7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
Determination and study of hold up and flow patterns in two-phase flow
liquid-liquid systems for horizontal and inclined pipes using image
Montova, Gustavo, Valecillos, Maria; Romero, Carlos; Garcia, Janneth; Gonzalez, Dosinda
Department of Thermodynamics and Transport Phenomena
University Simon Bolivar
Valle of Sartenejas, Caracas, Venezuela, Postal Code 1080
Abstract: Several flow patterns for horizontal and inclined pipes were visualized, in order to calculate hydrodynamics
parameters using a computational algorithm that was generated for this investigation. In order to accomplish this, the
images were acquired using a high speed camera in a tube bank equipment of the Transport Phenomena Laboratory of
the University Simon Bolivar for the horizontal configuration, and in different tube bank equipment in the Unitary
Operations Laboratory for 45 degrees of inclination. Diverse flow patterns for the horizontal pipes were characterized
following the classification used by Trallero (1996)  and Flores , then the images were processed in order to
obtain the height's phase and hold up. The resulting patterns that were obtain for 0 degrees consist in four models: two
segregates (ST and ST&MI) and two disperse (o/w and Do/w&w). For 45 degrees, three dominate by water: Oil in
water dispersion-pseudo pattern (PS), oil in water dispersion-cocurrent (CC), very fine dispersion of oil in water (VDF
o/w), and one dominates both by water and oil: transition (TF). were the pattern that was observed. Finally, a flow
pattern map that which depends of the superficial velocities was elaborated for the horizontal pipes in order to relate
the hydrodynamics parameters behavior using the presented parameters. This data was compared with correlations and
previous experimental results.
Key-Words: Flow pattern, horizontal pipes, inclined pipes, computational algorithm, tube bank, height's phase, hold
up, flow pattern map.
The obj ective of this work is to relate
hydrodynamics parameters like the height's phase and
hold up, calculated by image processing, with the flow
patterns presented by different configurations of the
water-oil system's superficial velocities, and use the
obtain data to generate a flow pattern map that allows to
predict flow changes at the operational conditions.
2 Theoretical Fundaments and Applied
2.1 Flow Patterns
In the flow pattern characterization for the liquid-liquid
systems there has been yet no consensus, because of the
numerous variable on which depends the geometric flow
configurations of the two fluids. The similarities of
density and viscosity of the two fluids has make difficult
the prediction on which of the two phases moves faster
in stratified flow or which has more propensity to create
drops and coalesce. That is why the investigators who
The biphasic systems liquid-liquid are presented in a
number of operations in the industry, which is the
reason why the study of this system, for more than 30
years, has been so important for the investigators
working in this area. Recently, the multiphase's system
has gain some interest thanks to its applications in
catalytic wall reactors, because of the transport
phenomena of heat, momentum, and mass that are
implicit in the fluid dynamics of these systems.
Even more frequently, the principle scenario in
which the liquid-liquid flow in pipes is presented is in
the crude oil industries, both in deposits and in the crude
transport, which can be accompanied by water, natural
gas, and solids suspensions.
The main problem is presented when trying to
predict the configuration that the crude will take through
the pipe, because the associated pressure drop will
depends on this, as well as the pipe dimensions, and the
work on these hydrodynamic phenomena based
themselves in the qualitative appreciation of their
observation of this phenomenon, and identification of a
wide variety of patterns.
Oglesby (1979)  proposed fourteen different flow
patterns, while Russel et al  and Malinowsky (1975)
 described only three or four.
One of the most complete works in this area was that
made by Trallero et al. in 1996 , in which he
described six flow patterns classified in two groups:
segregate, and disperse.
The segregated flow is characterized by the
continuity in the axial direction of the two phases
separate by gravity, where the denser one flows by the
bottom of the pipe. The disperse flow is characterized by
the mutual penetration of the interface that separate both
fluids and this turbulence leads to the dispersion of the
Table 1 Description of the
Trallerc's flow patterns (1996).
Stratified Both phases are completely
Flow separated by a flat interface
(ST) without wave.
Stratified The interface begins to curl and
flow with penetrate the oil droplets in
mixing at water and vice versa, forming a
the interface third laver on the interface. This
(ST & MI) occurs by increasing the mixing
Water as dominant phase. It has
Dispersion of a dispersion of oil in water at
oil in water and the top of the pipe while in the
water bottom of it, flows a continuous
(Dolw & w) laver of water.
