Multiphase Flow

Two-phase ow modelling within expansion and
contraction singularities
V. G. Kourakos
1
, P. Rambaud
1
, S. Chabane
2
, D. Pierrat
2
and J. M. Buchlin
1
1
von Karman Institute for Fluid Dynamics, Brussels, Belgium,
2
Centre Technique des Industries M ecaniques, Nantes, France.
Abstract
An experimental study is performed in order to describe the single- and two-phase (air-water)
horizontal ow in presence of pipe expansion and contraction. Three types of singularities are
investigated; smooth convergence and sudden and progressive enlargement. The opening angles
for progressive singularities are 5, 8, 9 and 15 degrees. The surface area ratios tested are =0.43,
0.64, 0.65 and 1.56. Bubbly ow is the dominant ow regime that is investigated for volumetric
quality up to 30 %. The pressure distribution for both single and two-phase is examined versus
axial position. For expansion geometries, it is found that the smallest the enlargement angle,
the largest the recovery pressure for the same ow conditions; the pressure drop caused by the
singularity is higher in the case of a sharper expansion. The comparison of the experimental
results to published models leads to proposed corrective coefcient for Jannsens [1] correlation.
Flow visualization is also performed; the ow patterns downstream the different singularities are
identied in each conguration and plotted in Bakers (1954) map for horizontal ow.
Keywords: Two-Phase Flow, Singularity, , Sudden expansion, Contraction, Pressure Drop,
Bubbly Flow, Flow Visualization
1 Introduction
Two-phase ow can be frequently met in nuclear, chemical or mechanical engineering where
gas-liquid reactors, boilers, condensers, evaporators and combustion systems are often used.
The presence of geometrical singularities in pipes may affect signicantly the behavior of two-
phase ow and subsequently the resulting pressure drop. Therefore, it is an important subject of
investigation in particular when the application concerns industrial safety valves. The studies of
two-phase ow in straight pipes existing in the literature are numerous. However, investigations
of two-phase ow in divergence, convergence, bends and other types of singularities are rather
sparse. The aim of studying these geometries is to nd how these geometrical accidents inuence
the two-phase ow pattern and pressure distribution. In particular, the understanding of the ow
in such basic geometries can lead to a better design of safety systems.
Some of the authors that have analyzed two-phase ow in expansion geometries are Jannsen
[1], McGee [2], Chisholm [3], Chisholm [4] and Lottes [5]. Correlations for estimating the
pressure change in two-phase ow in this type of piping geometry are reported by these authors.
These correlations can be extracted from the conservation equations applied downstream of the
sudden expansion. The equations used take into account different parameters of the geometry
and the ow such as surface area ratio , mass quality x and mass velocity G. More recently,
Aloui [6], Aloui [7], Schmidt [8], Hwang [9], Ahmed [10] and Ahmed [11] have evaluated
the pressure change in a sudden expansion duct. Moreover, some of them (Aloui [6]; Ahmed
[10]) have measured the bubble velocities and local void fraction to characterize the ow regime
downstream the singularity. The lack of studies in progressive enlargements in two-phase ow in
the literature makes such an investigation more appealing. In this paper, progressive contraction
and divergence geometry of different opening angles is considered. The latter is compared to
the case of sudden expansion. The two uids are air and water in isothermal conditions. The
volumetric quality of the air varies from 0-30 % and bubbly ow is the dominant regime. Four
surface area ratios; =0.43, 0.64, 0.65 and 1.56 are tested. The opening angles for the case of
progressive singularities are 5, 8, 9 and 15 degrees. The Reynolds number Re of the liquid
is ranging from 810
4
to 2310
4
. The determination of the recovery pressure for each of the
aforementioned geometries is the one of the main objectives of this investigation.
von Karman Institute for Fluid Dynamics-Centre Technique des Industries Mcaniques
1
Singularities
Flow
d
D
Reattachment length-L/d
d

