Journal of Chemical Technology and Biotechnology J Chem Technol Biotechnol 80:545–552 (2005)

DOI: 10.1002/jctb.1205

Drop size distribution in highly concentrated

liquid–liquid dispersions using a light back
scattering method
J Lovick,1 AA Mouza,2 SV Paras,2 GJ Lye3 and P Angeli1∗
1 Department of Chemical Engineering, University College London, Torrington Place, London, WC1E 7JE, UK
2 Department of Chemical Engineering, Aristotle University of Thessaloniki, Univ Box 455, GR 541 24, Thessaloniki, Greece
3 Department of Biochemical Engineering, University College London, Torrington Place, London, WC1E 7JE, UK

Abstract: New data are presented on drop size distribution at high dispersed phase fractions of organic-
in-water mixtures, obtained with a light back scattering technique (3 Dimensional Optical Reflectance
Measurement technique, 3D ORM). The 3D ORM technique, which provides fast, in-situ and on-line
drop distribution measurements even at high concentrations of the dispersed phase, is validated using
an endoscope attached to a high-speed video recorder. The two techniques compared favourably when
used in a dispersion of oil (density (ρ) = 828 kg m−3 , viscosity (µ) = 5.5 mPa s, interfacial tension (σi ) =
44.7 mN m−1 ) in water for a range of 5–10% dispersed phase fractions. Data obtained with the ORM
instrument for dispersed phase fractions up to 60% and impeller speeds 350–550 rpm showed a decrease
in the maximum and the Sauter mean drop diameters with increasing impeller speed. Phase fractions did
not seem to significantly affect drop size. Both techniques showed that drop size distributions could be
fitted by the log-normal distribution.
 2005 Society of Chemical Industry

Keywords: liquid–liquid dispersions; drop size; stirred vessels

1 INTRODUCTION but also decrease turbulence intensity) correlations in

The mixing of immiscible liquids is commonly encoun- the form of eqn (2) have been suggested for dmax :
tered in process industries, with many applications
such as extraction, multiphase (bio)reactions and dmax /Di = c1 (1 + c2 φ)WeT −0.6 (2)
suspension polymerisation. In these processes drop
diameter distribution is an important parameter as it where φ is the volume fraction of the dispersed phase
defines the available interfacial area for mass transfer. and c1 and c2 are constants.
Drop size will also affect phase separation after the The maximum drop diameter is important since it is
completion of the multiphase process. generally considered proportional to the Sauter mean
Drop diameter distribution depends on the rate of diameter, d32 ,2,3 although this has been questioned.4,5
drop break-up and coalescence. In turbulent systems The Sauter diameter, commonly used in processes
the theory developed by Hinze1 is being used to predict depending on interfacial area,2,3,6 is defined as the
the maximum drop size, dmax , that can resist break-up. ratio of the third to the second moment of the drop
In stirred tanks dmax can be related to the tank Weber size distribution:
number, WeT , as follows:

k
ni di 3
ρc N 2 Di 3 d32 =
i=1
(3)
WeT = (1)
k
σi 2
ni di
i=1
where ρc is the continuous phase density, N is the
impeller speed, Di is the impeller diameter and σi is where di is the drop diameter in bin i and ni is the
the interfacial tension. number of drops with diameter in bin i.
To account for the effect of increasing dispersed Based on the proportionality to dmax , correlations
phase volume fraction (which can cause coalescence have been proposed for d32 that take the form of

∗ Correspondence to: P Angeli, Department of Chemical Engineering, University College London, Torrington Place, London, WC1E 7JE, UK
E-mail: [email protected]
Contract/grant sponsor: Royal Society
Contract/grant sponsor: EPSRC
(Received 11 June 2004; revised version received 7 October 2004; accepted 21 October 2004)
Published online 2 March 2005
 2005 Society of Chemical Industry. J Chem Technol Biotechnol 0268–2575/2005/$30.00 545
J Lovick et al.