Oil in water Water as dominant phase. At
emulsion high speeds the oil is mixed in
(olw) water to form a dispersion
across the pipe cross area.
Dispersion of Oil as a dominant phase. On the
water in oil and top of the pipe flow water
oil in water droplets contained in the oil
(Dolw & D while the bottom, drops of oil
w/o) flowing on water.
Oil as a dominant phase. It is
Water in oil noted at high speed mixing and
emulsion high oil input relations, which
(w/o) allows to completely dispersing
Fi. lowpttm b rllr (96
In lieaue h atm aerpre hog h
Figur 2 shw h low pattemnb mraplr devlopd b
Traleow en a, which ca eeriete transition bewena
the patterns according to the relation between the axial
surface velocities of water and oil.
Fig. 2 Flow pattern map according
The operating conditions used by Trallero for the
realization of the map are shown below:
Table 2 Conditions of Trallero
D [cm] 5,04
p,, [cP] 0,97
p,, [kg/m3] 1037,00
Po [kg/m3] 884,00
When the pipe is inclined, the flow pattern will start
to change. As the angle increase it is more difficult to
see stratified flow, and easier to observe sloops, and
drops. At 45 degrees of inclination, there is not yet
enough force to produce only drops or annular flow, but
a transition between the horizontal flow patterns, and the
vertical ones. At low superficial velocities the flow
patterns can be describe as those observed by Flores et
al. (1997) . The principle difference between the
flow patterns observed in the case where horizontal to
inclined, is due to the gravitational component is normal
to the direction of flow, together with the pressure and
viscous forces that are important due to the high
viscosities of the oil.
Table 3 Description of the Flores's
flcw patterns (1996).
Oil in Water It happens at low and moderate
Dispersion- Usw and Use The range of
Countercurrent occurrence depends on the
(CT) angle of inclination. It is
characterized by bubbles of oil
in the top of the pipe and the
flow of water in the bottom of
the pipe, due to the gravitational
force to reverse the flow
Oil in Water Observed at speeds greater
Dispersion- than Usw and moderate to
Pseudo low speeds of Use, which
pattern causes segregation of the oil
(PS) droplets in the top of the pipe
Caracterized by segregated oil
Oil in Water drops on top of the pipe and
Dispersion- bottom cocurrent flow. Usw
Cocurrent (CC) happens to increase, since it
divides the cap bubble flow
Very fine By increasing Usw breaks the
dispersion of oil oil droplets into smaller
in water (VDF droplets due to increased
o/w) turbulent forces. (Uniform
Transition Grouping characteristics and
(TF) flow configurations
dominated by oil and water.
Consists of water droplets in
Dispersion of oil, which reduced their size by
water in oil increasing Qo, uniformly
(Dw/o) distributed in the central region
surrounded by the oil pipeline
like a ring.
Very fine These are characterized by a
dispersion of uniform distribution of water
water in oil droplets in oil.
Besides the way they are distributed inside the
pipeline stages in the multiphase flow, other
hydrodynamic parameters are important in these
systems. One of them is the height of the phase, which
can be obtained visually by measurement of any stage
in-situ and processing through a program or software for
that purpose. Additionally, it is necessary to know the
effect of sliding, phase retention or Hold Up.
2.2 Hold Up
When the phases differ in density and / or viscosity, the
less dense tends to flow at an average speed higher than
the other. This causes one of the most important
characteristics of the two-phase flow, the slip from one
phase to the other, or retention of one phase with respect
to the other . The Hold Up can be determined using
H,= volume of phase a in a segment of pipe
Volume ofpipe segment
The Hold Up varies from 0 to 1, and its maximum
value corresponding to a monofluid phase.
The primary importance in determining the Hold Up
experimentally, is that several hydrodynamic parameters
depend explicitly or implicitly from this.
2.3 MZHEIGHT AND MZDROPS
The MZ Height is an algorithm made in MATLAB@)
using a simulation to find each phase's height and hold
up in a two phase system for stratified flows. First, the
image is save in a variable in MATLAB@) and send to a
simulation made in Simulinks) in which is first treated
with a median filter, then with a close box in order to
perform a morphological close, after what it is treated
again with two median filter expecting to reduce or
eliminate any possible noise in the image, then a data
conversion to double is apply. After this pre-treatment,
Canny Edge detection will be applied to the image in
order to binaries it and find the top and bottom limits of
the pipe and the interface border. After this, the now
binary image will be treating with close and dilate in
order to fill the spaces that were not close before.