D
Reattachment length-L/d
B)
Flow
A)
Figure 1: A)Progressive expansion of different opening angles-reattachment length L/d.
B)Sudden expansion-reattachment length L/d.
In Figure 1, the two different types of expansion geometries tested in this paper are presented.
Figure 1A shows the divergent pipe with opening angle and Figure 1B the sudden expansion.
The normalized reattachment length L/d, noticed in Figure 1, denotes the eventual recirculation
zone. In the case of convergence geometry, a contraction region can be observed; a vena contracta
is formed in the pipe downstream the singularity.
2 Experimental facility and conditions
2.1 Experimental facility
A schematic of the horizontal air-water ow facility used for the present study is shown in Figure
2. A centrifugal pump (1) with a maximum ow rate of 65 m
3
/h is sucking water from a reservoir
and is controlled with a frequency inverter. During the experiments, an air release valve (11)
connected to the tank is kept continuously open to the atmosphere to avoid bubbles entering
the circuit. A by pass valve (12) is used to prevent facility from water hammer phenomenon. A
temperature sensor is placed in the reservoir, to monitor the temperature for each measurement.
Two electronic ow meters are used to measure the water ow rate (2 and 3); their maximum
capacity 12 m
3
/h (3) and 32 m
3
/h (2), respectively. In the case of the desired maximum ow
rate, which is 40 m
3
/h, the two ow meters are used in series. A bourdon tube pressure gauge
(4) is placed upstream in the pipe to know the wall static pressure relative to atmosphere. This
indication helped to prevent excessive pressure that could lead to a breaking of the test section
(made in Polymethyl Methacrylate, PMMA). Moreover, the pressure has to be high enough to
allow the necessary purging of the pressure transducers. Therefore, the pressure is held constant
at around 200 kPa. The setup has an upstream calming section (5) consisting in stainless steel
pipe length of 50 diameters (50d). That assures a fully developed ow after the bend. Close to
the test section, the injection of the air is performed through a gas injector (6) as indicated in
Figure 2. A regulation valve (7) controls the air that is supplied from a compressor. The air ow
rate is measured by an electronic mass ow meter (8). The design and the positioning of the air
injection device are such that uniform bubbly ow is produced at the inlet of the test section. It
is found that the most suitable distance for the air injection is 20 pipe diameters upstream the
singularity. After the test section, a heat exchanger (9) is placed for maintaining the temperature
constant at around 21

C during the experiments. A draining valve is also located at the bottom
of the reservoir. Finally, a pressure regulation valve (10) controls the pressure of the system.
4
Sketch facility-FINAL
1 Pump
2 Big electronic water flow meter
3 Small electronic water flow meter
4 Bourdon tubepressuregauge
5 Calming length
6 Air injector
7 Regulation valve
8 Electronic air mass flow meter
9 Heat exchanger
10 Pressureregulation valve
11 Air releasevalve
12 By pass valve
T Temperaturemeasurement
Water tank
Test section
Inverter
T
PP
Water discharge
Compressed air
1
2
3
4
5
6
7
8
9
11 11
12 12
Figure 2: Schematic of the experimental facility.
A detailed view of the test section is presented in Figures 3, 4 and 5. The case of a DN
40/65 (=0.43) divergent section with an opening angle of 8
is exemplied. At each section

of measurement, four pressure taps are placed with an angle of 45
between them as shown

in Figure 3. Thus, any three dimensionality of the ow could be identied from pressure
measurement. The 4 taps are named as A, B, C and D according to Figure 3. Figure 4 depicts
an overview of the test section. The setup is built in PMMA to allow optical access. Pressure
taps are placed along the tube in several points as is shown in Figure 5. The distance between
pressure holes is normally equal to 1 tube diameter but becomes smaller when approaching the
singularity. The pressure taps are also more dense inside and downstream the singularity. This
allows better tracking of the ow behavior in the singularity. Pressure distribution is measured
upstream and downstream the divergence. The test matrix is summarized in Table 1.
7
Sketch test section
Aluminumtable
4 pressuretaps
(45anglebetween them)
Aluminumtable
4 pressuretaps
(45anglebetween them)
A B
C D
Figure 3: Pressure taps placed in 4 different
points of the tube with 45

between
them.
7
Sketch test section
Aluminumtable
4 pressure taps
(45angle between them)
Aluminumtable
4 pressure taps
(45angle between them)
A B
C D
Figure 4: Overview of the test section.
6
Sketch test section
Figure 5: Detailed view of the test section with the pressure taps and their position.
2.2 Flow conditions-measurement devices
The ow conditions of the experimental campaigns are listed in table 2. Table 2 presents the test
conditions for the pressure measurements and for ow visualization. It should be pointed out
that the Re
L1
number of the liquid is based on the upstream pipe diameter d. For the comparison
between single and two-phase ow, Re
L1
is kept constant. This is obtained by adjusting the
water ow rate when increasing the air to reach a higher volumetric quality . Consequently,
we can assume that the total mass ux is constant since the mass of the air compared to that of
water is negligible. Differential pressure transducers of type Rosemount are used. The uncertainty
associated to the pressure transducers varies from a minimum of 0.35% to a maximum of 0.75,
depending on the range of the measurement (100-20% of the scale of the range respectively). To
obtain the best accuracy possible, 4 different pressure transducers are selected:
1. Calibrated at 0-1.6 kPa
2. Calibrated at 0-4 kPa
Every transducer is used in range that gives the best accuracy in all the conditions covered.
Prior to the measurements, predictions of regular pressure drop are performed by means of
Blasius and Colebrook-White formulas for single-phase and Lockhart and Martinelli (1949) [12]
for two-phase ow. Thus, this P estimation allows the selection of the appropriate pressure
Table 1: Different test cases studied.
Singularity d
1
[m] D
2
[m] [-] L
s
[m] L
s
/d
1
Angle [
]
Smooth contraction 0.04 0.032 1.56 0.025 0.63 9
Divergence 0.032 0.04 0.64 0.025 0.78 9
Divergence 0.041 0.0627 0.43 0.041 1 15
Divergence 0.041 0.0627 0.43 0.07503 1.83 8
Divergence 0.041 0.0627 0.43 0.1238 3 5
Divergence 0.0627 0.078 0.65 0.0529 0.8 8
Sudden expansion 0.041 0.0627 0.43 - - 90
Sudden expansion 0.0627 0.078 0.65 - - 90
Table 2: Upstream conditions for pressure measurements and ow visualization.
d
1
[m] Fluid