eqn (2), which were verified experimentally. This size distribution directly to the rotational speed of the
aforementioned equation predicts an increase in drop impeller and the dispersed phase fraction. Gallego-
size with increasing dispersed phase volume fraction, a Linzon and Perez de Ortis18 applied a modified
phenomenon that was observed in investigations where Rosin–Rammler function to fit their data in emulsion
dispersed phase fraction increased up to about 40%.7 liquid membrane systems using flat blade turbines
At higher fractions (usually above 50%), however, a with different diameters.
further increase in the dispersed phase concentration The limited available data on average drop size
results in decreasing drop size.8,9 This behaviour was and distribution, especially in unstable dispersions
attributed to a change in the drop breakage mechanism and at high dispersed phase volume fractions, are
from turbulent eddy at low concentrations to boundary partly due to the difficulty in performing such
layer at high concentrations.8 measurements. Sample withdrawal was one of the
Similarly, in a dispersion, a minimum drop size first techniques to be used. The withdrawn drops
exists, dmin , which is the minimum drop diameter that can be measured continuously within the sampling
can resist coalescence.10 Compared with dmax , fewer tube by a microscope9,19 or a Coulter counter2,20
correlations have been proposed for predicting dmin , which records changes in electrical resistance caused
while the Kolmogoroff length scale of turbulence has by drops passing through. Alternatively, drops can
also been used as an estimate of the minimum drop be ‘frozen’ using a surfactant and measured ‘off-
size.4,11 Shinnar12 argued that in coalescing systems line’ with a microscope and a photometer.21 The
d32 would be proportional to dmin . problems associated with the withdrawal technique
Experimental data on average drop size mainly are coalescence of sampled drops inside the withdrawal
exist for low dispersed phase concentrations, where tube or during the measurement and biased sampling
a variety of measuring techniques can be used. Few of certain drop sizes. At high dispersed phase fractions
studies looked at high concentrations but only Brown further problems can arise from droplets overlapping
and Pitt6,13 and Pacek et al 14 obtained drop sizes in the sample tube, which can be overcome by
in unstable dispersions; in all other cases surfactant- sample dilution,19 or by ensuring that the diameter
stabilised emulsions were used.8,9 of the sample tube is smaller than the smallest
Information on drop size distribution is even more drop diameter.21 Absorption of light transmitted
limited while existing studies often propose different through a dispersion has been used to obtain average
statistical correlations to describe the range of drop interfacial area in stirred vessels.22,23 Weinstein and
sizes in stirred vessels. Normal distributions have been Treybal7 extended the technique by incorporating
suggested by Chen and Middleman15 and Brown and the light transmitter and detector into a probe
Pitt6 for standard configuration stirred tanks. Brown that could be put at different locations within a
and Pitt6 also reported a bi-modal distribution that dispersion; this enabled local average drop sizes to
can be modelled as a sum of two normal ones. Zhou be measured. Photography/video recording provide
and Kresta,16 for a range of impeller types, found additional information on the actual shape of the
that the distribution shape changed with increasing drops. If used outside the vessel, these methods are
impeller speed from a single peak at small drop non-intrusive but allow measurements away from the
sizes, to a double peak at small and large sizes walls only in dilute dispersions.24 Pacek et al 14 studied
and finally to a single peak at intermediate sizes. the area close to the wall using an internal light source
They concluded that a single distribution could and reported results for mixtures with dispersed phase
not describe their experimental data satisfactorily. fractions up to 70%. The recent use of endoscopes
In aqueous–organic systems close to phase inversion has allowed recording at different locations within
Pacek et al 14 found that the type of distribution would the container, overcoming the problem of dense
depend on the continuous phase, being uni-modal dispersions (up to 30%18 ) but in an intrusive way.
in the aqueous continuous systems and bi-modal The introduction of lasers resulted in a number of
in the organic continuous ones. In contrast Pacek drop-size measuring techniques, where the amount of
et al 5 using a turbine impeller, observed a bi-modal light scattered by the drops is related to their size.
distribution in aqueous continuous systems at low In diffraction, light scattered at low forward angles is
agitation speeds, which changed to a uni-modal one detected while in back scattering light reflected back
as the impeller speed increased. Interestingly, normal is collected by a detector mounted in the same guide
or log-normal functions that fitted the experimental as the light source.25 – 27 In the former technique light
data well when these were plotted in terms of has to pass through the dispersion which limits it
cumulative volume, could not represent satisfactorily to low dispersed phase fractions, while in the latter
the same data when they were plotted in terms of drop there is no such limitation. Simmons et al 26 suggested
number frequency. The Swarz–Bezemer function that light diffraction was only suitable for dispersed
also failed to predict the cumulative volume drop phase fractions less than 3%, while back scattering
distribution. Some less known distributions that have could only be used at fractions greater than 5%.
been used for liquid–liquid systems are the Erlang, the Comparisons of the two techniques at low volume
Weibull and the Gamma.16 Gal-Or and Hoelscher17 fractions gave inconclusive results, which may suggest
proposed a Gamma-type function that relates drop that considerable error can incur when the techniques

