Flow Measurement and Instrumentation 28 (2012) 16

Contents lists available at SciVerse ScienceDirect

Flow Measurement and Instrumentation

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

Numerical investigation of wet gas ow in Venturi meter

Denghui He, Bofeng Bai n
State Key Laboratory of Multiphase Flow in Power Engineering, Xian Jiaotong University, Xian 710049, China

a r t i c l e i n f o

abstract

Available online 8 August 2012

Study of the Venturi meter over-reading in wet gas is of considerable importance for the wet gas
metering. Although the impacts of different parameters (e.g., liquid fraction, pressure and gas ow rate)
on the over-reading have been widely investigated, the underlying mechanism on how these
parameters act on the over-reading is still not fully understood. In this investigation, ve types of
turbulence models, including the standard k-e model, the RNG k-e model, the realizable k-e model, the
standard k-o model and the Reynolds stress model were examined. It was found that the standard k-e
model was in better agreement with the experimental data. From the simulations, how and why the
over-reading produced was explained. Then the liquid phase distributions and its impact on the
velocity eld and the pressure proles were discussed. The results indicated that the liquid
accumulated in the convergent section of the Venturi tube, where an annular liquid jet was formed.
The static pressure in the throat declined along the throat, which made the static pressure in the throat
unstable. To reduce their adverse effects on the over-reading of the wet gas ow, it was suggested that
the classical Venturi tube should extend the length of the throat and decrease the convergent angle.
This study gained a more comprehensive understanding of Venturi meter wet gas over-reading and
provided a reference for the design of a wet gas Venturi meter prototype.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Wet gas
Over-reading
Turbulence model
Discrete phase model
Liquid jet
Venturi meter

1. Introduction
Wet gas ow measurement is becoming increasingly important to
the production of natural gas [1,2]. The wet gas meter designs employ
mostly the Differential Pressure (DP) meter technology [38], especially the Venturi tube due to its proven advantages, including safety,
economy, convenience and clear physical interpretation. Although the
general understanding of the Venturi tube performance in wet gas is
widely accepted [3], little is known about the internal interactions of
the wet gas ow in a Venturi meter.
When the Venturi meter is used in the wet gas ow, the DP with
wet gas ow is usually larger than it would be if there was no liquid
present with the gas. This usually causes a positive error of the gas
ow rate prediction of the DP meter. Therefore, it is said that the
meter is over-reading (thereafter OR). The OR is the ratio of the
apparent gas mass ow, mg,Apparent , to the gas mass ow rate mg .
q
pb2 D2 C d e 2rg DPtp
q
mg,Apparent
1
4
4 1b
mg,Apparent
OR
mg

where, d and D are the throat diameter of the Venturi and the pipe
diameter, respectively, b the diameter ratio b d=D, C d the
n

Corresponding author. Tel.: 86 29 8266 5316; fax: 86 29 8266 5316.

E-mail address: bfbai@mail.xjtu.edu.cn (B. Bai).

0955-5986/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.owmeasinst.2012.07.008

discharge coefcient, e the expansibility factor, rg the gas density,

and Ptp is the wet gas differential pressure.
Many investigators [6,912] reported that the OR of the Venturi
meter is dependent on the LockhartMartinelli parameter (X LM ,
dened by Eq. (3)), the operating pressure P, the gas densiometric
Froude number (Fr g , dened by Eq. (4)) and the diameter ratio (b).
It is generally accepted that the OR increases with the increase of
X LM or Fr g keeping other parameters constant and decreases with
the increase of P, and OR also decreases as b increases. In addition,
several researchers [13,14] found that the Venturi OR was correlated with the liquid phase property and the pipe diameter.
s
ml rg
3
X LM
mg rl
U sg
Fr g p
gD

U sg

mg 4

rg pD2

rg
rl rg

where, ml is the liquid mass ow rate, rl is the liquid density, and

U sg is the supercial gas velocity.
In the last few years, Computational Fluid Dynamics (CFD) has
been applied increasingly in wet gas metering and various models
have been used. Reader-Harris et al. [15] examined wet gas ow
through Venturi tubes, in which the EulerEuler multiphase
model was employed. They noted that it was possible to model