Finally, it will be apply three or more (depending of the
necessities of the user) submatrix to choose the limits
between the top, bottom and interface of the image, and
with this eliminate any error that could be gain when the
next box is use. After this, the picture passes by a blob
analysis box in which is found the centroid of each white
line with a restriction of length. Then an embedded
Matlab@) function is used to define the Y parameter of
Fig. 3 Flow patterns by Flores in the
respectively order of table 2 (1996)
each centroid, and finally this is send to workspace as
the height for each line from the origin. After that, with
vector difference operations it is found the height of
each phase, and with this parameter by the ratio of each
phase's height and the diameter of the pipe make
possible to calculate the hold up.
The MZ DROPS works in a similar way, and it is
use in order to find the hold up of each phase by the
ratio of the area occupied by each one, and the total area
of the pipe in the image.
Fig. 4 MZ HEIGHT Simulation
Fig. 5 MZDROPS Simulation 1
Fig. 6 MZ DROPS Simulation 2
Before the image go from the workspace to the
simulation, it is treated with an algorithm where it is use
a padded size function, a high frequency filter, a gscale
function and a histogram equalization filter. After this,
the image is send to the first simulation, in which is
apply a median filter, a gamma correction to improve the
bright of the image, a closing box, another median filter,
a contrast adjustment with a similar intention than the
gamma correction, a double data type conversion, a
Canny Edge Detector, a closing, and a dilatation box,
respectively, and send to workspace. Then, two
approximately 1*100 pixels vertical line is make, in
order to close any drop that has not yet completely
appear in the pipe, after what an imfill function is apply.
Then it will be send to another simulation in which the
area of the drops will be calculate and after send it back
to the work pace, its hold up will be calculate.
Fig.7 Processed and unprocessed image.
In the case of inclined pipes, it is use an imrotate
command, and then a new section of study in the now
horizontal picture is taking in the algorithm.
2.3 HASAN AND KABIR CORRELATION
This correlation will be use along with previously
experimental data, in order to validate the results
for 45 degrees of inclination.
rg a jp~~- p_~~'
Hasan, A. R y Kabir. C. S., "A new M~odel for Two
Phase Oil Water Flow: Production Log Interpretation
and Tubular calculations", Proceeding AnnualTechnical
Conference and Exhibition, 3, 369-376, 1988.
3 Equipment descriptions
There is a feed section composed mainly of a water
storage tank, brand RESINCA (T-101), an oil storage
tank, brand RESINCA (T-102), two progressive cavity
pumps (BCP), brand SEW DO BRASIL LTDA ( B-101
and B-102) and two rotameters mark OMEGA
ENGINEERING, INC. (FI-101 and FI-102)
A test section which consists of a plexiglass pipe with an
inner diameter of 0.0445 m (1.75 in) and outer diameter
of0.0508 m (2 in).
A display section that consists of a high speed
camera 4540mx model Ektapro Kodak brand image,
which allows recording of 30 to 4500 frames per second
in full screen mode. Furthermore it is a lens Nikon
60mm f/2.8D AF Micro to increase or decrease the focal
aperture. This high-speed camera is connected to a
computer where they are recorded and digitized images
taken with their respective software, and a section where
both phases are separate.
Table 4 Hidrovenoco S-100 Propierties
Viscocity a 400C [cSt] 100
Viscocity a 1000C [cSt] 11.0
Viscocity Index 95
Specific Gravety 0.884
Fig.9. Experimental equipment for 45 degrees
4 Experimental Procedures
The procedure is divided into three stages according to
the nomenclature of the experimental assembly:
* Stage of fluids pumping: in which the fluid was pump
to the pipes until it reaches the visualization cell.
* Stage of visualization where the images were took at
different speeds with a high speed camera and a
visualization cell full of glycerin where the fluid pass.
There is the need to wait until the fluid is completely
developed before taking the photographs.
* Stage of separation of the two fluids.
