific
Research (Grant no. 3 1266 9101).
NOTATION

A
d
E
Eo

f
Fr
9
h, H
k
L
MO

Q
Re
V
We

Ah
AP

AZ

Greek
&

e
p
P
d

cross-sectional area, m*
internal diameter, m
energy dissipation or input rate, W
Eijtvijs number, defined by eq. (13)
Fanning friction factor
Froude number, defined by eq. (10)
acceleration due to gravity, m/s2
height, m
power-law parameter, Pa s
length of the tube, m
generalized Morton number for power-law
liquids, defined by eq. (11)
power-law parameter
pressure, Pa
gas flow rate, m3/s
Reynolds number, defined by eqs (15) and
(20)
superficial velocity, m/s
Weber number, defined by eq. (14)
height difference, m
pressure drop, Pa
vertical distance of manometer connections, m
letters

gas hold-up,
friction loss coefficient
viscosity, Pas
density, kg/m3
surface tension, N/m

1.8
2.95
3.01
2.4

1.3
1.3
1.43
2.4

Subscripts
B
bottom of the reactor
d
gas-liquid dispersion
D
downcomer

et7
F

G
h

i
L
m

:
T

effective
friction
gas phase
head space
input
liquid phase
manometer
nozzle or hole of the gas sparger
riser
top of the reactor

REFERENCES

Bello, R. A., Robinson, C. W. and Moo-Young, M., 1984,


Liquid circulation and mixing characteristics of airlift contactors. Can. J. them. Engng 62, 513-577.
Bello, R. A., Robinson, C. W. and Moo-Young, M., 1985,
Gas hold-up and overall volumetric oxygen transfer coefficient in airlift contactors. Biotechnol. Bioengng 27,
369-381.
Blenke, H., 1979, Loop reactors. Adv. Biochem. Engng 13,
121-214.
Blenke, H., 1983, Biochemical loop reactors, in Eiotechnology. Vol. 2, Fundamentals of Biochemical Ehgineering,
pp. 4455464.
Calvo, E. G., 1989, A fluid dynamic model for airlift loop
reactors. Chem. Engng Sei. 44, 321-323.
Chisti, M. Y., Halard, B. and Moo-Young, M., 1988, Liquid
circulation in airlift reactors. Chem. Engng Sci. 43,
451-457.
Chisti, M. Y. and Moo-Young, M., 1987, Airlift reactors:
characteristics, applications and design considerations.
C/rem.Engng Commun. 60, 195-242.
Chisti, M. Y. and Moo-Young, M., 1988, Prediction of liquid
circulation velocity in airlift reactors with biological
media. J. Chem. Technol. Biotechnol. 42, 21 l-219.
Clift, R.. Grace, J. R. and Weber, M. E., 1978, Bubbles, Drops
and Particles, Chap. 7. Academic Press, New York.
Diab, A. M., 1991, An average gas hold-up and liquid circulation velocity in airlift bioreactors with external loop.
Ph.D. thesis (in Polish), Technical University of todi,
Poland.
Hsu, Y. C. and Dudukovie, M. P., 1980, Gas hold-up and
liquid recirculation in gas-lift reactors. Chem. Engng Sci.
35, 1355141.
Hills, J. H., 1976, The operation of a bubble column in high
throughputs 1. Gas hold-up measurement. Chem. Engng J.
12, 89-99.
Jones, A. G., 1985, Liquid circulation in a draft-tube bubble
column. Chem. Engng Sci. 40, 449-462.
Kubota, H., Hosono, Y. and Fujie, K., 1978, Characteristic
evaluation of ICI air-lift type deep shaft aerator. J. Chem.
Engng Japan 11, 319-325.

