Pergamon

PII:

(.'heml~l En(tmeerm4 S~ wnce. Vo|. 52, Nos. 21 '22. pp. 3979..3992, 1997
( 1997 Elsevier Science Lid All nghts reser'eed
Printed m Great Britain
S0009-2509(97)00241-8
0(~)92509.97 $17(x)a-0(X)

Flow regime identification in gas-liquid

flow and three-phase fluidized beds
J.-P. Zhang, J.R. Grace,* N. Epstein and K.S. Lim
Department of Chemical Engineering, The University of British Columbia,
Vancouver, B.C., Canada V6T 1Z4
{Accepted 1 July 19971
Abstract--Experiments were carried out in an 82.6 mm diameter by 2 m high vertical column
with air and water as the gas and liquid phases. Three different types of solid particles (1.5 mm
glass beads, 4.5 mm glass beads and 1.2 mm steel shot) were employed to investigate the effects
of particle size and density on the flow regime transitions. Superficial gas velocities ranged from
0.0018 to 7.7 m/s, while the superficial liquid velocity was varied from 0 to 0.4 m/s. Criteria for
determining flow regime transitions, developed in air water two-phase flow based on bubble
frequency, Sauter mean bubble chord length and the time taken by a bubble to pass a given
point, were successfully applied to gas-liquid -solid three-phase fluidized beds. These quantities
were determined by a two-element conductivity probe located at the centre-line of the column
0.65 m above the distributor. It is shown that churn fow, bridging flow and annular flow occur
at high gas velocities in three-phase fluidized beds. as in gas- liquid two-phase flow. Flow regime
maps are presented for the air-water two-phase flow and for the three different three-phase
systems. In addition, empirical correlations are provided for the flow regime boundaries in the
three-phase systems investigated. V 1997 Elsevier Science Ltd
K e y w o r d x : Three-phase fluidization; flow regimes; transitions.

INTRODUCTION
Gas liquid flows and three-phase fluidized beds are
widely used in industry (Fan, 1989). The performance
of these systems differs according to which of several
flow regimes is present. Discrete bubble flow, dispersed bubble flow, coalesced bubble flow, slug flow,
churn flow, bridging flow and annular flow have been
observed in gas-liquid two-phase flow, but only the
lower gas velocity flow regimes have received attention in three-phase fluidized beds. Numerous studies
have been devoted to flow regime identification and
flow regime transitions in cocurrent gas-liquid upflow
(e.g. Nakazatomi et al., 1992; Monji, 1993) and in
three-phase fluidization (e.g. Fan et al., 1985: Jean and
Fan, 1987; Song et al., 1989). The objectives of this
work are to develop a measurement technique and
criteria that can be used to determine gas-liquid and
gas-liquid-solid flow regime transitions objectively,
to explore new flow regimes at high superficial gas
velocities and to delineate the boundaries between
flow regimes in gas-liquid flow and in three-phase
fluidized beds.
Most previous studies of flow regimes and their
transitions have been based on visual observations

*Corresponding author. Tel.: 00 1 604 822 3121; fax: 00

I 604 822 60()3.

[e.g. in gas-liquid systems: Griffith and Wallis (19611,

Taitel et a/. (I 980), Fernandes et al. (1983), Hasan and
Kabir (1992); in gas liquid-solid systems: Ermakova
et al. {1970), Mukherjee et al. (1974), Muroyama et al.
(1978), Fan et al. (19861, Song et al. (1989), Nacef et al.
(1992. 1996)]. Although visual observation provides
some information on the flow patterns, it is often
difficult to identify the flow regime transitions without
quantitative measurements, even in transparent columns, due to the relatively opaque nature of multiphase flow.
Pressure fluctuations are sometimes used to identify flow regimes (e.g. Tutu, 1982, 1984: Matsui, 1984,
1986: Fan et al., 1986; Kitano and lkeda. 1988), but
pressure signals are transmitted (with some attenuation) from great distances so that they are very difficult to interpret.
Various probes can also be used to determine the
local character of the multiphase flow and flow regimes can be inferred from these measurements. For
example, conductivity probes have been used in several studies (e.g. Mao and Dukler, 1989; Han and
Kim, 1990: Lee et al., 1990; Franca et al., 1991; Soria
and de Lasa, 1992). The transition criteria can be
based on various statistical parameters derived from
the data. For example, Vince and Lahey (1982) employed the variance of the probe output signal as an
indicator, while the shape of the probability density
function (PDF) was used by Jones and Zuber (1975)

3979

3980

J.-P. Zhang et al.

and integration of the PDF within a certain range by

Kelessidis and Dukler (1989). In other cases (e.g.
Barnea et al., 1980), raw signals have been used to
distinguish flow regimes.
Only Das and Pattanayak (1993, 1994)in a twophase system and Lee et al. (1990) in a three-phase
system have sought to link the flow regimes and their
transitions directly to the properties of the bubbles
and voids which characterize these regimes. However,
the scope of these investigations was quite limited. No
previous studies have applied consistent criteria based
on bubble/void characteristics over very broad ranges
of gas and liquid superficial velocities. A new method
is proposed below which is objective, based on the
multiphase characteristics of the system, covers very
broad ranges of operating conditions, and is applicable to both two- and three-phase systems.