5 Results and Discussions
There were performed a total of 22 experiments by
changing flow of water and oil to produce a full map of
flow patterns, and process the images with the algorithm
developed in Matlab 9. In the images that were taken
for the 27 flows in study shown four flow patterns, for 0
degrees, two of them which are dominated by the
aqueous phase (dispersion water, and oil in water
emulsion of oil in water) and the remaining flow
patterns corresponding to segregated (Stratified and
Stratified mixed in the interface). For 45 degrees, there
were observed four flow patterns according to Flores et
al. , three dominate by water: Oil in water
dispersion-pseudo pattern (PS), oil in water dispersion-
cocurrent (CC), very fine dispersion of oil in water
(VDF o/w), and one dominate both by water and oil:
transition (TF). For simplicity in order to name the
games of flow, it will be work with a nomenclature that
begin with the volumetric flow in GPM of water, by the
letter "w" followed by the oil volumetric flow in GPM
with the letter "o".
For horizontal pipes at low water surface velocities
(UWS) and low to moderate surface speeds of oil (UOS)
the flow pattern consists of two phases segregated
flowing through the pipe with a well defined interface.
Figure 10 shows the flow pattern Stratified (ST) for
these operating conditions in the system oil-water:
Table 5 Hidrovenoco S-68 Propierties
Density a 210C
Viscocity a 230C [Pa~s1 0.1310
Fig. 8. Experimental equipment for 0 degrees
T-101 water storage tank.
T-102 oil storage tank.
B-101/102 progressive cavity pumps.
T-103/104 senaration tanks
For 45 degrees of inclination, similar equipment was
used. The diameter of the pipe was 3.175 degrees. This
experimental equipment has a possible inclination from -
90 to 0 degrees (ascendant flow).
The oil used was Hidrovenoco S-68.
Fig 13. Flow pattern with velocity 0.324 m/s
oil and 0.081 m/s of water.
Depending on the surface speeds of both the water
and oil, a map with the flow patterns observed in the 27
experiments was built, being distributed as follows:
0.0 0.1 01.2
Fig. 10. Flow pattern with velocity of 0.122 m/s oil and
0.08 1 m/s of water.
As the flow of water circulating in the pipe increase,
maintaining the intervals of operation for the oil, drops
of oil begins to appear in the aqueous phase, and water
droplets of equal size in the layer of oil. At this point it
begins to appear the stratified flow with mix interface
(ST & MI). (Figure 11)
Fig. 11. Flow pattern with velocity 0.325 m/s oil and
0.325 m/s of water.
For high speed water and low speed oil, it can be
observed the transition to a disperse flow pattern,
characterized by the incorporation of multiple oil drops
of small-diameter in the aqueous phase. With the
increased of the flows, the turbulence in the pipe
increases as well and the droplets collapse to form a
dispersion oil in water and water (Do / w & w) in which
the wavy interface is barely visible and there is not
continuity in the layer of oil. (Figure 12)
Fig. 14. Obtained map flow patterns.
Comparing this map with the flow patterns
set by Trallero , it can observe certain
correspondence in terms of flow regimes which are
presented. For low superficial speeds for both water and
oil, it presents a stratified flow area (ST), followed by a
vast region in which the Stratified flow with mix at the
interface dominates (ST & MI) for surface velocities of
water and oil above 0.1 m/s. In the map presented by
Trallero  it can be note that the stratified flow
pattern is defined by the VKH line resulting from the
stability analysis of the biphasic mix, which indicates
the region where droplet formation begins. As it is
increases the phase's velocity, it can be observed in the
diagram observed two IKH analysis lines that separate
the flow patterns segregated from dispersed. In the flow
map obtained from experimental data, we can
distinguish the areas enclosed by lines IKH and VKH.
The comparison between the analysis made by
Trallero  and the one obtained for this work might be
due to similarities in the operating conditions and the
physical properties of both fluids, observing the
overlaps in terms of viscosities and densities
fluids used in the experiments and also the
resembles of the diameter of the pipe. This,
allowed to extrapolate the lines of stability analysis that
Trallero and colleagues obtained by the results of
Fig. 12. Flow pattern with velocity of 0. 122 m/s
oil and 0.284 m/s of water.
However, if the surface velocity of oil increases,
maintaining low water flow into the pipe, it
will be seen as the oil droplets spread around
the whole pipe by simultaneous action of
dynamic and thrust forces. In this case, there will be
observed a turbulent mixing of the aqueous and oleic
phase and the disappearance of the interface. Figure 13
shows an image took with the high speed camera for a
pattern of oil in water emulsion. Because the drops of
oil that are formed during the emulsion have a very
small diameter, they could not be captured with
sharpness by the camera.