Q [l/s] J [m/s] [%]

G [kg/m
2
s] Re
L1
10
4
Flow regime
0.032
Water 2 2.5
1-40
2500 9 Laminar Min
Water 4.7 5.8 5850 20 Turbulent Max
Air 0.017 0.02 0.03 0.005 Laminar Min
Air 1.8 2.2 2.61 0.46 Turbulent Max
0.041
Water 2.3 1.8
5-30
1750 8 Turbulent Min
Water 7 5.4 5300 23 Turbulent Max
Air 0.4 0.3 0.38 0.09 Laminar Min
0.0627
Water 6 1.9
5-25
1950 13 Turbulent Min
Water 10.5 3.4 3400 23.5 Turbulent Max
Air 0.4 0.1 0.15 0.05 Laminar Min
transducers for each test. Additionally, for the prediction of the singular pressure change in
single-phase, the coefcients given by Idelcik [13] are used. The uncertainty related to the ow
rate measurements varies from a minimum of 0.5 % to a maximum of 1.10%. The temperature
variation during the experiments is of the order of 4

C with an average value of 21
C. Although a heat exchanger is used for reducing this variation, a small uctuation of the
temperature could not be avoided. A variation of 5
C will change and by 0.1 and 11

% respectively. Therefore, a correction of the liquid density and viscosity is performed. The
sampling frequency of the measurements is f
sampling
=2 Hz and the acquisition time for each
measurement point is t
acq.
=1 minute with the aim of assuring a more accurate average. In some
cases (for sudden and progressive enlargement of =0.65), a higher uctuation of the signal is
observed; in this occurrence an acquisition time of 2 minutes is chosen.
3 Results-discussion
3.1 Pressure measurements
One of the main objectives of the study is the determination of the pressure distribution through
the different singularities. Figure 6 indicates how the measurements are performed and how
the singular single and two-phase pressure change is determined (the case of divergence is
chosen). As the graph of Figure 6 shows, following a normal decrease upstream the geometrical
accident, the pressure will increase to a maximum value inside the divergent section and will
start decreasing after a certain length in a regular way. We can split the whole phenomenon in
three regions; the upstream fully developed ow, the transitional region with a recirculation zone
and the downstream fully developed ow. The length of the transitional region varies with Re
L1
,
, and the type of the singularity. In all the tests, the measurement of the regular and singular
static pressure changes is referred to the pressure measured at 10d upstream the singularity
(Figure 6). The singular pressure change P can be nally determined by extrapolating the
regular static pressure drop from the start of the singularity to the reattachment point. Since the
points downstream the singularity are not enough to obtain fully established ow, the regular
pressure drop is calculated by means of the Blasius and Colebrook-White formulas for single-
phase and the model of Lockhart and Martinelli [12] for two-phase. The nal singular pressure
change is calculated by a simple summation of these three terms (|P
regularmeasured
| +
|P
singularmeasured
| +|P
regularcalculated
|). The reattachment length is determined as the
location of the maximum recovery pressure.
15
Singularity
P
reference
P
max
P [mbar]
Axial position [z/d]
P
regular
(measured)
P
regular
(calculated)
z/d=0
P
singular
(measured)
P
SINGULAR
-FINAL
P=0
Flow
Fully developed flow
Inlet
Transitional flow Fully developed flow
Outlet
Results paper
Figure 6: Explanation of the way to determine the singular pressure change in expansion
geometry.
3.1.1 Expansion singularities
3.1.1.1 Sudden expansion In Figure 7, the two-phase pressure change along the pipe and the
singularity is plotted for sudden expansion of =0.43 and at Re
L1
=1.8210
5
. The single-phase
result is also drawn on the same graph. The pressure is measured at the four peripheral taps on the
tube sections close to the singularity (points A, B, C and D) as well as their average (point M).
The two-phase experimental data are compared with prediction of the singular pressure change
for axisymmetric sudden expansion geometry obtained from the two following models:
Jannsen (1966) [1]:
P
tot
=
G
2
1
2
L
(1 )
2
_
1 +x
_
G
1
__
, (1)
where G
1
the mass ux upstream the singularity,
L
the density of water, the area ratio, x
the mass quality of air and
G
the density of air.
Chisholm (1969) [4]:
P
st
=
G
2
1
2
L
(1 ) (1 x)
2
_
1 +
C
X
+
1
X
2
_
, (2)
where
X
2
_
1 x
x
2
_