~21
~e~lJ ~.- Overflow

Data
aoqutsition
system

.;~

""5
2~
SZ
ConducUvit probe

EXPERIMENTAl.APPARATLISAND MEASUREMENT
METHODS
The main component of the experimental apparatus consisted of a 0.0826 m diameter by 2 m high
Plexiglas column with a perforated gas-liquid distributor plate containing 62 2 mm diameter circular
holes (covered by a stainless steel screen). The region
below the distributor plate was a 0.5 m long gasliquid calming section packed with 13 mm ceramic
intalox saddles as shown in Fig. 1. A three-phase
separator, attached to the top of the column, allowed
experiments to be carried out for a very wide range of
operating conditions. Air and tap water were used as
the gas and liquid phases, respectively. The water
temperature was measured at the outlet, while gas and
liquid flow rates were measured by inlet rotameters.
The superficial gas velocities ranged from 0 to 7.7 m/s,
while the superficial liquid velocity was varied from
0 to 0.4 m,s. Three different types of particles, glass
beads (p~ ~ 2510 kg/m 3) of diameters 1.5 and 4.5 mm
as well as steel shot (p.~= 7510 kg/m 3) of diameter
1.2 mm, were used, all with sphericities close to unity.
Conductivity probes, with only the tips of two
0.31 mm diameter wires electrically exposed to the
multiphase mixture, were used to detect the gas phase.
A stainless steel tube of 5.0 mm outer diameter served
as the common ground electrode for both wires. Two
probes of different wire spacings were used. The one
with the smaller wire interval (spacing = 1.1 mm) was
used in the discrete and dispersed bubble flow regimes, i.e. at relatively low gas velocities, while the
other (spacing = 4.0 mm) was utilized for the larger
bubbles/voids that occurred at higher gas velocities.
Each of these probes was inserted into the column
horizontally so that the ends of the two wires were
aligned vertically along the axis of the column, 0.65 m
above the distributor. The bubble frequency,.L was
obtained by counting the number of pulses within
a given time. The bubble time, t~, is defined as the
duration of the pulse, i.e. the time during which the
probe is immersed in the bubble. This parameter is
a function of the bubble size, bubble velocity, bubble

Disb-ibutor

.~'i:

Liquid

V1
Gas

Fig. 1. Schematic diagram of the experimental apparatus.

shape, position where the probe pierces the bubble,

and the angle of travel relative to the vertical
direction.
The time-average local gas holdup, %, was calculated from
1

'go = ~

i~_.l ti.

(1)

With a two-element probe, the bubble velocity, Ub,

and chord length, Ih, can be computed if the vertical
distance, L, and the time shift, "r, between the two
elements are known, providing all bubbles rise vertically, i.e.
L
Uh = (2)
r
In = Ub x ti.

(3)

A logic circuit was set up to select only signals caused

by bubbles which are travelling upwards (Zhang,
1996; Zhang et al., 1997).
CRITERIA FOR FLOW REGIMETRANSITIONS
The experimental criteria adopted for flow regime
transitions, based on measured bubble characteristics,
were developed in an air-water two-phase flow after
considering various alternative criteria. These criteria
have also been shown to be applicable for gas-liquidsolid three-phase fluidized beds. Details follow together with typical results. Other confirmatory results
are given elsewhere (Zhang, 1996). The criteria are
fully consistent with visual observations.

3981

|:low regime identification

between dispersed bubble flow and slug flow. However, this transition criterion may be difficult to apply
when discrete bubble flow exists only within a narrow
gas velocity range.

Transition between discrete bubble flow and coalesced

bubble flow reqimes
At low liquid flow rates, when small amounts of gas
are introduced into the column, small bubbles appear.
The bubble size distribution is narrow and strongly
dependent on the gas distributor. The number of
bubbles and the gas holdup are also small. Bubbles do
not have sufficient time to coalesce since their separation distance is large compared to their size, and the
regime is referred to as discrete bubble flow. With
increasing gas velocity, larger bubbles of wider size
distribution are encountered. As their population increases, the distance between individual bubbles decreases. Some bubbles coalesce as they ascend. The
overall behavior of the multi-phase mixture is then
categorized as coalesced bubble flow.
It was found that in the discrete bubble flow regime,
bubble frequency increased linearly with gas velocity,
with a proportionality constant of 545 m - ~. The frequency deviated from this linear relationship when
the transition to coalesced bubble flow took place.
Hence, the discrete coalesced bubble flow transition
can be obtained from a plot of bubble frequency
versus superficial gas velocity, corresponding to the
point where the bubble frequency deviates significantly from a linear relationship. Figure 2 displays
some typical results for the air water 1.5 mm glass
bead system showing this transition.
The same discrete coalesced flow transition criterion can also be applied to flow regime transitions at
high ~_'~, i.e. to the transition between dispersed
bubble flow [described below) and coalesced bubble
flow or. if the coalesced bubble ttow regime is absent,

Transition between coalesced bubble.flow and slug flow

re,qime.~
As the gas velocity is increased further, bubbles
become larger and more elongated, and some bubble
cross-sectional dimensions approach the diameter of
the column. The appearance of Taylor (bullet-shaped)
bubbles indicates that the flow pattern has changed to
slug flow. In the slug flow regime, many small bubbles
with diameter approximately equal to 4 mm are observed in the liquid plugs which follow the Taylor
bubbles. Both the "raylor bubbles and the trailing
small bubbles are detected by the conductivity probe.
The small bubbles result in a small arithmetic mean
bubble chord length. In order to detect the point at
which Taylor bubbles begin to appear in the column,
the Sauter mean bubble chord length,
?d

l~.,~. 2/~, ~/~,

i-I

~4~

t-I

was used in this study since it gives more weight to

large bubbles than to small ones. The transition between coalesced bubble flow and slug flow was set as
the point at which the Sauter mean bubble chord
length, /b.,..,, measured at the center of the column,
reached the column diameter. Figure 3 shows experimental results identifying this transition for the
4.5 mm glass bead air- water system.

