ARTICLE IN PRESS

Chemical Engineering Science 65 (2010) 1407–1414

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: www.elsevier.com/locate/ces

Droplet size distribution after liquid entrainment in horizontal stratified

two-phase three-field dispersed flow
L.E. Patruno a, J.M. Marchetti a, C.A. Dorao b, H.F. Svendsen a, H.A. Jakobsen a,
a
Department of Chemical Engineering, Norwegian University of Science and Technology, N-7491 Trondheim, Norway
b
Department of Energy and Process Engineering, Norwegian University of Science and Technology, N-7491 Trondheim, Norway

a r t i c l e in fo abstract

Article history: Traditional liquid entrainment correlations for two-phase flow systems only provide information about
Received 17 July 2009 the net mass transfer between the phases and not much data are available regarding the drop size
Received in revised form distribution of entrained, low surface tension liquids (i.e. hydrocarbons) when the working pressure is
16 October 2009
higher than the atmospheric pressure.
Accepted 17 October 2009
In this paper we study the additional drop size distribution information and we provide with a
Available online 27 October 2009
probability function that can be taken into consideration in suitable mathematical frameworks in order
Keywords: to increase the level of detail in describing multiphase systems. The presented probability function is
Drop case dependent and developed for the current data.
Entrainment
The actual dataset provides new results for liquid entrainment of a mixture of Exxsol D60TM as the liquid
Films
phase and pure nitrogen as the gas phase. The calculated corresponding surface tension was 24.4 mN/m.
Fluid mechanics
Hydrocarbon & 2009 Elsevier Ltd. All rights reserved.
Population balance

1. Introduction an accurate droplet size description. More complex models, like those
presented by Patruno et al. (2009c, d) make use of a droplet density
The description of the interaction between phases in multiphase function which takes into account the drop size distribution. This
flow is of major interest for the oil and gas industry as the presence actual work provides the reader with the entrainment shape
of droplets in the gas phase can cause erosion and breakdown of function, and Section 3.3 outlines the framework and the applic-
equipment. The use of conventional separators may not be ability of an entrainment shape function to model a multiphase
sufficient since very small droplets may be created from entrained dispersed flow.
liquid films at the walls. The droplet size plays a major role Ambrosini et al. (1991) present a correlation for the resulting
in choosing which separation technology is to be used, as smaller droplet size in annular flow but it predicts only a mean drop diameter
droplets are more difficult to separate than bigger drops given and does not account for statistical variations generating a droplet
the smaller ones have low inertia and will follow the gas size distribution around the mean. Kataoka and Ishii (1982) developed
stream lines. entrainment models for various regimes and modified the Paleev and
If the goal is to model the gas scrubbers in order to improve their Filipovich (1966) correlation to come up with a model closure for
design and increase their efficiency, a mathematical framework that entrainment and deposition. The advantage of this closure against
can take into account the phase interaction phenomena is of major those of Pan and Hanratty (2002a, b) is that it gives the net
significance. As a first approach, we could identify the presence mass transfers as a global parameter, but if droplet size discrimination
of liquid both as a continuum (i.e. a liquid film) and as a dispersed is to be modelled, then an entrainment spectrum has to be
phase (i.e. drops). Considering also the gas, the system is described considered. Recently, Sawant et al. (2008) have found that the Pan
as a two-fluid, two-phase, three-field problem. Models such as and Hanratty (2002a) correlation was satisfactory for low pressures,
Whalley (1990), Morud (2007) or Chatzikyriakou et al. (2009) could but underpredicted the entrainment in the high pressures regimes
be used to predict the overall velocities and mass transfers, but a (400 kPa). Kataoka et al. (1983) and Tatterson et al. (1977) presented a
drawback comes forward in these mentioned models as they lack of study on drop distributions related to entrainment for air–water
systems.
As the liquid film and the gas phase are in contact at the film
 Corresponding author. Tel.: + 47 73 594132; fax: + 47 73 594080. surface, the shear will be given in terms of the velocity difference.
E-mail addresses: patruno@nt.ntnu.no (L.E. Patruno), Given the velocity difference is above a certain threshold,
hugo.jakobsen@chemeng.ntnu.no (H.A. Jakobsen). instabilities (i.e. waves) will be created. Further on, the gas can