D. He, B. Bai / Flow Measurement and Instrumentation 28 (2012) 16

10D
Flow
Venturi tube
3D

Fig. 2. Sketch of geometry used for computational domain.

mm.

wet gas ow through Venturi tubes and provided good tendency

with the experimental over-reading data. Xu et al. [16] presented a
simulation method based on the Discrete Phase Model (DPM) to
predict the OR characteristics of the Venturi tube. In their study, the
Renormalization Group (RNG) k-e model was used. They reported
that the maximal relative error of OR was 5.14%, and the average
relative error was less than 2.8%. The effects of installation, tapping
length and different types of gases on Venturi tubes and the
derivation of the discharge coefcient were investigated in [17].
It was found that the standard k-e model appeared in better
agreement with the test data than the Reynolds Stress Model
(RSM). Moreover, in several investigations [16,18,19], the RNG k-e
model and DPM were used to investigate wet gas ow in a V-cone
meter. Comparisons between the simulations and the experiments
suggested that the CFD model worked well on the OR prediction.
Although the single-phase uid ow in the Venturi tube is well
known to us, the wet gas ow is still under exploring. The
objective of this research is to simulate the ow eld characteristic of the wet gas ow in Venturi meter by proposing a
numerical model. Five turbulence models were compared. And
the SKE model was selected. Then the liquid phase concentration,
velocity and pressure distributions were investigated. On the
basis of the results, some advice on the improvement of the
classical Venturi tube were provided to get a more preferable
performance in wet gas ow.

2. Computational details

Table 1
Test envelope for wet gas Venturi at NEL [11].

Pressure (MPa, gauge)

Frg

X LM

0.75

1.6
3.1
6.1

1.5, 2.5, 3.5

1.5, 2.5, 3.5, 4.5
1.5, 2.5, 3.5, 4.5

00.30
00.30
0-0.30

10
Frg = 1.5
Frg = 2.5
Frg = 3.5
Frg = 4.5

Slug

Frl

Fig. 1. Geometric prole of classical Venturi tube with b 0:75

Annular-Mist
0.1

0.01
0.01

Stratified

0.1

10

Frg

2.1. Geometry and experiment

Fig. 3. Flow map showing conditions for wet-gas Venturi tests at NEL [11].

The experiments to validate the simulations are from a report

made by the NEL (National Engineering Laboratory) in the UK [11].
The Venturi tube is shown in Fig. 1 [20]. The diameter ratio (b) of
the Venturi tube is 0.75 and the pipe diameter D is 100 mm, the
convergent and divergent angles are 213 and 73 , respectively. The
high pressure tappings (upstream pressure tappings) are 50 mm
away from the entrance of the conical convergent, the low
pressure tappings (throat pressure tappings) are in the middle
of the cylindrical throat.
The geometry consists of three parts, i.e. the upstream pipe,
the Venturi tube and the downstream pipe, as the 3D view of
the modeled computational ow domain shown in Fig. 2. The
upstream straight length is three times the pipe diameter from
the entrance of the conical convergent and downstream straight
length is 10 diameters downstream from the end of the cone,
which enables the ow to fully develop and the pressure building
to nish. The ow domain was meshed with structured hexahedral meshes and the boundary layer meshing scheme was used
for grid generation in the region proximate to the wall. Moreover,
the mesh size of the throat was kept ne enough to achieve better
convergence and greater accuracy. The grid independency was
tested using computational grids among 300 000 and 1200 000
cells. The computational grid of approximately 660 000 cells was
selected here because of its economic computation and perfect
prediction.

The test envelope for wet gas in Venturi is shown in Table 1.

The simulation that was conducted agreed with the tests. Fig. 3 is
the wet gas ow distribution in the ow pattern map. The wet gas
ows mainly lies in the Annular-Mist ow pattern.