100

//

f = 545 Ug----./
M

90 -

80

....... . . . . . . . . . .

~ v ..........

:.

70 Transition Points//

;>T,

60

Ut, m/s

.~-/~

//

50
~'~

"~

JJ

20-"

d~
r"

0.00

0.05

.- 0.0638

/ n~Sz
o ~ - ~_: : e.

0.10

..~.1

o-00729

.....

o.~---~:Jv--~

0.15

0.20

0.0154
0.0184
0.0273
0.0455

~-q~/~-~-

N~"-"*,

....z~/'A'lx/

;/^.A/'
,~Z~-:,

10- ~

-e

//
1A

"

~ . _ A//

l:

40-

..

0.25

0.30

0.35

0.40

- ~

o 1oo

---v

0.155

0.45

0.50

0.55

Superficial Gas Velocity, m/s

Fig. 2. Transition between discrete and coalesced bubble flow regime in a three-phase fluidized bed
containing 1.5 mm glass beads at different superticial liquid velocities.

J.-P. Zhang et al.

3982

o~
,..]

0.1

IJo//',,J
. J J/

-.-0.0364

/
^ / ...-v"
/ ~ - ~ X / / v

/
0.01

0.0455
- - A - - 0.0546
--e--

v./v

--~1,-- 0.0729
- - o - - 0.100
- - o - - 0.128

0.001
0.001

. . . . . . . .

0.01

. . . . . . . .

--A--

0.155

--v--

0.218

0.1

Superficial Gas Velocity, rn/s

Fig. 3. Transition between coalesced bubble flow and slug flow in a three-phase tluidized bed containing
4.5 mm glass beads at various superficial liquid velocities.

Transition between slu~ flow and churn flow

The transition between slug and churn flow can be
taken to occur when the gas holdup within the liquid
plugs corresponds to a maximum volumetric packing
of bubbles (Brauner and Barnea, 1986; Mao and
Dukler, 1989). Beyond this point, the bubble frequency decreases with increasing Ug due to coalescence of the small trailing bubbles within the liquid
plugs between the Taylor bubbles. Hence, the maximum bubble frequency in a plot of bubble frequency
vs superficial gas velocity can be used to distinguish
between slug flow and churn flow. Figure 4 shows
experimental results for this transition with 1.2 mm
steel shot as particles.

Transition between churn flow and bridging flow

Bridging flow occurs when liquid spans the core
region of the column linking the liquid film occupying
the annular region of the column in any cross-section.
In churn flow, some distorted Taylor bubbles were
observed. As the gas velocity was increased, the distorted Taylor bubbles became elongated. Some successive Taylor bubbles coalesced, forming larger
bubbles, while the liquid plugs between successive
Taylor bubbles became thinner. The measured bubble
frequency continued to decrease (as shown in Fig. 4),
and the average bubble chord length showed a sharp
increase with increasing gas velocity. This abrupt
change in slope can be taken to denote the onset of
bridging flow. The variation of the average bubble
time with the gas velocity showed the same tendency
as the average bubble chord length since the latter is
related to the bubble time. The average bubble chord

length is defined as
1

= "M ,~=1(ti Ubi).

(5)

If Ubi is constant and equal to Uh, then eq. (5) simplifies to

7b = Ufii.

(6)

Equation (6) indicates that the average bubble

chord length is proportional to the average bubble
time, providing that the bubble velocities are the
same. Because the measurement of Tb requires a twoelement probe, while i~ requires only a single element,
the average bubble time was used to determine the
transition between churn flow and bridging flow
throughout this work. This avoids problems associated with two-element probes, such as errors resulting from bubbles rising obliquely. Figure 5 shows the
variation of the average bubble time with respect to
U~ for the air-water-l.5 mm glass bead system. The
transition to bridging flow is quite clear in this figure,
and similar results were obtained for the other systems studied.

Transition between bridging flow and annular flow

Annular flow is characterized by a continuous gas
region at the core surrounded by a continuous liquid
region at the wall. In some cases, small bubbles are
found in the liquid annular region. At the center of the
column, in the annular flow regime, bubble frequency
or liquid bridge frequency should be zero, and measured bubble chord length should approach infinity.

Flow regime identification

180

3983

Slug/Churn

160 -

U e , m/s

--m-- 0.0364

140-.

--e--0.0455
--&--0.0546

120 o

ti
t.~
-~
d3

Transitions\

-v-

,00-

Y\

0.0638

? \

.V

- - O i 0'155

-,,-o.2oo

0 - - 0 . ~v~

Onset of

"o~

/
/,,-~

"

o o

..X2/ v /
.