0009-2509/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2009.10.011
 ARTICLE IN PRESS
1408 L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414

entrain liquid mass from the crests of these waves and create Chemical Engineering. The rig basically consists of a high pressure
droplets which will be carried over onto the gas phase. Azzopardi cell which is continuously fed with a gas and a liquid flow through
(1997) gives an extensive review on available data of droplet–wave separated ducts. The cell acts as a mixture chamber and disposes
interaction, being noticeable the redundancy of droplet entrainment the gas–liquid mixture to a damp tank. The gas is vented while the
data for air–water systems at atmospheric pressure and showing liquid is recirculated. All the following descriptions are based on
the lack of data in the low surface tension-hig pressure range. the process and instrumentation diagram (P&ID) shown in Fig. 1.
Moreover, an already existing droplet could either be unstable The gas line is fed by a gas bottle (TANK 1) and the pressure is
and break into smaller droplets, or collide with a neighbor droplet reduced and regulated with a pressure regulator. The gas flow can
and coalesce. Jakobsen (2008) and Patruno et al. (2009a) outline be adjusted with a needle valve and it is measured with (FI 1). The
the mathematical description of such behaviors within the temperature is measured after the flow measuring device with a
population balance framework. resistance thermometer (TEMP SENSOR 1) which is a three-wire
This paper provides a droplet size distribution function for PT100. After a last pressure measurement (PI 1), the gas is injected
drops created due to entrainment of a mixture of Exxsol D60TM (a into the high pressure cell. This gas line is pressurizing both the
light hydrocarbon) and nitrogen at slightly high pressure cell and the damp tank (DAMP TANK).
(800 kPa). This case dependent function is implemented in a An approximate liquid inventory of 2.5 L is stored in a vessel
suitable mathematical framework showing a possible coupling (LIQUID TANK) pressurized by a second gas bottle (TANK 2)
mechanism of the entrainment phenomena with an in detail containing the same gas as TANK 1. The pressure is reduced from
dispersed phase description. TANK 2 to the LIQUID TANK with a set of pressure regulators,
measured at PI 2 and kept slightly above the pressure indicated in PI
2. Experimental setup and methodology 1. The liquid flow is regulated with a needle valve. A filter for 40 mm
particles is placed upstream the liquid flow measuring device (FI 2).
This section gives an overview of the experimental setup and The liquid temperature is measured with a second three-wire PT100
equipment which was used to perform the measurements. (TEMP SENSOR 2) and then injected into the high pressure cell.
As indicated, the pressure in PI 2 is set higher than the pressure
2.1. Circulation rig in PI 1 in order to have the liquid flowing inside the cell;
nevertheless, the pressure difference is to be maintained as small
The circulation rig is placed at the Laser Lab-II facility at the as possible (2002300 kPa) in order not to influence the gas–liquid
Norwegian University of Science and Technology, Department of equilibrium. Further details are given in Appendix A.2.

Fig. 1. Process and instrumentation diagram of the rig used to study liquid entrainment in gas–liquid systems.
 ARTICLE IN PRESS
L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414 1409

The gas–liquid mixture is flushed through the bottom end of the 2:5  103 m2 and the equivalent hydraulic diameter is
high pressure cell and disposed into the DAMP TANK, which acts as 50  103 m. The pipe length is 0.5 m to minimize pressure
a separator/gravity coalescer. The pressure in the DAMP TANK is related safety issues. There are six openings through the cell’s
kept constant with a backpressure regulator which is also used as a wall: three for pipings, one for instrumentation and two for
second measurement the tank’s pressure. The gas is vented to the viewing glasses.
atmosphere while the liquid is stored in the DAMP TANK. The overall geometry of the cell is presented in Fig. 2 and the
When the LIQUID TANK runs out of liquid, an offline liquid most relevant dimensions are shown in Fig. 3.
transfer between the DAMP TANK and the LIQUID TANK is The liquid is introduced through a small pipe (internal
possible. Since the composing substances of the liquid could diameter of 2:9  103 m) from one of the front sides which is
evaporate to the gas phase, and since the gas is continuously directed to the bottom of the cell where a channel (10  103 m
renovated while the liquid is re-used, the liquid could change its wide and 3  103 m high) will guide it to the exit hole at the end
composition in time. In order to avoid strong changes in of the pipe. As the cell is slightly tilted, the channel allows the
composition the liquid is re-used only 4 times. An assessment of liquid to run through the cell as a thin film. No accurate
the evaporation effects is reported in Appendix A.2. measurement of the film height is done while it is visually
All pressures, temperatures and the liquid flow rate are estimated in 3 72  103 m.
displayed locally, while the gas flow rate is displayed by a PC. The gas is introduced through a small pipe (internal diameter
The gas flow meter used is a Bronkhorst mass meter, model F-113- of 2:9  103 m) from the top side of the cell, and a 903 bend
AC-1M0-AGD-44-V, calibrated for nitrogen at 500, 1000 and places it parallel to the liquid film. The gas injection point is the
1500 kPa. The liquid flow meter is a turbine meter, using a KEM plain-cut pipe, and there are no nozzles in use. The distance
VTE frequency pulse amplifier. between the gas injection pipe and the cell bottom (contact
The whole experimental setup (that is pipings, tanks and cell) height) is 3:0 7 0:2  103 m.
is flushed with gas from TANK 1 in order to clean it from other gas Two lateral openings in the cell allow for the multi-phase flow
contaminants. visualization along the pipe. The square of the opening is 40 
103 m high and 200  103 m wide. A pressure certified glass is
2.2. High pressure cell placed covering each of the openings.
No droplet supplies (such as spray nozzles) are present
The high pressure cell is a squared-section metallic pipe with a upstream inside the high pressure cell, making the liquid
square side length of 50  103 m (50 mm), hence the pipe area is entrainment from the film the only droplet source.