0 500 1000 1500 1000 2500
Fig. 15. Height of the oil phase
different flow rates to 2 GPM of water.
a on -
Fig. 16. Hold up of oil phase
different flow rates to 2 GPM of water.
the calculation of the centroid carrying an small error in
Plotting the hold up for a flow of
3 GPM of water, it is interesting to analyze the
trend exhibited since both patterns presented both
stratified flow (a low flow of oil)
and stratified mixed in the interface (for
high flows of oil). That makes possible to observe
the influence of the undulation of the interface
system when calculating the hold up:
0 500 1000 1500 2000 2500
Fig. 17. Hold up of the oil phase
different flow rates to 3 GPM of water
Inspecting Figure 17, it can be verified that the
3w20 flows and 3w30 exhibit an stable behavior for the
hold up, which comes into consistent with their
classification as stratified flow (ST). As can be seen,
with the increasing of the flow of oil flowing through
the pipe, the behavior of hold up tends to become more
unstable as the droplet formation began, corresponding
to regimes Stratified flow with mixing at the interface
(ST & MI), for 3w50 flows and 3w70.
For the rest of the water streams similar graphics
were generated, which are presented below. As was
mentioned already, the singularities presented by the
graphics are due to the presence of oil droplets of large
diameter that interfere with the reading of the interface
done by the program.
Similarly, the lighting condition is another key factor
that affect the proper processing of the images. The
incandescent bulbs used and a good management of the
the camera and its lens shutter allow us to obtained a
large amount of high quality images for the analysis
with computer, but there were others where the good
lighting condition decreased and processed with the
program yielded significant errors that are reflected
in the graphics as the oscillator behavior
of the hold up of the interface. Under these conditions
The program MZHEIGHT was used to analyze
with a sample between 500 photos and 2,000 varying for
The data obtained from processing the images,
allowed to develop graphics to report the variations for
each image of the height and hold up for the oil phase,
for a constant water flow and a variable oil flow. The
water and oil flows are expressed in gallons per minute
(GPM). For 2 GPM of oil earned the following graphic:
In these figures is evidence the analog behavior of
the phase's height and Hold Up for each set of flow
graphed. This trend is evident in the same way
for all the experiments performed, since the
Hold Up was estimated by the program based on the
obtained phase's heights, ensuring that their sum
never exceed the unit.
Notably, the game flow plotted in
Figures 15 and 16, corresponds to a flow pattern
stratified (ST). Although the water-oil interface
appears flat in the pictures, are evident
periodic fluctuations of the phase height (and
accordingly, the Hold Up), where the values obtained by
the program vary in the range of 0.04 along the
photos analyzed. This trend is due to irregular
appearance of droplets of large diameter oil
travel through the system interface and appear
reflected in the photographs taken. While the algorithm
done in Simulink@ for the analysis of images is quite
accurate in giving the phase's height for a set of images,
when drops are linked to the system interface, it alters
Table 4 shows a discrepancy between some data of
hold up respect to the real flow running through the
pipe. One would expect that, at higher proportions of
water in relation to the oil, the first phase has to have
greater hold up, but this does not happen
presumably due to the phenomenon of "wetting" which
refers to the ability of oil in this case,
to keep in contact with the surface of the pipe
as a result of a balance between adhesive and
cohesive forces. This happened because it was not
worked gradually increasing the flow of oil caused a
layer of oil added to the inner walls of the pipe.
This laver is reflected in some images and had a
significant influence on the reading by the program.
Table 6. Averae values of Hold Up for eachphs
2w3o 0,152 0.818697
2w4o 0,087 0.90425
2w5o 0,127 0.902871
2w6o 0,134 0.866238
3w2o 0,314 0.679688
3w3o 0,265 0.763087
3w5o 0,216 0.738906
3w7o 0,243 0.749274
4w2o 0,359 0.721757
4w3o 0,315 0.680738
4w4o 0,323 0.532045
4w6o 0,253 0.670653
5w2o 0,437 0.581455
5w3o 0,369 0.59337
5w4o 0,329 0.667953
5w6o 0,339 0.639228
6w2o 0,505 0.502898
6w4o 0,381 0.6148
6w6o 0,314 0.686088
7w2o 0,525 0.475411
7w7o 0,316 0.684159
8w2o 0,558 0.442177
Fig. 18. Hold up of the oil phase
different flows of 4 GPM of water.
a 50e 1oon 150o 2000 Zsoo
Figure 19. Hold up of the oil phase
different flows to 5 GPM of water.