1o ^ /v
.

a=~

i,lr~.

v..w~l

"~\~I~\ l/

,~-~_/

0.01

L--r,ow

o.O

~ /

200

L2

oo-

,0-

v /

--n--OlO0

80-

~/V
vv-v

0.1

'

"

" - ' " ' ' '

10

Superficial Gas Velocity, m/s

Fig. 4. Local bubble frequency at axis of column versus superficial gas velocity showing transition between
slug flow and churn flow and onset of annular flow for three-phase fluidized bed containing 1.2 mm steel
shot at different superficial liquid velocities.

1000
zx

. ~11 ~ 1 1 = ' ~ " l -~IL=. II,,

t/'3

c-q

100-

,~//ffffff~

~D
I-

_~

ra

----------t~-o
X~-_____A--

w---"'J_ _./o*:~Ag,=-[~o~.~.o~';~.,w~Rmlk
~
_~v

gx--Lx-'-,~, / z ~ x
A

'
--&-- 0.0273

V/v

V~ 10 ~V____----V--V--~v~V'V~v/V V.V/V

--v--. 0.o455
- , - 0.0638
_ n _ 0.0728
- o - 0.0729
--A-- 0.100
--V--0.155

<

. . . . . .

0.01

0.1

Ut m/s

10

Superficial Gas Velocity, m/s

Fig. 5. Average bubble time versus superficial gas velocity showing transition between churn flow and
bridging flow for a three-phase fluidized bed containing 1.5 mm glass beads at different superficial liquid
velocities.

The former was used to determine the onset of annular flow in this work. Since the probe was always on
the axis of the column, the transition velocity between
bridging flow and annular flow could be obtained

experimentally by extrapolating the measured bubble

frequency to zero in a plot o f f v s Ug. Results for one of
the systems studied, the 1.2 mm steel shot system, are
shown in Fig. 4.

J.-P. Zhang et al.

3984
'

3.0

2.5.

Ug, n'ffs
0

2.0.

~O 0/

0.t0

/ o

0.070 o . . ~

' 6o
t~

o~ I
/o/#~

00,0
1.5.

0.032 .

////, /

"~

1.0-

(/3
0

0.5-

0.0050

~ g ~ - " ' - "

.e
Discrete / Dispersed Transitiom
0.0
0001

'

'

'

'

'

'

' ' 1

0.01

"

'

'

'

'

'

' ' ]

0.1

Superficial Liquid Velocity, Ue, m/s

Fig. 6. Dimensionless standard deviation of bubble time plotted against superficial liquid velocity showing
transition between discrete bubble flow and dispersed bubble flow for air water two-phase flow at different
superficial gas velocities.

Transition between discrete or coalesced bubble flow

and dispersed bubble flow
As liquid flow is increased at a constant low gas
flow rate, bubbles become smaller and their size distributions narrower. This regime is the dispersed bubble
flow regime. Unlike discrete bubble flow, the bubble
behavior in this regime is relatively independent of the
gas distributor.
It was found that both the average and standard
deviation of Ib (or t~) decreased with increasing liquid
flow in the discrete and dispersed bubble flow regimes.
However, the ratio of standard deviation of ti to the
average bubble time, i.e. a,/fi, increased for both discrete and coalesced bubble flow, but decreased for
dispersed bubble flow. The transition from the discrete (or the coalesced bubble flow regime when the
discrete regime does not exist) to the dispersed bubble
flow regime was assumed to correspond to the point
where o,/'?~ begins to fall sharply. This parameter is
equal to ad~, provided that uniform bubble velocity
can be assumed. Figure 6 shows the variation of a,/i,
with respect to U~ for the air-water system. Similar
curves were obtained for all three three-phase systems
(Zhang, 1996).
FLOW REGIME MAPS
The experimental criteria for determination of flow
regime transitions, applicable to both gas-liquid flow
and three-phase fluidized beds, are summarized in
Fig. 7. The flow regime boundaries based on conductivity measurements for air-water two-phase flow are
plotted in Fig. 8. Note that this map, like those below

for the three-phase systems studied, is applicable only

for the size of column and type of distributor studied.
In addition, the map is strictly only applicable at the
(0.65 m) height of the measurement probe, although in
practice visual observations suggest that regime
transitions generally occur throughout the height of
the column at nearly the same values of gas and liquid
superficial velocities.
It can be seen that the discrete bubble flow regime
predominates at low gas and liquid velocities, while
the dispersed bubble flow regime is encountered at
higher liquid velocities. Both the discrete and dispersed bubble flow regimes are characterized by small
bubbles with relatively uniform size distributions. The
bubbles in the dispersed bubble flow regime, however,
are smaller and more uniform as a result of liquid
turbulence, while the bubble size and its distribution
in the discrete bubble flow regime may be influenced
by the gas distributor. The coalesced bubble flow
regime in Fig. 8 is confined to a very narrow gas
velocity range due to the small column diameter used
in this study. The slug flow regime spans a wide range
of gas velocities. At low U~, the onset of slug flow is
almost independent of U~, while at high liquid velocities the transition from dispersed bubble flow to
slug flow is a function of Ut. With increasing superficial liquid velocity, the transition gas velocity from
dispersed bubble flow to slug flow increases. The
transitions among the slug, churn, bridging and annular flow regimes are nearly independent of Uj. For the
conditions studied in this work, the churn flow regime
exists only within a small range of Ug.