546 J Chem Technol Biotechnol 80:545–552 (2005)

 Drop size distribution in liquid–liquid dispersions

are used outside their range of applicability.25,26 A 10

Phase Doppler Particle Analyser (PDPA), where the 10 min
phase difference between light scattered is measured 8
20 min
40 min
from two collection angles, has also been used but

Number frequency
requires very dilute dispersions.4
6
In this work a light back scattering technique was
applied in unstable kerosene–water dispersions to
obtain drop size distributions. The light back scattering 4
probe can be used on-line and be placed at any location
within the dispersion thus allowing fast measurements 2
at high dispersed phase fractions even in unstable
systems. Results from the technique were evaluated
0
against data from high-speed video recording which 0 50 100 150 200 250
was combined with an endoscope. The technique was Drop size, µm
subsequently used to study drop sizes in oil-in-water
dispersions at volume fractions up to 60%. Figure 2. Evolution of number frequency drop size distribution with
time for 350 rpm impeller speed and 40% oil fraction. Data obtained
by ORM instrument.

2 EXPERIMENTAL APPARATUS
the mixture. At high dispersed phase fractions
The set-up for measuring drop sizes consisted of
phase continuity was monitored with a conductivity
a standard configuration baffled cylindrical tank (id
probe. Fluids were mixed for 15 min before any
138 mm) equipped with a six-bladed Rushton turbine
measurements were taken; initial measurements over
(65 mm diameter) with its centre located 65 mm from
a period of 1 h had shown that after 15 min the drop
the bottom of the vessel (Fig 1). The six blades had
size distribution as well as d32 remained constant. The
dimensions of 15 mm × 15 mm, while the disc had a
change in size distribution with time is illustrated in
diameter of 50 mm. The impeller was constructed of
Fig 2 for 40% oil and 350 rpm impeller speed.
1 mm thick stainless steel. The test fluids used were
A 3 Dimensional Optical Reflectance Measurement
tap water and kerosene (EXXSOL D140 with density
(3D ORM) technique, based on light back scattering,
ρ = 828 kg m−3 , viscosity µ = 5.5 mPa s, interfacial
was used for drop size distribution measurements.
tension σi = 44.7 mN m−1 ), with the kerosene forming
The results were evaluated against a method based on
the dispersed phase. No surfactant was added to
in-situ video recording. Both techniques are described
in detail below. Data were collected at 65 mm from
138 mm the bottom of the vessel at the height of the impeller
(Fig 1). The dispersed phase volume fraction varied
from 10% to 60%; in this range water was the
continuous phase at all impeller speeds used as
confirmed by the use of the conductivity probe.
Impeller speeds ranged from 350 rpm to 550 rpm.
The minimum impeller speed was chosen so that
there was complete mixing of the two phases at all
volume fractions used, while the maximum speed was
the one that gave no air entrainment.
65 mm
2.1 Optical Reflectance Measurement (ORM)
technique
The Optical Reflectance Measurement (3D ORM)
particle size analyser (by Messtechnik Schwartz
GmbH28 ) can provide in-situ and on-line drop size
distribution measurements. The technique uses a laser
beam (with 785 nm wavelength) with a focal point
0.6 µm in diameter at a maximum distance 1 mm from
ORM
178 mm

the instrument front. By having a focal point close

to the probe, the laser beam does not have to travel
Endoscope
far through the liquid mixture and can therefore be
used to measure chord lengths in high concentration
65 mm

dispersions. This focal point is rotated at a known

velocity within the sample and tracks a 3D volume with
8.5 mm diameter. When the rotating beam intercepts
Figure 1. Schematic diagram of the experimental set-up. a drop, light is scattered back from the chord length