Fig. 2. Side, top view and cross section of the high pressure test section acting as a mixture chamber for the gas–liquid system.
 ARTICLE IN PRESS
1410 L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414

Fig. 3. Most relevant dimensions of the high pressure test section.

2.3. Image analysis—droplet identification parameters as constant as possible in order to check for
reproducibility, the numerical value of the physical properties
The optical part of the experimental setup consists of a 100 W are given in Table 1. The selected target pressure was 800 kPa
white light source illuminating through the high pressure cell (gauge) as it is the value placed in the middle of the rigs
towards a high speed camera. The used camera is a Photron capabilities, allowing these measurements to be center points for
FASTCAM 1024PCI doted with magnification lenses achieving a further perturbation experiments. The atmospheric pressure was
resolution of 11:61 70:06 mm=pixel. found to be 102.2 kPa.
The working principle is based on the fact that when a droplet The gas flow was set to 45 nL/min, only reaching approximate
is placed in the optical path between the light source and the values in practice. The liquid flow was set to 0.1 L/min, yielding an
camera, the camera will register the shadow of the droplet. The easy to achieve setpoint. The flow pumping design, which actually
high magnification reduces the depth of view of the camera giving has no pumps at all, proved to be a good solution to get rid of the
droplets in and out of focus. common pulsations generated by pumps. The economic costs of a
After the camera has taken a set of pictures of the droplets, high pressure pump were replaced by the cost of the gaseous
these pictures are analyzed and the droplet edges are detected. nitrogen needed to pressurize the tanks.
When an edge is diffuse (outcome of an out of focus droplet), the
information is discarded. Every accepted edge datum is consider
to be a droplet, thus its diameter, position and elongation are 3.1. Image analysis
calculated and stored.
A paper by Rodriguez and Shedd (2004) shows the applicability of With the above commented setpoints, the high speed camera
this measurement technique when measuring gas entrainment onto was set to 1000 frames per second with a shutter speed of 1/
liquid films. Further on, a previous use and validation of the droplet 122 000 s in order to avoid motion induced blur in the pictures.
identification method is outlined by Patruno et al. (2009b). The picture size was set to the maximum allowed by the camera
hardware, namely 1024  1024 pixel. The frames were taken close
to the gas outlet in order to reduce the shape changes of the PDF
3. Measurements and results induced by other physical means such as breakage, coalescence
and gravitational deposition.
Three different runs were performed in order to test the rig and As an example, Fig. 4 shows one of the images captured by the
acquisition system’s capabilities. Efforts were made to keep the high speed camera. The picture’s full color (RGB) are real colors
 ARTICLE IN PRESS
L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414 1411

Table 1
Physical properties for the gas and the liquid phases.

Run Qg (nL/min) Ql (L/min) P (kPa) Tg (3 C) s^ (mN/m) r^ g (kg=m3 ) r^ l (kg=m3 ) m^ g (cP) m^ l (cP)

1 48.8 70.20 0.100 70.001 790 710 15.67 0.05 24.54 9.256 743.4 1:801  102 1.2
2 43.5 70.20 0.100 70.001 8007 0.10 16.27 0.05 24.49 9.236 743.4 1:814  102 1.2
3 43.07 0.20 0.104 7 0.001 790 710 18.47 0.05 24.31 9.165 743.4 1:814  102 1.2

The calculated parameters are noted with an accent (i.e. s^ ) and no related error is given.