An average gas hold-up

and liquid circulation

van der Lasts, R. G. J. M., 1985, Hydrodynamics


of a bubble
column loop reactor. Ph.D. thesis, Technical University of
Delft, The Netherlands.
Maksimov.
G. A. and Orlov, A. I., 1949. Insulation aad
Ventilation. Part 2. Ventilation, pp. 20, 236-240. Stroiixdat, Moscow (in Russian).
Merchuk, J. C.. 1986, Gas hold-up and liquid velocity in
a two-dimensional
air-lift reactor. Chem. Engng Sci. 41,
11-16.
Merchuk, J. C. and Stein, Y., 1981, Local hold-up and liquid
velocitv in air-lift reactors. A.f.CL.E. J. 27. 377-388.
Metkin, V-. P. and Sokolov, V. N., 1982, Hydraulic resistance
to flow of gas-liquid
mixtures having non-Newtonian
nronerties. 3. ADDS. Chem. USSR 55. 558-563.
Miyahara, T., Ha-maguchi, M., Sukeda Y. and Takahashi,
T., 1986, Sire of bubbles and liquid circulation in a bubble
column with a draft tube and-sieve plate. Can. J. them.
Engng 64,718-725.
Nicol, R. S. and Davidson,
J. F., 1988a, Gas hold-up in
circulating bubble columns. Chem. Engng Res. Des. 66,
152-158.
Nicol. R. S. and Davidson, J. F., 1988b, Effect of surfactants
on gas hold-up in circulating bubble columns. Chem.
Engng Res. Des. 66, 159-164.
Onken, U. and Weiland, P., 1980, Hydrodynamics
and mass
transfer in an airlift loop fermenter. Eur. J. Appl. Microbiol. Biotechnol. 10. 3 l-40.
Philip, J., Proctor, J. M., Niranjan, K. and Davidson, J. F.,
1990, Gas hold-up and liquid circulation in internal loop
reactors containing highly viscous Newtonian and nonNewtonian liquids. Chem. Engng Sci. 45. 651-664.
Popovi& M. K. and Robinson, C. W., 1984, Estimation of
some important
design parameters for non-Newtonian
liquids in pneumatically-agitated
fermenters, in Proceedings of the 34th Canudian Chemical Engineering Confirence, 31 September-3 October, Quebec City, pp. 2588263.
Popoviti, M K. and Robinson, C. W., 1988, Externalcirculation-loop airlift bioreactor: study of the liquid circulating

velocity

4035

velocity in highly viscous non-Newtonian


liquids. Biotechno!. Eiwngng 32, 301-312.
Popovih M. K. and Robinson, C. W., 1989, Mass transfer
studies of external-loop
airlifts and a bubble column.
A.1.Ch.E. J. 35, 393-405.
Roberts, J., 1979, Operation and predicted performance of
a recirculating bubbly liquid air-lift column, in Proceedings of the 7th Australian Conference on Chemical Engineering, 22-24 August, pp. 63-67.
Schiigerl, K., Locke, J. and Oels, U., 1977, Bubble column
bioreactors. tower bioreactors without mechanical agitation. Adv. Eiochem. Engng 7. l-84.
Sokolov, V. N. and Metkin, V. P., 1981, Turbulent flow of
pseudoplastic
liquids in tubes. J. Appl. Cheat. USSR 54,
1103-1107.
Vatai, G. and TekiC, M. N., 1986, Effect of pseudoplasticity
on hydrodynamic characteristics of airlift loop contactor,
Paper presented at the 2nd Conference
of European
Rheoloeists EURO-RHEO86.
Prauue.
Verlaan, &., Tramper,
J., van% R&
K. and Luyben,
K. Ch. A. M., 1986a. A hydrodynamic model for an airliftloop bioreactor with extemii
loop. Chem. Engng J. 33,
43-53.
Verlaan, P., Tramper,
J., van? Riet, K. and Luyben,
K. Ch. A. M., 1986b, Hydrodynamics
and axial dispersion
in an airlift-loop bioreactor with two and three-phase
flow. Paper 07 presented at the International Conference
on Bior&tor
Fluid Dynamics, Cambridge, Enghmd.
Wachi, S., Jones, A. G. and Elson, T. P., 1991. Flow dynamica in a draft-tube bubble column using various
liquids. Chem. Engng Sci. 46, 657-663.
Weiland, P. and Onken, U., 1981, Fluid dynamics and mass
transfer in airlift fermenter with external loop. Get. Chem.
Engng

4.42-50.

Young, M. A., Carbonell, R. G. and Ollis, D. F., 1991, Airlift


bioreactors: analysis of local two-phase hydrodynamics.
A.I.Ch.E. J. 37. 403-428.

You might also like