3985

Flow regime identification

U e increased
Dispersed
Bubble Flow

Ug increased

Discrete

Coalesced

Bubble Flow

Bubble Flow

Slug
Flow

Churn
Flow

/ / I H
a,/i- '

log (tb.,O

log (U t)

Us

k~g (i',)

log (U g)

Bridging
Flow

log (U g)

Annular
Flow

lOg ~ |1

log (U g)

Fig. 7. Schematic diagram of flow regimes and their transition criteria.

lo0

Legend
Dispersedbubbleflow

[ ~

1st criterion in Fig. 7

2nd criterion in Fig. 7
3rd criterion in Fig. 7

10 "I

4th criterion in Fig. 7

5th criterion in Fig. 7
6th criterion in Fig. 7
"3

10.2 -:

r.~

10 -3

........
10-3

lff 2

.... i
10"1

........

i
100

. . . . .

l,,

101

Superficial Gas Velocity, Ug, m/s

Fig. 8. Flow regime map for air water two-phase system.

A flow regime map for the 1.5mm glass bead

three-phase fluidized bed is presented in Fig. 9. In this
case the discrete bubble flow regime does not exist. It
has been found (Henriksen and Ostergaard, 1974:
Darton, 1985) that bubble coalescence is more intensive in air-water systems containing glass beads less
than 2.5 mm in diameter than in the corresponding
gas-liquid system, since the particles increase both the
emulsion phase viscosity and density. Figure 9 indicates that a dispersed bubble flow regime exists at
high Ui. This is because the solids holdup of the
three-phase fluidized bed decreases towards zero as
Ut is increased, with the system then approaching the
corresponding gas-liquid two-phase limit.
Figure 10 presents a flow regime map for the
4.5 mm glass bead three-phase fluidized bed. A discrete bubble flow regime is observed, and the domain
of the coalesced bubble flow regime is decreased due

to the large particles that break up bubbles, it has

been found (Ostergaard, 1971; Lee eta/., 1974) that
dispersed bubbles of uniform small size occur for glass
beads larger than 2.5 mm in air water systems. The
findings of the present study are in good agreement
with previous investigations. The coalesced bubble
flow regime exists for U~=0.02-0.04m/s and
Ut = 0.02-.0.05 ms. In this regime, the gas passes
through the liquid -solid mixture as bubbles of irregular shape and large size. This differs from the 1.5 mm
glass bead system where bubbles are spherical or
spherical-caps in the coalesced bubble flow regime.
Figure 11 shows the corresponding flow regime
map for the 1.2 mm steel shot three-phase fluidized
bed system. Both the discrete bubble flow and coalesced bubble flow regimes are present. Since the density
of the steel shot, 7510 kg/m 3, is much greater than
that of glass beads, the particles are able to penetrate

3986

J.-P. Zhang et al.

10 0

10 "1

10 .2

'E
r#3
10 .3

0
0

10 .3

10 .2

10 1

10 0

10 ~

Superficial Gas Velocity, Ug, m/s

Fig. 9. Flow regime map for air water-l.5 mm glass bead three-phase fluidized bed. Shaded portions refer
to pre-fluidized and post-fluidized regimes vis fi ,,is the motion of the particles. Symbols as in Fig. 8.
supplemented by HI') based on pressure gradient measurements and () based on visual observation.
100

10 -1

10 -2

rJ3
10-3:

O,
0

10-3

10-2

10-1

10o

101

Superficial Gas Velocity, Ug, m/s

Fig 10. Flow regime map for air- water-4.5 mm glass bead three-phase fluidized bed. Shading as in Fig. 9:
symbols as in Figs 8 and 9.

and break bubbles at lower gas and liquid superficial

velocities, causing the fluidized bed to operate in the
discrete bubble flow regime. As U~ is increased further
at low Ut, coalesced bubble flow is encountered. At
high liquid velocities, coalesced bubble flow is also
found between dispersed bubble flow and slug flow at
relatively high gas velocities. This may be related to

the solids holdup. Since the particle density is larger

than that of the glass beads, the solids holdup in the
dispersed bubble flow regime is expected to be higher
for the same operating conditions. The higher emulsion phase density and viscosity cause gas bubbles to
coalesce and form Taylor bubbles at relatively high
gas velocities.

Flow regime identification

3987

10 0

g
10 -I

10 .2

r/3
10 .3

0
0

10 .3

10 .2

10 -~

10 0

101

Superficial Gas Velocity, Ug, m/s

Fig. 11. Flow regime map for air--water-l.2 mm steel shot three-phase fluidized bed. Shading as in Fig. 9:
symbols as in Figs 8 and 9.
Our findings show that churn flow, bridging flow
and annular flow are observed in three-phase
fluidized beds, as in two-phase gas-liquid flow, at high
gas velocities. Since the glass beads are already entrained from the column before the transition from
bridging to annular flow occurs, the boundaries between the bridging and the annular flow regimes in
Figs 9 and l0 are similar to those for the corresponding gas-liquid two-phase system. On the other hand,
for the 1.2 mm steel shot three-phase system, threephase annular flow is observed, with the annular
region containing liquid, particles and small
(db= 2-4 mm) gas bubbles.

experimental values. The scatter is no doubt due, at

least in part, to the different (often subjective) criteria
used to distinguish regimes in previous studies.
At low liquid velocities, the gas velocity for onset of
the slug flow regime increases with both particle size
and density in the three-phase systems. At high Ut, the
onset of the slug flow regime does not differ significantly for the systems tested. For low superficial
liquid velocity, with