100 100
200 200
Image height (pixels)

300 300

Image height (pixels)

400 400
500 500
600 600
700 700
1 mm 1 mm
800 800
900 900
1000 1000
100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000
Image width (pixels) Image width (pixels)
Fig. 4. Raw image captured by the high speed camera. Fig. 6. Matlab representation of the image. Two blobs representing droplets of
13.6 mm (left) and 21.4 mm (right).

100 gets rid of the undesired impurities but also provides a better
contrast between the blobs and the background.
200 An ad hoc Matlab script was written and used to analyze the
300 complete series of pictures, such as Fig. 5, and recognize each of
Image height (pixels)

the blobs. Blobs were interpreted as droplets after the sharpness

400
(in focus droplets) and the elongation (close to spherical droplets)
500 criteria were satisfied. Assuming the recognition script has a size
independent efficiency, every accepted blob will contribute
600
positively to the statistics. The droplet diameter is calculated as
700 the Feret mean diameter of each of the accepted blobs. An image-
1 mm representation of the accepted data is shown in Fig. 6. We can see
800
that from the initial amount of blobs, only two were
900 acknowledged as droplets. In this case, the drop on the right
hand side has a diameter of 21:4 mm and the one on the left hand
1000
size has a diameter of 13:6 mm.
100 200 300 400 500 600 700 800 900 1000
Probability density functions (PDFs), which can be interpreted
Image width (pixels) as normalized histograms, were created for each of the experi-
Fig. 5. Processed image; original image converted to 8 bits and background ments distinguishing the droplet data with a functionality with
eliminated. respect to the droplet diameter.

3.2. Entrainment shape function

and, even though it is being illuminated by a white light source,
the image’s reddish tone is due to the high speed shutter; many The PDFs for each of the runs are shown in Figs. 7–9. In order to
dark blobs appear in the picture, some corresponding to in-focus provide the reader with a reference function a lognormal
drops and others to out-of-focus drops or even drops sticked to probability function was fitted to the data (see for example
the glasses’ surface. The gray regions of the picture are due to a Kataoka et al., 1983 or Xie et al., 2004). Ninety five percent
film created on the inner side of the viewing ports (glasses). confidence intervals are also shown in the graphs. Note that there
The image is then converted to 8 bits (unsigned) and is no emphasis on the physical meaning of a lognormal
background measurement is subtracted from the original picture. distribution reproducing the experiments, but rather given as a
The resulting image is shown in Fig. 5. This new image not only basis for further work.
 ARTICLE IN PRESS
1412 L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414

order to provide prospective users of the presented results with a

formula to include in the mathematical frameworks, a lognormal
fit was done for the three cases and the numerical values of
the parameters describing the lognormal function are given in
Table 2; the parameters are suitable to be used along with
" #
A lnðx=xc Þ2
wðxÞ ¼ w0 þ pffiffiffiffiffiffi exp  ð1Þ
2pox 2o2

From Table 2 we see that the droplet mean diameter (xc ) has been
measured to be 64 7 7 mm. The correlation for the droplet
diameter given by Tatterson et al. (1977) yields a diameter
which decreases with increasing gas velocity. Even when
considering the best case scenario given by a gas velocity of
14 m/s (this value is the gas velocity inside the inlet pipe), the
calculated mean diameter is 309 mm. This overestimation could
find its explanation in the fact that the correlation was derived for
an air–water system which presents a higher surface tension.

Fig. 7. Experimental results and curve fitting for the first run.
3.3. Usage of the data

One of the possible mathematical frameworks used to describe

the interaction between the droplets and the film is the one
presented by Patruno et al. (2009c, d). The model makes use of the
population balance framework to describe, in a statistical manner,
the droplet phase and its interaction with the film phase and has
already been validated for the net mass transfer.
Neglecting the droplet breakage, coalescence and deposition it
is possible to arrive to a simplified case of the previously
commented model. The mass conservation for the droplet and
film phases is given by
@
8
> ud fd ðx; zÞ ¼ wðx; zÞbðzÞff ðzÞ in O ¼ ðxmin ; xmax Þ  ðzmin ; zmax Þ
@z
>
>
>
>
< @
>
>
u f ðzÞ ¼  bðzÞff ðzÞ in Oz ¼ ðzmin ; zmax Þ
> @z f f
fd ðx; z ¼ zmin Þ ¼ 0 on Gx ¼ ½xmin ; xmax 
>
>
>
>
: ff ðz ¼ zmin Þ ¼ f 0
>
on Gz ¼ ½z ¼ zmin 
>
f