J Chem Technol Biotechnol 80:545–552 (2005) 547

J Lovick et al.

in the beam path which through the optical system

of the instrument is transmitted to a detector where
it is transformed to an electrical signal. From the
signal duration and the known velocity of the rotating
beam the chord length that was intercepted can be
determined. A chord length distribution is therefore
built up over a period of time. If the drops are assumed
to be spherical, then their chord length distribution
can be found analytically. It is however, the inverse
problem that needs to be solved, ie to derive from a
(measured) chord length distribution the (unknown)
drop diameter distribution. The solution of the inverse
problem is achieved numerically with the software
DISPAS 2.52e. Since the probe can be used in-situ,
the problems associated with sample withdrawal are
also avoided. Also a very large number of drops, of
the order of thousands, is sampled with the probe,
much larger than in the commonly used photographic
or sampling techniques. This large number ensures Figure 3. A characteristic image obtained by the high-speed video
recording technique for 10% oil fraction at 450 rpm impeller speed.
that the maximum drop size, which is taken equal
to the maximum chord length measured, has been
captured. The drawback of this technique is the size the side of the vessel wall and at the same height as
of the probe (30 mm) in relation to the vessel. In this the 3D ORM instrument (Fig 1) (with a 90◦ viewing
work, however, the probe was placed in parallel to the direction). The imaging system used is capable of
flow stream that leaves the impeller, ensuring that the recording up to 1000 full frames per second, but
measured drops did not interfere with the instrument in the present experiments pictures were taken with
and their size was least affected by its presence. Light proper lighting at a speed of 500 fps, considered to
back scattering techniques have previously been used be a suitable recording rate, and at a shutter rate
in aqueous–organic dispersions formed in stirred tanks of 1/10 000 s. It must be noted that the endoscope
where they were found to respond swiftly to any used has a very narrow depth of field and its focusing
changes in process conditions which affect drop size. distance is approximately 2 mm from the front lens.
In all these investigations, however, the techniques The length calibration, needed to ensure the accurate
were implemented in either low dispersed phase measurement of the drops, was carried out before each
fractions25,27 or in surfactant-stabilised emulsions.29 run by measuring a microscale placed at the focusing
Validation experiments were carried out initially plane. Subsequent image analysis provided the sizes
with standard diameter latex and polystyrene mono- of the dispersed drops. Approximately 800 drops
sized spheres ranging from 20.6 µm to 230 µm were measured in each experimental run, which is
(±0.1 µm) (from Polymer Laboratories, UK and considered to be adequate for statistical calculations.14
Dynal Particles AS, Norway). These materials were The images collected with this technique (see Fig 3)
chosen as they have optical properties close to those of proved the validity of the assumption used in the
oil. Particle concentrations up to 10% were used and measurements with the 3D ORM instrument that
the comparisons between the nominal particle diam- drops are spherical. The repeatability of the method
eters and those obtained by the 3D ORM instrument was good and Sauter mean diameters measured in
can be seen in Table 1. repeated experiments were found to be less than 10 µm
different.
2.2 High-speed video recording technique
The results from the 3D ORM probe were compared
with those from in-situ video recording obtained with 3 RESULTS AND DISCUSSION
an endoscope (Schoelly) (od = 4 mm) attached to 3.1 Comparison of the two techniques
a high-speed camera (Redlake MotionScope PCI ). Due to the problems associated with video recording,
The endoscope was inserted into the dispersion at only volume fractions up to 10% were used as volume
fractions greater than this resulted in individual drops
Table 1. Comparison between the nominal sizes of the calibration not being clear in the photographs. Conversely since
particles and their average diameters obtained by the ORM instrument
ORM is not reliable for dispersed phase fractions
Nominal particle Measured average below 5%, measurements to validate the ORM
size (µm) particle size (µm) % difference technique against the video recording one were
restricted to dispersed phase volume fractions in the
20.6 19.23 −6.65
range 5% to 10%.
58.5 58.6 0.17
230 225 −2.17
Drop size distributions from both techniques are
shown in Figs 4a and b, for number frequency and