Ut <~8"S2(PIdPt
\ Pl /

('qD)312Arl"41O\dpJ
(SI units)

(8)

C O R R E L A T I O N S O F F L O W REGIME TRANSITIONS

The following empirical correlation is proposed

to predict the transition liquid velocity for the onset
of the dispersed bubble flow regime in three-phase
fluidization:

Values of U~ from eq. (7) are plotted against the

experimental data in Fig. 12. The average absolute
percentage deviation is 8.0% for the 44 sets of experimental results from the current study. Many earlier
investigators have also studied the transition between
the coalesced bubble flow and dispersed bubble flow
regimes using a range of particle sizes and densities.
The parity prediction also goes through the middle of
the 102 data points from the literature, but there is
considerable scatter, with an average absolute percentage deviation of 70.3% between the predicted and

the transition gas velocity for the onset of slug flow in

a three-phase system was correlated by
--=U~ 6.9710

5Re-9~Ar8s{D~l
,

!9,

For high superficial liquid velocity, when U~ is greater

than the right-hand side of eq. (8), the transition gas
velocity for the onset of slug flow was correlated by
.Uuut. . 20'3r'~3ZArt-861(~)
.
- 3 1 8 ' m

(10)

Comparisons between the experimental data and the

predictions ofeqs (9) and (10) are presented in Figs 13
and 14, respectively. The average absolute percentage
deviation between eq. (9) and the experimental data of
the present study is 6.7%, based on 20 sets of data.
For the 8 sets of available literature data obtained by

3988

J.-P. Zhang eta/.

0
0

0.1

DO

V
D~. of this ~.ork

n~

I.=e1al. (1990)

ltad Ii~,x~ (1988)

M alsutl'a ~ld Fan (1984)

N a c , ef et al. (1992)

0.01

0.01

Fan et al. (1986)

t at (1970)

Faneta.l.(1985)

M ~ a

I'~.ef et al. (1996)

S c m and de La~ (1992)

et al. (1978)

0.1

Ext:ea'imentalLiquid Transition Velocity,m/s

Fig. 12. Comparison between predictions of eq. (7) and experimental data for onset of dispersed bubble
flow in three-phase fluidization.

0.1

".,E

1.5 rran glass beads

! 4,mm s

1.2 f~r~ steel shot

Matmawa and Fan (1984)
0.01

.
0.01

.
.
. . . . . .

0'11

Expemmntal Transition Gas Velocity, m/s

Fig. 13. Comparison between predictions of eq. (9) and experimental data for transition to slug flow in
three-phase fluidized beds at low liquid velocites.

visualization, the transition gas velocity is overestimated with a m e a n deviation of 52.3%. The
average deviation between eq. (10) a n d o u r 19 sets of
data is 12.4%. F o r 14 sets of data from the literature,
the m e a n deviation is 22.3%.

The b o u n d a r y between the slug a n d c h u r n flow

regimes is correlated by

Flow regime identification

1.5 mm glass beads

4.5 mm glass beads

1.2 nun steel shot

..y

~7 Matsuura and Fan (I 984)

O

3989

Muroyama et al. (1978)

_~,

.So .

0.1

y"

I--

0.01
0.01

. . . . . . .

011

. . . . . . .

Expema~tal Transition Gas Vdocity, rrfs

Fig. 14. Comparison between predictions of eq. (10) and experimenlal data for transition to slug flow in
three-phase fluidized beds at high liquid velocities.

a;

1.5 , m i glass beads

4.5 iiim glass b e a d s

oo.~-

.f

1.2 ii=ii steel shot

~

0.1

o
~- 0.01

F-

0.001
0.001

. . . . .

. . . . .

'1

0.1

0.01

Extxrimental Transition Liquid Velocity,m/s

Fig. 15. Comparison between predictions of eq. 111) and experimental data for onset of churn flow in
three-phase systems.

This equation is based on 30 sets of data for the

three-phase systems of the present study with an average absolute percentage deviation of 14.3%. Results
are plotted in Fig. 15. No earlier data are available for
the slug/churn flow transition in three-phase systems.

The following equation correlates the boundary

between the churn and bridging flow regimes:

U. = 219Refl.2aArO.36~(p~ ~ o.*.~,~
UI

\PC

(12)

3990

J.-P. Zhang

r~o

et al.

1.5 mm glass beads

4.5 n.ml___glass beads

01

o=
[~ 0.01

0.001

0.001

0.01

0.1

Experimental Transition Liquid Velocity, m/s

Fig. 16. Comparison bct~vcen predictions of eq. (12) and experimental data for onset of bridging flow in
three-phase systems.

Two-phase

_ Q _ 1.5 mm Glass Beads

r./3

.A...

4.5 mm Glass Beads

_.7..