G
L
,
C =
_
1 + 0.5
_
L
_
0.5
__
_
G
_
0.5
+
_
L
_
0.5
_
.
Both models rely on the assumption of a homogeneous ow. Figure 7 shows that Jannsens
model [1] ts satisfactorily with the experimental results while Chisholms [4] model overesti-
mates the pressure change. This was also reported by Velasco [14].
18
Results paper
-20
-15
-10
-5
0
5
10
15
20
25
30
35
40
45
-10 -8 -6 -4 -2 0 2 4 6 8 10 12 14
z/d [-]
P

[
m
b
a
r
]
Single-phase-Experimental
L-M&Chisholm (1969)
L-M&Jannsen(1966)
Point M
Point A
Point B
Point C
Point D
Sudden enargement =0.43
Two-phase-20%air-Re
L1
=1.82E5
A B
C D
A B
C D
Figure 7: Two-phase static pressure change versus axial position for sudden enlargement of
=0.43 and for Re
L1
=1.8210
5
-comparison with experimental single-phase and with
models of Jannsen [1] and Chisholm [4].
To better emphasize the effect of two-phase ow we dene the dimensionless pressure change
L
as follows:
L
=
P
TP
Singular
P
SP
Singular
, (3)
where P
TP
Singular
is the singular two-phase pressure change as explained in Figure 6 and
P
SP
Singular
the single-phase one. Figure 8 displays the evolution of the experimental
L
versus
19
Results paper
0.90
0.95
1.00
1.05
1.10
1.15
1.20
1.25
1.30
1.35
1.40
0 5 10 15 20 25 30 35
Volumetric quality [%]
L

[
-
]
Experimental
Chisholm (1969)
Jannsen (1966)
Re
L1
=2.0E5
Figure 8: Dimensionless singular pressure change L versus volumetric quality. Comparison with
models of Jannsen [1] and Chisholm [4].
volumetric quality at Re
L1
=2.010
5
. The data are compared to the model of Jannsen [1] and
Chisholm [4], respectively. As it was previously mentioned, Jannsens [1] correlation agrees
better than Chisholms [4] correlation with the experimental results. The comparative graphs
given in Figures 9 and 10 indicate that the maximum deviation from the experimental data for
the model of Jannsen [1] is limited to 5% while it reaches 10% for Chisholm [4] model.
Measurements with the same ow conditions are repeated for a sudden enlargement of surface
area ratio =0.65. A summarizing graph of the static pressure recovery measured in both
geometries of =0.43 and 0.65 for different Re
L1
and for volumetric quality, , varying from
0 to 35% is presented in Figure 11. The singular pressure change is increasing for higher and
Re
L1
. Furthermore, for the same Re
L1
lower results in a lower P (up to three times smaller).
3.1.1.2 Progressive and sudden enlargement - Comparison Compared to sudden expansion,
a progressive enlargement will create for the same ow conditions, less pressure loss and
accordingly will exhibit a higher pressure recovery as depicted in Figures 12 and 13. Figure
12 shows a single-phase P diagram along sudden expansion and divergent of angles 5, 8 and
15
, of surface area ratio =0.43 and at Re

L1
=1.810
5
. In Figure 13, the same type of plot is
built for =20% of air. It can be seen that, for single-phase, the pressure drops 17% passing
from divergent section of 5
to 15
and 29% from 5
to sudden expansion. For two-phase

20
Results paper
5%
10%
40
45
50
55
60
65
70
75
80
40 45 50 55 60 65 70 75 80
P
singular
experimental [mbar]
P
s
i
n
g
u
l
a
r

J
a
n
n
s
e
n

[
m
b
a
r
]
Single-phase
Air 5%
Air 10%
Air 15%
Air 20%
Air 25%
Air 30%
Air 35%
Re
L1
=2.0E5
Figure 9: Deviation of Jannsen [1] model from
experimental results.
21
Results paper
5%
10%
40
45
50
55
60
65
70
75
80
40 45 50 55 60 65 70 75 80
P
singular
experimental [mbar]
P
s
i
n
g
u
l
a
r

C
h
i
s
h
o
l
m

[
m
b
a
r
]
Single-phase
Air 5%
Air 10%
Air 15%
Air 20%
Air 25%
Air 30%
Air 35%
Re
L1
=2.0E5
Figure 10: Deviation of Chisholm [4] model
from experimental results.
22
Results paper
0
10
20
30
40
50
60
70
80
90
80000 100000 120000 140000 160000 180000 200000 220000 240000
Re
L1
[-]
P
S
i
n
g
u
l
a
r