548 J Chem Technol Biotechnol 80:545–552 (2005)

 Drop size distribution in liquid–liquid dispersions

10 10
ORM ORM
endoscope log-normal dist.
8 8
Number frequency

Number frequency
6 6

4 4

2 2

0 0
0 100 200 300 400 500 0 100 200 300 400 500
(a) Drop size, µm (a) Drop size, µm

1 10
ORM endoscope
endoscope
0.8 8 log-normal dist.
Cumulative number

Number frequency
0.6 6

0.4 4

0.2 2

0
0
0 100 200 300 400 500
0 100 200 300 400 500
(b) Drop size, µm
(b) Drop size, µm
Figure 4. Drop size distribution for 450 rpm impeller speed and 10%
Figure 5. Experimental data and log-normal distribution curve for
oil fraction. (a) Number frequency distribution; (b) cumulative number
450 rpm and 10% oil fraction. (a) ORM instrument; (b) high-speed
distribution.
video recording technique.

cumulative number distribution respectively. It can

100
be seen that the two techniques are in reasonable
agreement. Measurements at the other conditions also
showed good agreement between the two techniques. 90
Small differences are probably due to the small number
of sampled drops that can be collected with the video 80
d32, mm

recording technique compared with those sampled

by the ORM (in the order of thousands). Both 70
techniques also gave drop number distributions that
could be fitted by the log-normal function, as can be
60
seen from Figs 5a and b respectively. The log-normal 10% 20% 30%
distribution is described by the following equation: 40% 50% 60%
50
 2 300 350 400 450 500 550 600
−0.5
ln d − ξ Impeller speed, rpm
1 δ
y(d) = √ e (4)
δ 2π Figure 6. Sauter mean drop diameter vs impeller speed for different
oil fractions. Data obtained by ORM instrument.
where d is the drop size and δ and ξ are parameters
of the log-normal distribution, with δ affecting the
distribution height and ξ affecting the distribution all experiments less than about 4 µm and on average
width. about 2.5 µm.
The above results suggest that the 3D ORM
instrument, which provides fast, on-line drop mea- 3.2 Results at higher dispersed phase
surements, can be used reliably to study unstable concentrations with the ORM technique
liquid–liquid dispersions. Further studies with the The Sauter mean diameters, obtained from 3D ORM,
ORM for different impeller speeds and dispersed at all experimental conditions used, are summarised in
phase concentrations are detailed below. The stan- Fig 6. It can be seen that there is an overall trend for d32
dard deviation in the Sauter mean diameter was in to decrease with increasing impeller speed. However,

J Chem Technol Biotechnol 80:545–552 (2005) 549

J Lovick et al.

10 400
350 rpm
450 rpm 350
8
Number frequency

550 rpm
300
6

dmax, mm
250
4
200

2 150

0 100 10% 20% 30%

0 100 200 300 400 500 40% 50% 60%
(a) Drop size, µm 50
300 350 400 450 500 550 600
10 Impeller speed, rpm
350 rpm
8 450 rpm Figure 8. dmax vs impeller speed for different oil fractions. Data
Number frequency

550 rpm obtained by ORM instrument.