1.2 mm Steel Shot

Annular Flow
".-VI

0.1

,-]

Bridging Flow
0.01

q
I

o
0

0.001

10

Superficial Gas Velocity, Ug, nVs

Fig. 17. Boundaries between bridging and annular flow regimes.
Figure 16 compares the experimental values with predictions from this equation. The average absolute
percentage deviation for the 30 sets of three-phase
data of the present study is 11.1%. No other data have
been found in the literature for this transition.
Figure 17 presents the transition between bridging
flow and annular flow. For the two-phase system, the
transition gas velocities decrease with increasing U~ at
high liquid flow rates, but remain almost constant at

lower Ut. For the glass beads, before this transition

occurs, the particles have all been entrained from the
column, so the bridging-annular transitions are virtually identical to those for the corresponding gas
liquid system. For the 1.2 mm steel shot. however, the
solid particles cause a decrease in the transition gas
velocities. Most experimental data for this transition
obtained in the present study correspond to absence
of particles. There are insufficient data to construct an

3991

Flow regime identification

empirical correlation for this transition in three-phase
systems.
CONCLUSIONS

A new set of criteria based on conductivity probe

measurements of bubble characteristics, summarized
in Fig. 7, was developed to determine flow regime
transitions in upward air-water two-phase flow and
then applied successfully to air-water-solid threephase fluidized beds. Flow regimes at high gas velocities, not previously studied in three-phase systems,
were explored. Thus complete flow regime maps were
obtained for cocurrent upward three-phase systems.
The flow regimes in a three-phase fluidized bed are
qualitatively the same as for air--water systems in the
same column, i.e. dispersed bubble flow, discrete
bubble flow, coalesced bubble flow, slug flow, churn
flow, bridging flow and annular flow. Empirical
correlations are developed to predict most of the
flow regime boundaries in the three-phase systems
investigated.
NOTATION

Ar I

D
db
dp
Frg
Frgo

liquid Archimedes number (dZpth(p~ - Pt)g/l~

column diameter, m
bubble diameter, m
equivolume sphere diameter of particles, m
gas Froude number [U2/(gdp)]
gas Froude number based on column diameter, [U ~/(gD)]
local bubble or liquid bridging frequency,
S -1

L
lb

lb. rs
M
Ret

/tot

Ub
U,
UI

gravitational acceleration, m/s 2

distance between two elements of conductivity probe, m
local individual bubble chord length, m
Sauter mean bubble chord length, m
local number average bubble chord length, m
number of bubbles observed during observation period
particle Reynolds number based on superficial liquid velocity, (Uidppt,.%)
duration of conductivity probe in gas bubble
or bubble time, s
average bubble time, s
total test time, s
local number average bubble velocity, m/s
superficial gas velocity, m's
superficial liquid velocity, m/s

Greek letters
%
gas holdup, m3/m 3
#t
liquid viscosity, N s/m 2
p~
liquid phase density, kg/m 3
p.~
particle density, kg/m 3
a~
standard deviation of bubble chord length, m
a,
standard deviation of bubble time, s
time shift between two elements of conductivity probe, s

REFERENCES

Barnea, D., Shohan, O. and Taitel, Y. (1980) Flow

pattern characterization in two phase flow by electrical conductance probe. Int. Multiphase Flow 6,
387-397.
Brauner, N. and Barnea, D. (1986) Slug/churn
transition in upward gas-liquid flow. Chem. Engn#
Sci. 41, 159-163.
Darton, R. C. (1985) The physical behaviour of threephase fluidized beds. In Fluidization, 2nd edn., eds
J. F. Davidson, R. Clift and D. Harrison, Chap. 15,
pp. 495-528. Academic Press, London.
Das, R. K. and Pattanayak, S. (1993) Electrical impedance method for flow regime identification in vertical upward gas-liquid two-phase flow. Meas. Sci.
Technol. 4, 1457-1463.
Das, R. K. and Pattanayak, S. (1994) Bubble to slug
flow transition in vertical upward two-phase flow
through narrow tubes. Chem. Enyny Sci. 49,
2163-2172.
Ermakova, A., Ziganskin, G. K. and Slin'ko, M. G.
(1970) Hydrodynamics of a gas-liquid reactor with
a fluidized bed of solid matter. Theor. Found. Chem.
Engng 4, 84-89.
Fan, L.-S. (19891 Gas-Liquid-Solid Fluidization
Engineeriny. Butterworth, Boston, MA.
Fan, L.-S., Matsuura, A. and Chern, S.-H. (1985)
Hydrodynamic characteristics of a gas-liquid-solid
fluidized bed containing a binary mixture of particles. A.I.Ch.E.J. 31, 1801-1810.
Fan, L.-S., Satija, S. and Wisecarver, K. (1986) Pressure fluctuation measurements and flow regime
transitions in gas-liquid-solid fluidized beds.
A.I.Ch.E.J. 32, 338--340.
Fernandes, R. C., Semiat, R. and Dukler, A. E. (1983)
Hydrodynamic model for gas-liquid slug flow in
vertical tubes. A.I.Ch.E.J. 29, 981-989.
Franca, F., Acikgoz, M., Lahey, R. T. and Clausse, A.
(19911 The use of fractal techniques for flow regime
identification. Int. J. Multiphase Flow 17, 545-552.
Griffith, P. and Wallis, G. B. (1961) Two-phase slug
flow. J. lteat Tran.sfer 83, 307-320.
Hart, J. H. and Kim, S. D. (19901 Radial dispersion
and bubble characteristics in three-phase fluidized
beds. Chem. Engng Commun. 94, 9-26.
Hasan. A. R. and Kabir. C. S. (1992) Two-phase flow
in vertical and inclined annuli. Int. J. Multiphase
Flow 18, 279-293.
Henriksen, H. K. and Ostergaard, K. (1974) On the
mechanism of break-up of large bubble in liquids
and three-phase fluidised beds. Chem. Engnq Sci. 29.
626 629.
Jean, R.-H. and Fan, L.-S. (1987) On the particle
terminal velocity in a gas liquid medium with
liquid as the continuous phase. Can. J. Chem.
En,qnq 65, 881-886.
Jones, O. C. and Zuber, N. (1975) The interrelation
between void fraction fluctuations and flow patterns in two-phase flow. Int. J. Multiphase Flow 2,
273-306.
Kelessidis, V. C. and Dukler, A. E. (1989) Modeling
flow pattern transitions for upward gas liquid flow
in vertical concentric and eccentric annuli. Int. J.
Multiphase Flow 15, 173 191.
Kitano, K. and Ikeda, K. (1988) Flow regimes of
three-phase fluidized beds. Proceedings of the Asian
Con/erence on Fluidized-Bed and Three-Phase