ð2Þ

Fig. 8. Experimental results and curve fitting for the second run. in which ud and uf are the droplet and film velocity and fd and ff
are the droplet and film mass distributions, respectively. wðx; zÞ is
the entrainment spectrum and bðzÞ the entrainment kernel.
The first equation, the droplet equation, is valid inside the
domain O ranging from the smallest droplet (xmin ) to the biggest
droplet (xmax ) in the system, and from the pipe’s initial position
(zmin ) to the final position (zmax ). The second equation, the film
equation, is independent of the droplet size and it is only
modelling the film phase, then the equation is valid inside Oz
constrained to the z variable.
The boundary conditions, third and fourth equations, are given
in Gz ¼ zmin both for fd and ff . As fd is a two variables equation
evaluated at z ¼ zmin , the equation is valid inside Gx , meaning for
all x.
Assuming that the entrainment distribution function is
independent of the position (wðx; zÞ ¼ wðxÞ), the data fitted to
Eq. (1) can be taken as the entrainment shape function and further
implemented in Eq. (2). As Eq. (1) is normalized, mass will be

Table 2
Parameters for the lognormal fitting of the three experimental runs.

Fig. 9. Experimental results and curve fitting for the third run. Parameter Value Error

w0 0.000 7 0:002
An application of the data obtained from the experiments is xc 64 77
o 0.6 7 0:1
the use of the normalized PDF as entrainment shape function for A 3.7 7 0:4
the models previously presented by Patruno et al. (2009c, d). In
 ARTICLE IN PRESS
L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414 1413

700 The measured droplet mean diameter differs from the one
predicted by existing correlations. This difference could be due to
600 the gas liquid pair systems used for the experiments. Not a mean
diameter but rather a drop size distribution was presented,
500 enabling a more in depth description of the dispersed system
Droplet
total mass (kg/m3)

Film and the interaction with the continuous phase.

400 The capabilities of the experimental facility, as well as the
Total
suitability of the chosen measuring system, were tested in three
300
different runs at a center point for the tunable physical values. As
200 the presented entrainment shape function is case and system
dependent, running similar experiments with different operating
100 conditions will open the possibility for exploring how the droplet
size distribution is affected by different flow rates, pressure and
0 surface tension.
Further studies are being carried on in order to investigate the
−100 changes in the entrainment shape function due to spatial effects
0 0.1 0.2 0.3 0.4 0.5 and changes in the fluids properties which could help explaining
z (m) the differences seen between experiments and already existing
correlations.
Fig. 10. Mass evolution for the droplet and the film phases and total mass
conservation.

Notation

A normalization constant, dimensionless

fd droplet size distribution, kg=m3 mm
ff film distribution, kg=m3
h contact height, m
P pressure, kPa
Qg gas flow rate, nL/min
Ql liquid flow rate, L/min
T temperature, 1
ud droplet velocity, m/s
uf film velocity, m/s
xN2 nitrogen molar fraction in liquid, mol%
yN2 nitrogen molar fraction in gas, mol%
z pipe length, m
Greek symbols
bðzÞ liquid entrainment kernel, 1/s
mg gas viscosity, cP
Fig. 11. Droplet size distribution as a function on the droplet size (x) and the pipe ml liquid viscosity, cP
length (z).
x droplet diameter, mm
xc median of the lognormal distribution, mm
rg gas density, kg=m3
conserved and the droplets resulting from the entrainment of
rl liquid density, kg=m3
mass from the liquid film will be forced to follow a lognormal
distribution. When the entrainment is the dominating mechan- s surface tension, mN/m
ism, the whole droplet phase itself will look closer to a lognormal wðxÞ entrainment shape function, 1/mm
distribution. o standard deviation, dimensionless
Given the fact that no droplet exists before the entrainment
phenomena occurs, fd ðx; z ¼ 0Þ ¼ 0, and the mass is conserved as a
whole between the liquid and the droplets, the modelling results Acknowledgments
are shown in Fig. 10. The net droplet mass increases and the film
mass decreases due to entrainment, while the total mass is The PhD fellowship (Patruno, L.E.) and post-Doc fellowship
constant throughout the pipe length. (Marchetti, J.M.) financed by the Research Council of Norway
The droplet size distribution is expected to grow due to the Petromaks Programme through the (HiPGLS) project are gratefully
mass transfer rate, bðzÞ, and to take the shape given by wðxÞ. The appreciated. The authors would also like to thank Hui Nie Nilssen
results from the model are presented in Fig. 11. (MSc) for her help with the calculation of the surface tension.