2.2. Mathematical model

2.2.1. Turbulence model
The commercial CFD software, FLUENT 6.3, was used here. The
continuum gas phase (nitrogen) was predicted under steady-state
conditions. Five types of turbulence models [20], i.e., the SKE, the
RNG, the Realizable k-e model (hereafter RKE), the Standard k-o
model (hereafter KWM) and the RSM were compared in this
study. The SKE, RNG, and RKE models have similar forms, with
transport equations for k and e. The major differences in the three
models are as follows: the method of calculating turbulent
viscosity, the turbulent Prandtl numbers governing the turbulent
diffusion of k and e and the generation and destruction of the
turbulence in the e equation.
The KWM model contains the modications for low-Reynoldsnumber effects, compressibility, and shear ow spreading. This
model is in close agreement with measurements for far wakes,

D. He, B. Bai / Flow Measurement and Instrumentation 28 (2012) 16

mixing layers, and plane, round, and radial jets, and is thus
applicable to wall-bounded ows and free shear ows [21].
The RSM [2224] accounts for the effects of streamline
curvature, swirl, rotation, and rapid changes in strain rate in a
rigorous manner, and it has great potential to give accurate
predictions for complex ows, such as cyclone ows, highly
swirling ows in combustors, rotating ow passages, and the
stress-induced secondary ows in ducts [21].

2.2.2. Multiphase model

In the Annular-Mist ow, the liquid consists of two types: the
droplet and very thin liquid lm. In the present simulation, the
maximum volume fraction of the liquid is 8.23%. Thus the DPM
model is fully capable of simulating the conditions (the DPM
model usually used for the liquid fractions is less than 10% [21]).
In addition, to simulate the thin liquid lm, the wall-lm model
was also used as the boundary condition of the wall. The
equations of the motion for droplets can be written as
g x rl rg
dul
F D ug ul
Fs
dt
rl

Furthermore, the stochastic tracking (random walk) model

was used to predict the dispersion of droplets due to turbulence
in the gas phase.
2.3. Numerical procedure
The governing transport equations were discretized with a
nite-volume approach. The second-order upwind discretization
scheme was used for the pressure equation and the third-order
QUICK scheme was adopted for other terms.
The mass ow inlet boundary condition was used to dene the
gas ow rate at the ow inlet and pressure outlet boundary
condition was adopted at the end of the pipeline. The turbulence
intensity at the inlet and outlet was dependent on the empirical
correlation for fully-developed duct ows. The boundary condition of the wall employed the wall-lm model. The temperature
was set as 291.15 K.
The liquid (liquid kerosene) was injected from the surface at
the inlet. The distribution of the droplet sizes employed the
RosinRammler type, the mass fraction Y d of the droplets
diameter greater than d0 was given by
Y d expd0 =dn

18mg C D Re
FD
r d2 24

l l

where ug , ul are the gas and liquid velocity, respectively, rg the

gas density, rl the liquid density, g x the gravitational acceleration,
F D ug ul the drag force per unit droplet mass, and F D is
determined by Eq. (7), F s is the Saffman lift force due to the shear
between phases, mg is the gas molecular viscosity, dl the liquid
droplet diameter, and Re is the relative Reynolds number dened
as Eq. (8), C D is the drag coefcient [25].
Re

rg dl 9ul ug 9
mg

10

where d is the droplet mean diameter, n is the spread parameter.

Table 2 shows the distribution of the liquid droplet diameter in
present simulation. The minimum and maximum diameters of
the droplet are 0.05 mm and 0.5 mm, respectively, and the mean
diameter is 0.248 mm [26].
To increase the calculation efciency, the continuous gas
phase ow eld was obtained rstly and the liquid phase
simulation was then carried out based on the converged solution
of the gas ow. The convergence criteria were assumed to be met
when the iteration residuals were reduced by 106 .

8
3. Comparison of turbulence models

a2
a3

C D a1
Re Re2

where a1 , a2 and a3 are empirical constants for smooth spherical

droplets over several ranges of droplet Reynolds number.
Table 2
Size distribution of liquid droplet diameter.
Diameter range (mm)

Mass fraction in range

Diameter, d0 (mm)