6
100
4
90
2
80
d32, mm

0
0 100 200 300 400 500
(b) Drop size, µm 70

10 400 rpm
60 450 rpm
350 rpm 550 rpm
8 450 rpm
Number frequency

550 rpm 50
0 20 40 60 80
6
Dispersed phase fraction, %

4 Figure 9. Sauter mean drop diameter vs dispersed phase fraction for

various impeller speeds. Data obtained by ORM instrument.
2
previous investigators.8,9 As the oil fraction increases
0 further there is no clear trend between the impeller
0 100 200 300 400 500
speeds.
(c) Drop size, µm
The increase in the impeller speed will increase
Figure 7. Drop size distribution at different impeller speeds for drop breakage and lead to smaller drops. The effect
dispersed phase oil fractions: (a) 20%, (b) 40% and (c) 60%. Data on coalescence is less clear as the drop collision
obtained by ORM instrument. frequency will increase but the coalescence efficiency
which depends on the contact time of colliding drops
in some volume fractions an increase with the impeller will decrease. As can be seen, the overall effect in the
speed can be seen. Interestingly, when the actual drop current system is a slight decrease in the drop size.
size distributions are used, in all cases increasing the On the other hand an increase in the volume fraction
impeller speed resulted in a shift in the distribution would be expected to result in an increase in the drop
to smaller drop sizes (see Fig 7 for different dispersed size. A large amount of dispersed phase, however,
phase volume fractions). This indicates that average can change the properties of the continuous phase
drop diameters may not always be able to represent turbulence.30 Currently there is little work available
distributions satisfactorily. The maximum diameters on the interactions between continuous and dispersed
were also found to decrease with impeller speed (see phase in liquid–liquid systems but in the current
Fig 8), while the ratio of dmax to d32 varied between mixture they seem to result in small changes in drop
3.5 and 4.5 in all cases. size with volume fraction.
There is little overall trend in d32 values with All drop distributions were found to be uni-modal
increasing oil fraction at a constant impeller speed and were represented satisfactorily by the log-normal
(Fig 9). However, local phenomena do seem to exist. correlation. It was found that as the impeller speed
After an initial decrease, Sauter mean diameters increased, parameter ξ , the log-normal distribution
increase again up to 40% oil fraction. A peak at high width, decreased from 4.7 to 4.4, indicating narrower
dispersed phase volume fraction has been reported by distributions. There was no trend for parameter δ,