[
m
b
a
r
]
Single-phase
Air 5%
Air 10%
Air 15%
Air 20%
Air 25%
Air 30%
Air 35%
Figure 11: P
singular
for several Re
L1
from 0-35% of air for sudden enlargement of surface
areas =0.43 and =0.65.
ow, the pressure drop is 11% and 21% respectively. Additionally, we can notice that all the
curves in Figure 13 are shifted to the right, meaning that the ow becomes fully developed
further downstream the singularity and thus the recirculation zone is longer in two-phase ow.
In the case of sudden enlargement, contrary to smooth divergence, the pressure before starting
to increase slightly decreases at 1d and starts increasing again at 2d upstream of the singularity.
This is due to the presence of a secondary recirculation zone.
23
Results paper
-20
-15
-10
-5
0
5
10
15
20
25
30
35
40
45
50
-10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16
z/d [-]
P

[
m
b
a
r
]
Sudden enlargement
Divergent-angle 5
Divergent-angle 8
Divergent-angle 15
Singularity =0.43
Single-phase-ReL1=1.8E5
Figure 12: Pressure recovery diagram for single-phase ow and the same Re
L1
, a singularity of
=0.43 and for sudden enlargement and divergent of angles 5, 8 and 15
.
3.1.1.3 Proposed correlation for expansion singularities The proposed correlation relies
upon Jannsen [1] formulation. By tting this model to the experimental values, a corrective
coefcient is dened. It turns out that this parameter C is a function of the opening angle and
Re
L1
as shown by the 3D representation proposed in Figure 14. Although Jannsens [1] model
is chosen as the most accurate, attempts are made with Chisholm [4] model as well. Hence, the
corrective coefcient C for Chisholms [4] correlation is represented in a 3D plot in Figure 15.
In Table 3, the coefcients that are calculated for both models and for the different parameters
tested in progressive expansion are given.
24
Results paper
-20
-15
-10
-5
0
5
10
15
20
25
30
35
40
45
50
-10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16
z/d [-]
P

[
m
b
a
r
]
Sudden enlargement
Divergent-angle 5
Divergent-angle 8
Divergent-angle 15
Singularity =0.43
Two-phase 20 % air-ReL1=1.8E5
Figure 13: Pressure recovery diagram for two-phase ow (20% of air) and the same Re
L1
, a
singularity of =0.43 and for sudden enlargement and divergent of angles 5, 8 and
15
.
26
Results paper
Angle []
6
8
10
12
14
R
e
[
-
]
1.8x10
+05
2.0x10
+05
2.2x10
+05
C
[
-
]
0.3
0.4
0.5
0.6
0.7
0
.3
0
.3
5
0
.4
0
.4
5
0
.4
0
.4
5
0
.5
0
.5
5
0
.6
0
.6
5
X
Y
Z
C [-]
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
Jannsen (1966) model
Angle []
6
8
10
12
14
R
e
[-
]
1.8x10
+05
2.0x10
+05
2.2x10
+05
C
[
-
]
1.2
1.3
1.4
1
.3
2
1
.3
4
1
.3
6
1
.3
8
1
.4
1
.4
2
1
.1
8
1
.2
1
.2
2
1
.2
4
1
.2
6
1
.2
8
1
.3
1
.3
2
1
.3
4
1
.3
6
X
Y
Z
C [-]
1.42
1.4
1.38
1.36
1.34
1.32
1.3
1.28
1.26
1.24
1.22
1.2
1.18
Chisholm (1969) model
Figure 14: Coefcient C in function of and
Re
L1
for Jannsen [1] model.
26
Results paper
Angle []
6
8
10
12
14
R
e
[-
]
1.8x10
+05
2.0x10
+05
2.2x10
+05
C
[
-
]
0.3
0.4
0.5
0.6
0.7
0
.3
0
.3
5
0
.4
0
.4
5
0
.4
0
.4
5
0
.5
0
.5
5
0
.6
0
.6
5
X
Y
Z
C [-]
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
Jannsen (1966) model
Angle []
6
8
10
12
14
R
e
[-
]
1.8x10
+05
2.0x10
+05
2.2x10
+05
C
[
-
]
1.2
1.3
1.4
1
.3
2
1
.3
4
1
.3
6
1
.3
8
1
.4
1
.4
2
1
.18
1
.2
1
.2
2
1
.2
4
1
.2
6
1
.2
8
1
.3
1
.3
2
1
.3
4
1
.3
6
X
Y
Z
C [-]
1.42
1.4
1.38
1.36
1.34
1.32
1.3
1.28
1.26
1.24
1.22
1.2
1.18
Chisholm (1969) model
Figure 15: Coefcient C in function of and
Re
L1
for Chisholm [4] model.
The corrective coefcient C for Jannsen [1] formulation can be modeled as follows:
C = 0.061
0.8917
10717 Re
0.8283
L1
+ 0.378. (4)
This coefcient when applied to Jannsens [1] model gives a maximum deviation from the
model t of 58% for the case of =0.43, Re
L1
=1.8410
5
and 5
and minimum of 1.4% for