Yd

050
50100
100150
150200
200250
250300
300350
350400
400500

0.05
0.05
0.15
0.20
0.20
0.15
0.10
0.05
0.05

50
100
150
200
250
300
350
400
500

0.95
0.90
0.75
0.55
0.35
0.20
0.10
0.05
0

According to ISO 5167-4:2003 [20] the discharge coefcient C d

of the classical Venturi tube with a machined convergent section is
approximately 1.000 when the throat Reynolds number, Red , lies
between 106 and 2  106 and C d 1:010 with Red ranging from
2  106 to 108 . Table 3 shows the discharge coefcient with different
turbulence models compared with the values recommended by ISO
5167-4:2003. All turbulence models can predict the single phase gas
ow accurately. There are little differences in relative deviations of all
ve models and the maximum deviation is less than 2.6%.
As can be seen from Fig. 4, the ve models under predict the
OR for low X LM o 0:05 and over predict for high X LM at
Frg 1:5; whereas the simulations of the ve models under
predict the OR at Frg 3:5. The SKE model gives slightly higher
OR than the other four models. Fig. 5 shows the relative deviations of OR at Frg 1:5 and Frg 3:5 for different turbulence
models. At Frg 1:5, the KWM model gives the smallest relative
deviation of OR, the deviation of the SKE model is the largest, but

Table 3
Comparisons of discharge coefcient using different turbulence models.
Turbulence model

Calculated discharge coefcient C cd

Frg 1:5

SKE
RNG
RKE
KWM
RSM

0.9859
0.9877
0.9881
0.9870
0.9882

Red 1845088:3

Frg 3:5
0.9841
0.9850
0.9851
0.9851
0.9868

Red 4305206:2

ISO discharge coefcient C d

Relative deviation (%)

Red 106 2  106

Red 2  106 108

Frg 1:5

Frg 3:5

1.010

1.41
1.23
1.19
1.30
1.18

2.56
2.48
2.47
2.47
2.30

1.000

D. He, B. Bai / Flow Measurement and Instrumentation 28 (2012) 16

1.8
1.7
1.6

OR

1.5

Frg = 1.5
SKE
RNG
RKE
KWM
RSM
Test

it is still no more than 6.0%. When Frg 3:5, the SKE model
predicts the wet gas best among the ve models, while the
deviation of the KWM model is up to 8.5%. Hence, compared with
the other four models, the SKE model can predict the wet gas
better. And the relative deviation of the OR is with 76.0%.
From the above discussions, the SKE model is better than other
four models in the wet gas simulation. In addition, the convergence
of the other four models is more difcult than the SKE model,
especially the RSM model. Hence the SKE model is nally selected.

Frg = 3.5
SKE
RNG
RKE
KWM
RSM
Test

1.4
1.3

4. Results and discussion

1.2

4.1. Comparisons between simulations and test results

1.1
1.0
0.00

0.05

0.10

0.15
XLM

0.20

0.25

0.30

Fig. 4. Over-reading at Frg 1:5 and Frg 3:5 for different turbulence models
under 1.6 MPa gauge.

15
10

Frg = 3.5
RNG
SKE
RLK
KWM
RSM

1.7

+6.0%

1.6
1.5

0
1.4
OR

Deviation (%)

Frg = 1.5
RNG
SKE
RLK
KWM
RSM

The comparisons between the simulations and the test results

[11] under different pressure are shown in Fig. 6. The simulations
agree well with the experiments. The OR is closely related to the
X LM and increases with it for other parameters held constant.
Moreover, an increase in the pressure leads to a reduction in OR.
As shown in Fig. 7, the OR does not vary signicantly with the Frg .
On the one hand, the effect of the Frg on the OR is closely related

-6.0%

-5

1.3

-10
-15
0.00

3.1MPa
Experiment
1.5
2.5
3.5
4.5
Simulation
1.5
2.5
3.5
4.5

1.2
1.1

0.05

0.10

0.15

0.20

0.25

0.30

0.35

XLM

1.0
0.00

Fig. 5. Relative deviations of OR at Frg 1:5 and Frg 3:5 for different turbulence
models under 1.6 MPa gauge.

0.05

0.10

0.15
XLM

0.20

0.25

0.30

Fig. 7. Comparison between simulations and experimental results for different Frg
at 3.1 MPa gauge.

1.7
Frg = 2.5

1.5

OR

1.4
1.3

Experiment
1.6MPa
3.1MPa
6.1MPa
Simulation
1.6MPa
3.1MPa
6.1MPa

10
8
6
Relative Deviation (%)

1.6

1.2
1.1

1.6MPa
3.1MPa
6.1MPa

+5.5%

4
2
0
-2
-4
-6

-5.5%

-8

1.0
0.00

0.05

0.10

0.15
XLM

0.20

0.25

0.30

Fig. 6. Comparison between simulations and experimental results for different

pressures at Frg 2:5.