4. Conclusions Appendix A

The droplet size distribution generated by the liquid entrain- A.1. Fluid’s properties
ment was successfully measured for a mixture of light hydro-
carbons commercially know as Exxsol D60TM as the liquid phase The data presented in this paper were obtained using a
and pure nitrogen as the gas phase at a pressure of 800 kPa mixture of Exxsol D60TM for the liquid phase and pure nitrogen as
(gauge). the gas phase. Exxsol D60TM is convenient to use within
 ARTICLE IN PRESS
1414 L.E. Patruno et al. / Chemical Engineering Science 65 (2010) 1407–1414

Table 3 the gas phase is a stream of pure nitrogen, the molar fraction of
The composition of the commercial mixture Exxsol D60TM used as liquid phase for nitrogen in the gas and in the liquid streams should give a clear
the experiments. idea of the amount of nitrogen dissolved into the liquid phase, and
the evaporation of hydrocarbons to the gas phase. Table 4 gives
Component (dimensionless) Fraction (mol%)
the molar fraction for nitrogen in both of the phases for each of
nC9 (n-Nonedecane) 1.2424 the experimental runs. As it can be read from the two last
nC10 (n-Decane) 20.1849 columns, the amount of nitrogen dissolved to the liquid phase is
nC11 41.9108
around 1.3 mol% and the gas phase can be taken to be pure
nC12 28.9341
nC13 7.7278
nitrogen as the composition is around 99.99 mol% nitrogen. This
also means that the amount of liquid that was evaporated is
considerer insignificant for this present work.

Table 4 References
Calculated phase split for nitrogen in each of the runs.