Re
L1
=2.3610
5
and 15
. It should be stressed that further experimental data are needed to rene

the C modeling and improve the validation.
Finally, the modied Jannsens [1] correlation can be written as:
P
tot
=
_
0.061
0.8917
10717 Re
0.8283
L1
+ 0.378
G
2
1
2
L
(1 )
2
_
1 +x
_
G
1
__
.
(5)
Table 3: Coefcient for adaption of Jannsen [1] and Chisholm [4] models to t to the experimental results
for divergence geometry and for several , and Re
L1
.
Jannsen [1] Re
L1
10
5
[
] C [-] Chisholm [4] Re

L1
10
5
[
] C [-]
=0.43
1.84 5 0.4
=0.43
1.84 5 1.34
2.3 5 0.26 2.3 5 1.445
1.78 8 0.48 1.78 8 1.3
2.36 8 0.38 2.36 8 1.365
1.76 15 0.7 1.76 15 1.155
2.36 15 0.69 2.36 15 1.163
=0.65
1.79 8 0.3
=0.65
1.79 8 1.187
2.26 8 0.24 2.26 8 1.365
3.1.2 Contraction singularity
3.1.2.1 Measurements in progressive contraction Convergence geometry of =1.56 and
angle 9
is studied. The geometry is identical to the test section shown in Figure 5 with a scaling
factor of 1/2 (DN40/32). The experimental facility and ow conditions are described in section
2.2. Pressure transducers of type Validyne are used for this experimental campaign with the
same acquisition time (t
acq.
=1 min) and sampling frequency (f
sampling
=2Hz). The different
membranes that cover all the range of the pressure measurements are:
Additionally, numerical simulations are carried out with the commercial CFDcode Fluent. The
test parameters and conditions are: 2D axisymmetric computation, realizable k turbulence
model with enhanced wall treatment and second order discretization scheme. Convergence
criterion is set at 10
7
. In Figure 16, the experimental and numerical static pressure drop is
plotted against axial position for several Re
L1
in single and two-phase ow. The pressure is
decreasing in a regular way before the singularity; the contraction creates a high pressure drop
step and then starts decreasing regularly downstream.
26
Results paper
Convergence
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60
L [m]
P

[
m
b
a
r
]
Single-phase-Exp-Re=136000
Single-phase-Exp-Re=79300
Single-phase-CFD-Re=739000
Two-phase-11% air-Exp-Re=95100
Two-phase-10%air-CFD-Re=66500
Smooth convergence =1.56,
angle 9
Figure 16: Experimental and numerical single and two-phase static pressure change versus axial
position for convergence of =1.56 and angle 9
for several Re
L1
.
The ow is observed fully developed close to the singularity (at 2d upstream and
downstream) contrary to the case of divergence for which the reattachment length is detected at
10d. Therefore, the singular pressure change P
singular
for convergence geometry is determined
by measuring the static pressure at equal distance upstream and downstream the singularity (2d).
Asummarizing graph of all experimental and numerical results obtained for single and two-phase
ow is shown in Figure 17. The results concerning the case of sudden contraction for several
and G (Guglielmini [15]) are compared to the experimental data. The experimental results for
smooth contraction are plotted in terms of the dimensionless pressure change
L
, dened by
eqn.3. In Figure 17 Jannsens [1] correlation for sudden contraction is adapted with a correction
coefcient of C=0.81 to t with the results (G=1990 kg/m
2
s).
28
Results paper
Convergence
G=1990 kg/m^2s
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
1.4
1.45
1.5
1.55
0 5 10 15 20 25 30 35
Volumetric quality [%]
L