-10
0.00

0.05

0.10

0.15

0.20

0.25

XLM
Fig. 8. Relative deviation of predicted OR.

0.30

0.35

D. He, B. Bai / Flow Measurement and Instrumentation 28 (2012) 16

with the wet gas ow pattern [10]. In our investigation, the ow

pattern appeared as Annular-Mist ow in which most of the
liquid moved close to the gas velocity as small droplets. Under
such conditions, increasing gas velocity has little effect on the OR.
On the other hand, the droplet size distribution shown in Table 2
is the same for different Frg . However, the predicted OR varies
with droplet size and are close to each other for different Frg
under the equal droplet size [15].
The Relative deviations of predicted OR compared with the
experimental data [11] are displayed in Fig. 8. The relative deviation
of the OR is within 7 5:5% at the 95% condence level. The maximal
relative deviation of OR is 6.38%, and the average relative deviation
is less than 2.84%. The comparisons with the experiments show that
the model and its solution approach are reasonable.

Fig. 9. Contours of liquid phase concentration in Venturi meter (P 3:1 MPa,

Frg 2:5, X LM 0:1).

4.2. Liquid phase concentration distributions

The concentration of liquid is quite intense in the convergent
and throat sections of the Venturi tube, for the droplets impacting
on the wall of the convergent section and forming a liquid layer as
shown in Fig. 9. This liquid layer then separates at the end of the
convergent and forms an annular jet entering the throat, after
which it continues to pass through the divergent without reattaching to the wall. A similar phenomenon was also reported by
Reader-Harris et al. [15].
When the liquid fraction is low (e.g. X LM 0:01), the liquid jet is
not noticeable and most of the liquid is carried by the gas in droplet
form and disperses into the gas more homogeneously, as shown in
Fig. 10(a). The friction pressure drop and the acceleration pressure
drop are almost the same as that in the dry gas ow, so the
presence of the liquid produces very low OR (Fig. 6). The liquid jet
becomes obvious and lasts a much longer distance as the liquid
fraction increases and the throat area occupied by the liquid phase
increases. Thus, as shown in Fig. 11, the effective gas ow passage
in the core of the throat is decreased when the X LM is from 0.01 to
0.3, which leads to the increase of the acceleration pressure drop.
The OR increases correspondingly as shown in Fig. 6.
Under the conditions of xed pressure and X LM , the gas ow
rate has little effect on the distribution of the liquid phase
concentration as shown in Fig. 10(b). The liquid jet is not affected
by the increasing Frg . Under the conditions of xed Frg and X LM ,
it is shown from Fig. 10(c) that the liquid annular jet is more
obvious under lower pressure than that under higher pressure,
which leads to greater OR as shown in Fig. 6.

4.3. Wall pressure prole distributions

When the X LM is low (e.g. X LM r0:05), the liquid has little effect
on the wall pressure prole as shown in Fig. 12. Like the dry gas, the

(v/vmax)

XLM = 0
Throat

XLM = 0.01

XLM = 0.05

XLM = 0.10

XLM = 0.20

XLM = 0.30
Fig. 10. Contours of liquid phase concentration in Venturi meter (a) P 3:1 MPa,
Frg 2:5, X LM 0:0120:3, (b) P 3:1 MPa, X LM 0:1, Frg 1:54.5, (c) Frg 2:5,
X LM 0:1, P 1:626:1 MPa.

Fig. 11. Contours of normalized gas velocity v=vmax through the Venturi tube
under 3.1 MPa gauge (Frg 2:5, X LM 00:3), vmax is the maximum velocity in the
Venturi tube.

D. He, B. Bai / Flow Measurement and Instrumentation 28 (2012) 16

The decline of the pressure is bad for the pressure measurement in the throat of the Venturi.

3106
XLM
3104
Divergent

3102
Static Pressure (kPa)

0
0.01
0.05

0.10
0.20
0.30

Acknowledgments

3100

This work was nancially supported by the National Nature

Science Foundation of China for Creative Research Groups under
Contract No. 51121092.