Fig. 20. Hold up of the oil phase
different flow to 6 GPM of water.
Fig. 21. Hold up of the oil phase
different flows to 7 GPM of water.
Figure 22. Height of the oil phase
different flows of water to 8 GPM.
8w4o 0,423 0.57659
8w6o 0,370 0.629581
8w8o 0,234 0.766584
For pipes with 45 degrees of inclination, three
velocities where studied. Next, it is presented both the
graphics and a table with the superficial velocities, the
hold up calculated by the program, the hold up
calculated by an experimental work with the same
operational condition but using simultaneous close of
valves instead. It is also presented both the error compare
with the experimental and thorical values method.
Oil flow Water flow Uso Usw
rate (gpm) rate (gpm) (m/s) (m/s)
1 2 0,02 0,16
1 3 0,02 0,24
1 5 0,02 0,40
With respect to the goals outlined in this research on the
study of two-phase flow in oil-water systems were
unable to conclude that:
* We identified four flow patterns prevailing on the
model of Trallero (1996): two segregated patterns
stratified type (ST) and stratified mixed in the interface
(ST & MI), and two patterns flow dominated by the
aqueous phase dispersion of oil in water and water (Do /
w & w) and oil in water emulsion, and for 45 degrees,
three dominate by water: Oil in water dispersion-pseudo
pattern (PS), oil in water dispersion-cocurrent (CC),
very fine dispersion of oil in water (VDF o/w), and one
dominates both by water and oil: transition (TF).
* The flow pattern map obtained depending on surface
velocities of the phases exhibits trend similar to that
obtained by Trallero et al in 1996, who worked on
experimental conditions similar to those in this work.
*Image processing allowed the calculation of
automated hydrodynamic parameters biphasic
characteristic of these systems. However, should
improve the lighting and image quality acquired in the
experimental work to minimize errors in the results of
*Due to the fouling of the internal walls of
the pipe with a layer of oil, there were errors in phase
heights gave by the program image processing, being
difficult to link in some cases the results obtained with
the flow Actual oil and circulating water through the
 Gonzalez-Mendizabal, D., Sanchez E., Zeppieri, S.,
Romero, C., 2007, "Experimental Study on patterns
Liquid-liquid flow in vertical pipes. 8
Iberoamerican Congress of Mechanical Engineering.
 Irandoust, S., Andersson, B., 1988, "Mass transfer
and liquid-phase reactions in a segmented two-phase
flow monolithic catalyst reactor. Chemical engineering
 Russell, T. W. F., G.W. Hodson and G.W. Govier,
Horizontal Pipelmne Flow of Mixtures of Oil and Water,
The Canadian Joumnal of Chemical Engineering, 39, 9-
 Malinoswky, M. S., An Experimental Study of Oil -
An Air-Water Flow-Flowing Oil Mixtures in Horizontal
Pipes, The University of Tulsa, MS Thesis (1975).
 Charles, M.E., G. W Govier. and G. W. Hodson, The
horizontal pipeline flow of equal density oil-water
mixture, The Canadian Joumnal of Chemical
39, 27-36 (1961).
 Guzhov et al (1973)
Oil and water flow studied with the respectively
Table 8. Continuation of table 7, with the tehorical,
experimental, and predicted hold up with the respectevly
Fig. 14. Validation for the program for stratified flow
For the validation of the program for stratified flow'
it was measure the pixels of a photograph using the
image toolbox of MATLAB@, and calculating the hold up
as the number of pixels for the oil phase and the diameter of
the pipe. Giving a value of 0.85, which is very close to the
average value gave by the program for the respectively
 Oglesby, K. D., An Experimental Study on the
Effects of Oil Viscosity, Velocity and Water Mixture
Fraction on Horizontal Oil-Water Flow, The University
of Tulsa, MS Thesis (1979).
 Clarke (2001)
 Trallero, J. L., C. Shariah and J.P. Brill, A Study of
Oil Water Flow Patterns in Horizontal Pipes,
Proceeding Annual Technical Conference and
Exhibition, 1, 363-375 (1996).
 Flores et al. (1998)