[
-
]
Exp-G=1990 kg/m^2s
Exp-G=2786 kg/m^2s
Exp-G=1990-3424 kg/m^2s
CFD-G=1300-1700 kg/m^2s
Guglielmini et al.-G and varying
Janssen(1966) correlation-C=0.81
Smooth convergence =1.56,
angle 9
Figure 17: Experimental and numerical dimensionless singular pressure change
L
versus
volumetric quality. Comparison to literature (Guglielmini [15]) and to adapted
(C=0.81) Jannsen [1]model.
3.1.2.2 Proposed correlation for progressive contraction The correlation for sudden con-
vergence, as described from Jannsen [1], is recalled:
P
TP
=
G
2
2
2
L
_
_
1
C
C
1
_
2
+ 1
1
2
_
_
1 +x
_
L
G
__
. (6)
where C
c
is the contraction coefcient dened as C
c
= A
c
/A
1
where A
c
the ow area in the
vena contracta. A typical value of this parameter equal to 0.64 is considered for this investigation.
This correlation can be modied and then applied for the case of smooth contraction. The
parameter varying is the mass ux of water upstream of the singularity G
L1
. A t to the present
results is made and the resulting corrective coefcients are listed in Table 4.
A correlation to calculate the correction coefcient C is obtained as a function of G
L1
.
C = 2 10
8
G
2
L1
0.0001 G
L1
+ 0.9913. (7)
The relative discrepancy between experimental-numerical data and model t, when eqn.7 is
applied, varies from 5.72% to a maximum of 24.25%. The nal corrected correlation for the case
of smooth convergence of angle 9
is:
Table 4: Coefcient for adaption of Jannsens [1] formulation to t to the experimental and numerical results
for progressive contraction for several G.
Jannsen [1] correlation G
L1
[kg/m
2
s] Correction coefcient C [-]
Convergence-angle 9
, =1.56
1592 0.835
1990 0.800
2786 0.770
4378 0.754
P
TP
=
_
2 10
8
G
2
L1
0.0001 G
L1
+ 0.9913
G
2
2
2
L
_
_
1
C
C
1
_
2
+ 1
1
2
_
_
1 +x
_
L
G
__
.
(8)
3.2 Flow pattern maps and visualization
Flow regime maps are often considered in two-phase ow. A common chart is the one proposed
by Baker [16]. It has been established for horizontal ows in pipes of constant cross section. In
the present study, the ow is visualized both upstream and downstream the singularity. As it is
illustrated in Figure 18, four different ow patterns are identied downstream of the divergence;
Bubbly, Plug, Disperse and Annular ow.
30
Results paper
Flow regimes
Bubbly Plug
Annular
Disperse
Figure 18: Flow patterns identied downstream of the divergence geometry of =9
and =0.64.
For sudden and progressive enlargement (angles 5
and 8
) with =0.43 and =0.65, a normal

video camera is used to determine the condition for transition from bubbly ow to other types of
ow just after the singularity. The results are plotted on Bakers map and are reported in Figure
19. However, since the departure frombubbly owis decided on visual information, the transition
criterion remains rather subjective and the results given in Figure 19 are only indicative.
The second campaign of visualization is performed, using a high-speed camera, in a fully
transparent setup that allows better optical access (without pressure taps). Consequently, distinc-
tion between ow regimes is more straightforward. In this facility, a progressive enlargement of
31
Results paper
10 100 1000 10000 100000
G
L1
[kg.m
-2
s
-1
]
Sud.enl.-=0.43
Sud.enl.-=0.65
Div.angle 5 =0.43
Div.angle 8 =0.43
Div. angle 8 =0.65
Slug
Bubbly
Plug
Wavy
Annular
Stratified
Singularities =0.43
and =0.65
0.1
1
10
100
1
G
G
1

/

[
k
g
.
m
-
2
s
-
1
]
Figure 19: Modied Baker [16] map for pro-
gressive and sudden expansion of
=0.43 and 0.65.
32
Results paper
10 100 1000 10000 100000
G
L1
[kg.m
-2
s
-1
]
Bubbly
Disperse
Plug
Annular
Slug
Bubbly
Plug
Wavy
Annular
Stratified
Divergence =0.64
Angle 9
0.1
1
10
100
1
G
G
1

/

[
k
g
.
m
-
2
s
-
1
]
Figure 20: Modied Baker [16] map for pro-
gressive expansion of =0.64 and
=9
.
=0.64 for an opening angle of =9
is tested. The ow conditions for which these regimes

are visualized are reported in Figure 20. Finally, we should draw attention to the fact that all
ow conditions calculated refer to the upstream position. Indeed, for these test cases, the ow
regime upstream the singularity corresponds to bubbly ow (Baker map) while downstream three
additional ow patterns occur (plug, disperse and annular).
4 Conclusions
An investigation of horizontal air-water ow in sudden and progressive enlargements and
smooth contraction is performed. The static pressure evolution along these geometrical accidents
is measured and ow visualization is performed. The results are expressed in terms of the
dimensionless singular pressure change
L
. Compared to literature, a deviation of 5% is
found with Jannsens [1] model and 10% with Chisholms [4] model for axisymmetric sudden
expansion. For progressive enlargement of the same surface area ratio , the smallest the
opening angle, the highest the pressure recovery. For the same ow conditions, the minimum
pressure recovery occurs for sudden enlargement geometry. A modied version of Jannsens [1]
correlation is suggested for both progressive expansion and contraction. A corrective parameter
taking into account the different effects of the divergent angle and the liquid Reynolds number
of the divergent section and the upstream mass ux for convergence, is introduced. The
proposed correlation gives satisfactory results but needs further validation. In the convergence
conguration, the single and two-phase static pressure drop along the pipe is compared with
literature and CFD simulations; a satisfactory agreement is found. Finally, ow visualization
shows that departure from bubbly ow to plug, disperse or annular ow may occur in the
downstream section of a divergent singularity.
Acknowledgements
The support of the French company CETIM (Centre Technique des Industries M ecaniques) is
gratefully acknowledged. Mr. E.C. Bacharoudis and Miss R. Delgado-Tard aguila are thanked for
their contribution to the numerical and experimental study in this paper.
Nomenclature
A surface area [m
2
]
C correction coefcient [-]
d upstream diameter [m]
D downstream diameter[m]
f frequency [Hz]
G mass velocity [kg/m
2
s]
J supercial velocity [m/s]
L length of the pipe [m]
P pressure [Pa]
Q volumetric ow rate [l/s]