3992

J.-P. Zhang et al.

Reactors, eds K. Yoshida and S. Morooka, pp.

361-367. Sanjo-Kaikan, Tokyo.
Lee, J. C., Sherrard, A. J. and Buckley, P. S. (1974l
Optimum particle size in three-phase fluidized bed
reactors. Fluidization and Its Applications, ed. H.
Angelino, pp. 407.-416. Cepadues-Editions, Toulouse.
Lee, S. L. P., Soria A. and de Lasa, H. I. (1990)
Evolution of bubble length distributions in threephase fluidized beds. A.I.Ch.E.J. 36, 1763 -1767.
Mao, Z.-S. and Dukler, A. E. (1989) An experimental
study of gas--liquid slug flow experiments in fluids.
Exp. Fluids 8, 169-182.
Matsui, G. (1984) Identification of flow regimes in
vertical gas-liquid two-phase flow using differential
pressure fluctuations. Int. J. Multiphase Flow 10.
71 I- 720.
Matsui, G. (1986) Automatic identification of flow
regimes in vertical two-phase flow using differential
pressure fluctuations. Nucl. Eng. Des. 95, 221 231.
Matsuura, A. and Fan, L.-S. (1984) Distribution of
bubble properties in a gas liquid-solid fluidized
bed. A.I.Ch.E. J. 30, 894-903.
Monji, H. (1993) Transition mechanism from bubble
flow to slug flow in a riser. Fluid Dyn. Res. l l,
61--74.
Mukherjee, R. N., Bhattacharya, P. and Taraphdar,
D. K. (1974) Studies on the dynamics of three-phase
fluidization. Fluidization and Its Applications, ed.
H. Angelino, pp. 372379. Cepadues-Editions,
Toulouse.
Muroyama, K., Hashimoto, K. and Kawabata, T.
(1978) Axial liquid mixing in three-phase fluidized
beds. Kagaku Kogaku Ronbunshu 4, 622-628 (in
Japanese).
Nacef, S., Poncin, S. and Wild, G. (1996) Onset of
fluidisation and flow regimes in gas-liquid solid
fluidisation--influence of the coalescence behaviour of the liquid. Fluidization VIII. eds J.-F. Large
and C. Laguerie, pp. 631-638. Engineering Foundation, New York.

Nacef, S., Wild, G. and Laurent, A. (1992) Scale effects

in gas liquid-solid fluidization. Int. (_?hem. Enqng
32, pp. 51 61.
Nakazatomi, M., Shimizu, H., Miyake, G. and
Sekoguchi, K. (1992) Rising characteristics of
a single measure of gas slug in stagnant liquid (effect
of pressure). JSME Int. J., Ser. II 35, 388-394.
Ostergaard, K. (1971) Three-phase fluidization. In
Fluidization, eds J. F. Davidson and D. Harrison,
Chap. 18, pp. 751-782. Academic Press, London.
Song, G.-H., Bavarian, ['. and Fan, L.-S. (1989)
Hydrodynamics of three-phase fluidized bed containing cylindrical hydrotreating catalysts. Can. J.
Chem. Engn.q 67, 265 275.
Soda, A. and de Lasa, H. (1992) Kinematic waves and
flow patterns in bubble columns and three-phase
fluidized beds. (-'hem. Engn,q Sci. 47, 3403-3410.
Taitel. Y., Barnea, D. and Dukler. A. E. (1980) Modelling flow pattern transitions for steady upward
gas-liquid flow in vertical tubes. A.I.Ch.E.J. 26,
345-354.
Tutu, N. K. (1982) Pressure fluctuations and flow
pattern recognition in vertical two phase gas-liquid
flows. Int. J. Multiphase Flow 8, 443-447.
Tutu, N. K. (1984) Pressure drop fluctuations and
bubble-slug transition in vertical two phase
gas-liquid flow. Int. J. Multipha.~e Flow 10,
211-216.
Vincc, M. A. and Lahey, Jr. R. T. (1982) On the
development of an objective flow regime indicator.
Int. J. Multipha,w Flow 8, 93-124.
Zhang, J.-P. (1996) Bubble columns and three-phase
fluidized beds: flow regimes and bubble characteristics. Ph.D. Thesis, University of British Columbia,
Vancouver, Canada.
Zhang, J.-P., Epstein, N., Grace, J. R. and Lira, K.-S.
(1997) Experimental study of bubble characteristics
in vertical gas liquid two-phase upflow using
a conductivity probe. ASME Fluids Engineering
Division Summer Meeting, Vancouver, Canada.