550 J Chem Technol Biotechnol 80:545–552 (2005)

 Drop size distribution in liquid–liquid dispersions

the distribution height, which ranged between 4.5 The authors suggested a predictive model similar to
and 5.5. that by Hinze1 that accounts for drops close to the tur-
Correlations of the form of eqn (2) did not predict bulence dissipation range. In the current work average
satisfactorily either the Sauter mean or the maximum drop sizes were larger than the eddy dissipation range
diameters. Interestingly, d32 and dmax were found (between 24 µm and 36 µm) and the model failed to
to be proportional to WeT −0.22 , and to WeT −0.21 predict the experimental data.
respectively, rather than to WeT −0.6 suggested by
eqn (2). As was pointed out in the literature, however,
experimental data against which these correlations 4 CONCLUDING REMARKS
were validated are mainly available for low volume A light back scattering technique (3D ORM tech-
fraction or surfactant-stabilised systems. Furthermore, nique) was used to obtain on-line drop size distri-
the above dependence on WeT has been questioned butions in unstable kerosene-in-water dispersions for
by a number of investigators. Pacek et al 5 and Cull dispersed phase fractions from 10% to 60%. Drop
et al,27 found experimentally larger power than (−0.6) size data from this technique compared favourably
at dispersed phase fractions up to 20%. Baldyga et al 31 with those obtained with a high-speed video record-
had shown that in dilute systems powers smaller than ing method that implemented an endoscope. Using
(−0.6) should be expected because of the phenomenon the ORM technique, drop diameter distributions were
of turbulence intermittency. Desnoyer et al 32 also found to become narrower and shift to smaller sizes
found higher values for the power in the WeT for with increasing impeller speed, a trend that was not
dispersed phase concentrations up to 60%, as well as always obvious from the average drop sizes. No sig-
a dependency of this value from the dispersed phase nificant effect of the dispersed phase volume fraction
fraction. The value tended to that of the classical Hinze was found. Log-normal distributions represented sat-
theory at very low phase fractions. According to the isfactorily the number frequency of the measured
authors a different mechanism of drop–turbulent eddy drops. Standard literature correlations of the form
interaction compared with that of the Hinze theory of eqn (2) were unable to predict average drop sizes.
may be needed for dense dispersions. Brauner33 has
also argued that while in dilute turbulent systems the
drop size is determined by a static force balance on ACKNOWLEDGMENTS
each drop in dense dispersions the turbulent kinetic Dr Angeli, Dr Lye and Dr Paras would like to thank
energy flux should exceed the rate of surface energy the Royal Society for the award of the Partnership
generation required for the break up of drops. It is Grant that made this collaboration between UCL and
possible that the turbulent energy may not be sufficient the Aristotle University of Thessaloniki possible. J
to create the small drops attained in dilute dispersions. Lovick would also like to thank EPSRC for providing
In addition, very few experimental data exist in the financial support for his studentship.
literature on drop size of unstable liquid–liquid dis-
persions at high dispersed phase fractions to compare
with the current work. Brown and Pitt6 investigated REFERENCES
1 Hinze JO, Fundamentals of the hydrodynamic mechanism of
drop size distributions of oil (with viscosity 1.6 mPa s)
splitting in dispersion processes. AIChE J 1:289–295 (1955).
in water, in a standard configuration vessel equipped 2 Sprow FB, Distribution of drop sizes produced in turbulent
with a disc turbine. Dispersed phase fraction varied liquid–liquid dispersion. Chem Eng Sci 22:435–442 (1967).
from 5% to 20% and impeller speed ranged from 250 3 Calabrese RV, Chang TPK and Dang PT, Drop breakup in
to 400 rpm. In agreement with the findings of the turbulent stirred tank contactors, part I: effect of dispersed-
phase viscosity. AIChE J 32:657–666 (1986).
current work, the authors also reported that volume
4 Zhou G and Kresta SM, Correlation of mean drop size and
fraction and impeller speed had little effect on d32 . minimum drop size with the turbulence energy dissipation
Average drop sizes ranged between 100 and 200 µm and flow in an agitated tank. Chem Eng Sci 53:2063–2079
which are slightly different from the current ones, as (1998).
would be expected given the differences in fluids’ prop- 5 Pacek AW, Man CC and Nienow AW, On the Sauter mean
diameter and size distributions in turbulent liquid–liquid
erties and vessel sizes. Weinstein and Treybal7 carried dispersions in a stirred vessel. Chem Eng Sci 53:2005–2011
out experiments using three vessel sizes and eight com- (1998).
binations of organic liquids and water, agitated with 6 Brown DE and Pitt K, Drop size distribution of stirred non-
a six-flat-bladed turbine up to 59% dispersed phase coalescing liquid-liquid system. Chem Eng Sci 27:577–583
fractions. Experimental data (shown in the paper for (1972).
7 Weinstein B and Treybal RE, Liquid–liquid contacting in
dispersed phase fractions up to 36%) revealed that unbaffled agitated vessels. AIChE J 19:304–312 (1973).
average drop diameters increase with increasing dis- 8 Boye AM, Lo M-YA and Shamlou P, The effect of two-liquid
persed phase fraction and decreasing impeller speeds. phase rheology on drop breakage in mechanically stirred
Sauter mean diameters (between 200 and 550 µm) vessels. Chem Eng Comm 14:149–167 (1995).
were larger than the ones observed in the current 9 Kumar S, Kumar R and Gandhi KS, Alternative methods of
drop breakage in stirred vessels. Chem Eng Sci 46:2483–2489
work, probably because unbaffled vessels were used (1991).
that have lower power consumption for the same 10 Liu S and Li D, Drop coalescence in turbulent dispersions.
impeller speed. No distribution data were reported. Chem Eng Sci 54:5667–5675 (1999).

J Chem Technol Biotechnol 80:545–552 (2005) 551

J Lovick et al.