Run Qg (nL/ Ql (L/min) P (kPa) T (1C) xN2 yN 2 Ambrosini, W., Andreussi, P., Azzopardi, B.J., 1991. A physically based corre-
lation for drop size in annular flow. International Journal of Multiphase Flow
min) (mol%) (mol%)
17, 497–507.
Austrheim, T., Gjertsen, L.H., Hoffmann, A.C., 2008. An experimental investigation
1 48.8 7 0.20 0.1007 0.001 790 710 20.4 7 0.05 1.31409 99.9943
of scrubber internals at conditions of low pressure. Chemical Engineering
2 43.5 7 0.20 0.1007 0.001 8007 10 20.9 7 0.05 1.3096 99.9940 Journal 138, 95–102.
3 43.0 7 0.20 0.1047 0.001 790 710 20.9 7 0.05 1.3069 99.9931 Azzopardi, B.J., 1997. Drops in annular two-phase flow. International Journal of
Multiphase Flow 23, 1–53.
Chatzikyriakou, D., Walker, S.P., Hewitt, G.F., Narayanan, C., Lakehal, D., 2009.
Comparison of measured and modelled droplet-hot wall interactions. Applied
Thermal Engineering 29 (7), 1398–1405.
Jakobsen, H.A., 2008. Chemical Reactor Modeling—Multiphase Reactive Flows.
laboratory experiments as it contain very low levels of hazardous
Springer, Berlin, Germany.
air pollutants, and has a relatively low vapour pressure. It is not an Kataoka, I., Ishii, M., Mishima, K., 1983. Generation and size distribution of droplet
exact composition since it is a paraffin distillation cut, but the in annular two-phase flow. Journal of Fluids Engineering 105, 230–238.
Kataoka, I., Ishii, M., 1982. Mechanism and correlation of droplet entrainment and
composition does not vary a lot and the supplier provides typical
deposition in annular two-phase flow. NUREG/CR-2885, ANL-82–84.
values of its fluid properties (Austrheim et al., 2008), which are Miqueu, C., Mendiboure, B., Graciaa, A., Lachaise, J., 2005. Modeling of the surface
given in Table 3. Additional information regarding the tension of multicomponent mixtures with the gradient theory of fluid
hydrocarbons used in the experiments can be found in Whitson interfaces. Industrial & Engineering Chemistry Research 44 (9), 3321–3329.
Morud, J.C., 2007. Dilute gas–liquid flows with liquid films on walls. Progress in
(1994). Computational Fluid Dynamics 7, 170–175.
The properties of the system nitrogen-Exxsol D60TM were Paleev, I.I., Filipovich, B.S., 1966. Phenomena in liquid transfer in two-phase
calculated using the commercial software Aspen HYSYS v2006.5. dispersed annular flow. International Journal of Heat and Mass Transfer 9,
1089–1093.
The surface tension and liquid density in particular were found to Pan, L., Hanratty, T.J., 2002a. Correlation of entrainment for annular flow in vertical
be close within a 10% to the corresponding ones given by pipes. International Journal of Multiphase Flow 28 (3), 363–384.
Austrheim et al. (2008) and within 1% to the ones calculated Pan, L., Hanratty, T.J., 2002b. Correlation of entrainment for annular flow in
horizontal pipes. International Journal of Multiphase Flow 28 (3), 385–408.
using the Gradient Theory (Miqueu et al., 2005). Patruno, L.E., Dorao, C.A., H.F., Jakobsen, H.A., 2009a. Analysis of breakage
As the gas and liquid streams will mix inside the test section at kernel for population balance modelling. Chemical Engineering Science
different pressures, the physical properties will vary. The 64(3), 501–508.
Patruno, L.E., Dorao, C.A., Dupuy, P.M., Svendsen, H.F., Jakobsen, H.A., 2009b.
equilibrium properties are calculated with HYSYS using the
Identification of droplet breakage Kernel for population balance modelling.
Peng–Robinson equation of state (Whitson and Brulé, 2000). Chemical Engineering Science 64 (4), 638–645.
Table 1 shows a set of measured and calculated physical proper- Patruno, L.E., Dorao, C.A., Svendsen, H.F., Jakobsen, H.A., 2009c. On the modelling of
droplet-film interaction considering entrainment, deposition and breakage
ties for each of the experiments.
processes. Chemical Engineering Science 64 (6), 1362–1371.
Patruno, L.E., Marchetti, J.M., Dorao, C.A., Svendsen, H.F., Jakobsen, H.A., 2009.
A.2. Two-phase split calculation Modelling of droplet entrainment in annular two-fluid three-phase dispersed
flow. In: 13th International Topical Meeting on Nuclear Reactor Thermal
Hydraulics (NURETH-13), N13P1056.
As stated previously, the question remains on whether a Rodriguez, D.J., Shedd, T.A., 2004. Entrainment of gas in the liquid film of
system consists of one or two phases at a given temperature and horizontal, annular, two-phase flow. International Journal of Multiphase Flow
30, 565–583.
pressure. In addition, we need to calculate the composition of Sawant, P., Ishii, M., Michitsugu Mori, M., 2008. Droplet entrainment correlation in
each of the phases given the overall composition. According to vertical upward co-current annular two-phase flow. Nuclear Engineering and
Whitson and Brulé (2000) there are many algorithms for solving Design 238, 1342–1352.
Tatterson, D.F., Dallman, J.C., Hanratty, T.J., 1977. Droplet size in annular gas liquid
the two phase split calculation, either by satisfying the equal-
flows. A.I.Ch.e. Journal 23 (1), 68–76.
fugacity and material balance or by minimizing the Gibbs free Whalley, P.B., 1990. Boiling Condensation and Gas–Liquid Flow. Oxford University
energy of the mixture. Press, Oxford, ISBN: 0-19-856234-9.
Whitson, C.H., Brulé, M.R., 2000. Phase Behavior. In: SPE Monograph Series, ISBN:
Assuming all but the chemical potential force to be negligible,
1-55563-087-1.
the criterion of thermodynamic equilibrium is that the chemical Whitson, C.H., 1994. Petroleum Engineering Fluid Properties Data Book. Compila-
potential of a given component in one of the phases is equal to the tion by the Author within Reservoarfluider og strømning—Norwegian
chemical potential of the same component in the other phase. University of Science and Technology.
Xie, H., Koshizuka, S., Oka, Y., 2004. Numerical simulation of liquid drop deposition
It is possible to solve iteratively for the composition of each of in annular-mist flow regime of boiling water reactor. Journal of Nuclear Science
the phases using the Rachford–Rice procedure. For our case, since and Technology 41 (5), 569–578.