ReReynolds number [-]
t time [s]
x mass quality [-]
z axial position [m]
Greek symbols
opening angle [
]
volumetric quality [-]
kinematic viscosity[m
2
/s]
density [kg/m
3
]
surface area A
1
/A
2
[-]
dimensionless P [-]
Sub-Superscripts Abbreviations
c contraction SP single-phase
Ggaseous phase st static
L liquid phase tot total
1 upstream TP two-phase
2 downstream
s singularity
References
[1] Jannsen, E. & Kervinen, J.A., Two-phase pressure drop across contractions and expansions
of water-steam mixture at 600 to 1400 psia. Technical Report Geap 4622-US, 1966.
[2] McGee, J., Two-phase ow through abrupt expansion and contraction. Ph.D. thesis, North
Carolina State University, Raleigh, 1966.
[3] Chisholm, D., Prediction of pressure losses at changes of sections, bends and throttling
devices. Technical Report NEL rept. 388, 1968.
[4] Chisholm, D., Theoretical aspects of pressure changes at changes of section during steam-
water ow. Technical Report NEL rept. 418, 1969.
[5] Lottes, P., Expansion losses in two-phase ow. Nucl Sci Eng, 9, pp. 2631, 1960.
[6] Aloui, F. & Souhar, M., Experimental study of a two-phase bubbly ow in a at duct
symmetric sudden expansion. Part I: Visualization, pressure and void fraction. Int J
Multiphase Flow, 4, pp. 651665, 1996.
[7] Aloui, F., Doubliez, L., Legrand, J. & Souhar, M., Bubbly ow in an axisymmetric sudden
expansion: pressure drop, void fraction, wall shear stress, bubble velocities and sizes. Exp
Therm Fluid Sci, 18, pp. 118130, 1999.
[8] Schmidt, J. & Friedel, L., Two-phase pressure change across sudden expansions in duct
areas. Chem Eng Commun, 141, pp. 175190, 1996.
[9] Hwang, C.Y. & Pal, R., Flow of two-phase oil/water mixtures through sudden expansions
and contractions. Chem Eng J, 68, pp. 157163, 1997.
[10] Ahmed, W., Ching, C. & Shoukri, M., Pressure recovery of two-phase ow across sudden
expansions. Int J Multiphase Flow, 33, pp. 579594, 2008.
[11] Ahmed, W., Ching, C. & Shoukri, M., Development of two-phase ow downstream of a
horizontal sudden expansion. Int J Heat and Fluid Flow, 29, pp. 194206, 2008.
[12] Lockhart, R.W. & Martinelli, R.C., Proposed correlation of data for isothermal two-phase
two-component ow in pipes. Chem Eng Prog, 45, pp. 3948, 1949.
[13] Idelcik, I.E., Memento des pertes de charge. Editions Eyrolles: 61 Bd Saint-Germain Paris,
5th edition, 1986.
[14] Velasco, I., L ecoulement diphasique ` a travers un elargissement brusque, 1975.
[15] Guglielmini, G., Muzzio, A. & Sotgia, G., The structure of two-phase ow in ducts with
sudden contractions and its effects on the pressure drop. Experimental Heat Transfer, Fluid
Mechanics and Thermodynamics, 1997.
[16] Baker, O. Oil Gas J, 53, p. 185, 1954.

Multiphase Flow

Uploaded by

Copyright:

Available Formats

Multiphase Flow

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Multiphase Flow

Uploaded by

Copyright:

Available Formats

Two-phase ow modelling within expansion and

is exemplied. At each section

between them as shown

C will change and by 0.1 and 11

, of surface area ratio =0.43 and at Re

and 29% from 5

to sudden expansion. For two-phase

and minimum of 1.4% for

. It should be stressed that further experimental data are needed to rene

] C [-] Chisholm [4] Re

) with =0.43 and =0.65, a normal

is tested. The ow conditions for which these regimes

Q volumetric ow rate [l/s]

You might also like