11 Tsouris C and Tavlarides LL, Breakage and coalescence models 23 Urua IJ and Del Cerro MCG, Measurement of large gas–liquid
for drops in turbulent dispersions. AIChE J 40:395–406 interfacial areas by the light transmission method. Canadian
(1994). J Chem Eng 65:565–569 (1987).
12 Shinnar R, On the behaviour of liquid dispersions in mixing 24 Narsimhan G, Nejfelt G and Ramakrishna D, Breakage func-
vessels. J Fluid Mech 10:259–275 (1961). tions for droplets in agitated liquid–liquid dispersions. AIChE
13 Brown DE and Pitt K, Drop breakup in stirred liquid-liquid J 30:457–467 (1984).
contactor, in Proc Chemeca 1970 Conf, Melbourne and Sydney, 25 El-Hamouz AM and Stewart AC, On-line drop size distribution
pp 83–97 (1970). measurements of oil–water dispersions using a Par tec M300
14 Pacek AW, Moore IPT, Nienow AW and Calabrese RV, Video laser backscatter instrument, in Proc SPE Annual Technical
technique for measuring dynamics of liquid–liquid dispersion Conf and Exhibition SPE36672, pp 785–796 (1996).
during phase inversion. AIChE J 40:1940–1949 (1994). 26 Simmons MJH, Zaidi SH and Azzopardi BJ, Comparison of
15 Chen HS and Middleman S, Drop size distribution in agitated laser-base drop size measurement techniques and their
liquid–liquid systems. AIChE J 13:989–995 (1967). application to dispersed liquid–liquid pipe flow. Opt Eng
16 Zhou G and Kresta SM, Evolution of drop size distribution in 32:505–509 (2000).
liquid–liquid dispersions for various impellers. Chem Eng Sci 27 Cull S, Lovick JW, Lye GJ and Angeli PA, Scale down studies
53:2099–2113 (1998b). on the hydrodynamics of two-liquid phase biocatalytic
17 Gal-Or B and Hoelscher HE, A mathematical treatment of the reactors. Bioprocess Biosyst Eng 25:143–153 (2002).
effect of particle size distribution on mass transfer dispersions. 28 Messtechnik Schwartz GMBH, Apparatus for measuring
AIChE J 12:499–508 (1966). particle dimensions in fluids. UK Patent 2 325 973 (2000).
18 Gallego-Linzon T and Perez de Ortis ES, Drop sizes in liquid 29 Rimpler S and Daniels R, In-situ particle sizing in highly
membrane dispersions. Ind Eng Chem Res 39:5020–5028 concentrated oil-in-water emulsions. Pharm Tech Europe: 8:
(2000). 72–80 (1996).
19 Godfrey JC and Grilc V, Drop size and drop size distribution 30 Pal R, Pipeline flow of unstable and surfactant-stabilised
for liquid–liquid dispersions in agitated tanks of square cross emulsions. AIChE J 39:1754–1764 (1993).
section, in Proc 2nd European Conf on Mixing, Cambridge, 31 Baldyga J, Bourne JR, Pacek AW, Amanullah A and Nienow
England, pp C1: 1–20 (1977). AW, Effects of agitation and scale-up on drop size in
20 Koshy A, Das TR and Kumar R, Effect of surfactants on turbulent dispersions: allowance for intermittency. Chem Eng
drop breakage in turbulent liquid dispersions. Chem Eng Sci Sci 56:3377–3385 (2001).
43:649–654 (1988). 32 Desnoyer C, Masbernat O and Gourdon C, Experimental study
21 Verhoff FH, Ross SL and Curl RL, Breakage and coalescence of drop size distributions at high phase ratio in liquid–liquid
processes in an agitated dispersion. Experimental system and dispersions. Chem Eng Sci 58:1353–1363 (2003).
data reduction. Ind Eng Chem Fundam 16:371–377 (1977). 33 Brauner N, The prediction of dispersed flows boundaries in
22 McClaughlin CM and Rushton JH, Interfacial areas of liq- liquid–liquid and gas–liquid systems. Int J Multiphase Flow
uid–liquid dispersions from light transmission measurements. 27:885–910 (2001).
AIChE J 19:817–822 (1973).

552 J Chem Technol Biotechnol 80:545–552 (2005)