3098
Throat

3096

References

3094
3092

Convergent

3090
-200

200

400
600
X axis (mm)

800

1000

1200

Fig. 12. Wall static pressure proles under 3.1 MPa gauge (Frg 2:5, X LM 00:3).

pressure prole in the throat is at. While the throat static pressure
declined obviously along the throat at the great X LM . This is closely
related with the reduced effective gas ow passage along the throat
as shown in Fig. 11. The greater X LM is, the faster the throat pressure
declines. In fact, the pressure decline along the throat means that
the ow in the throat is not fully developed, and it is not favorable
to measure the pressure of the throat. Moreover, the inuence
distance of the liquid jet increases with the X LM increasing as shown
in Fig. 10. Hence the pressure recovery length increases with the
X LM as the other parameters keep constant. Longer pressure
recovery length is required in wet gas ow than that of in single
phase uid recommended by ISO 5167-4:2003 [20].
According to the above analysis, the liquid jet has great impact
on the wet gas measurement. To reduce the inuence of the liquid
jet and get more preferable wet gas metering performance with
the Venturi meter, several measures were proposed. First, increasing the length of the throat. Extending the Venturi throat can
reduce the inuence of the liquid jet and more stable throat
pressure is obtained. The throat-extended Venturi meters have
been adopted in multiphase ow measurement [2730]. Second,
reducing the convergent angle. The small convergent angle could
reduce the production of the liquid jet. Finally, extending the
outlet pipe length for the installation of Venturi meter. The pipe
length required for the installation of the classical Venturi tube in
the single phase uid ow is inapplicable in the wet gas ow.

5. Conclusions
In this investigation, the wet gas ow through a Venturi meter
was examined with the Discrete Phase Model. The standard k-e
model agreed with the experimental data better and was employed.
On the basis of the simulations, the liquid phase distributions and
the pressure proles and their impact on the over-reading of the
wet gas Venturi meter were discussed. The strategies to reduce
their adverse effect on the measurement were suggested. The main
ndings may be summarized as follows:
(1) The liquid accumulated in the convergent section of the
Venturi tube and formed an annular liquid jet. The liquid jet
was much more obvious under both greater liquid fraction
and lower pressure, which led to greater over-reading.
(2) The static pressure in wet gas ow is unstable and declines
much more along the throat than that in dry gas ow.
The greater the X LM is, the faster the throat pressure declines.

[1] Steven RN. Wet gas metering with a horizontally mounted Venturi meter.
Flow Measurement and Instrumentation 2002;12:361372.
[2] Falcone G, Hewitt GF, Alimonti C. Multiphase ow metering. 1st ed. UK:
Elsevier B.V.; 2009 pp. 22938.
[3] ASME (The American Society of Mechanical Engineers). Wet gas owmetering guideline. ASME MFC-19G-2008, Technical Report, 11 July, 2008.
[4] Lawrence PA. Wet gas measurement. In: 84th International School of
Hydrocarbon Measurement. 2009.
[5] Werven van M, Drenthen J, Boer de G, Kurth M. Wet gas ow measurement with
ultrosonic and differential pressure metering technology. In: 6th ISFFM. 2006.
[6] Steven RN. Wet gas ow metering with gas meter technologies. In: CIATEQ
2006. 2006.
[7] Evans R, Ifft SA. Wet gas performance of differential pressure owmeters. In:
25th North Sea Flow Measurement Workshop 2007: Gardermoen, Norway, 2007.
[8] Evans R, Ifft SA. Wet gas performance of differential pressure owmeters. In:
ASME Conference Proceedings. 2007. pp. 195200.
[9] Steven RN. A dimensional analysis of two phase ow through a horizontally
installed Venturi ow meter. Flow Measurement and Instrumentation
2008;19:342349.
[10] Stewart DG, Brown G, Hodges D, Kilbride E. Wet gas Venturi metering. In: 2002
SPE Annual Technical Conference and Exhibition. San Antonio, Texas 2002.
[11] Stewart DG. The evaluation of dry gas meters in wet gas conditions. National
Engineering Laboratory: East Kilbride, Glasgow. 2002; Report No: 2002/100.
[12] Stewart DG. Application of DP meters to wet gas ow. In: 2nd South East Asia
Hydrocarbon Flow Measurement Workshop. 2003.
[13] Reader-Harris MJ, Hodges D, Gibson J. Venturi-Tube performance in wet gas
using different test uids. TUL NEL. 2005; Report No: 2005/206.
[14] Steven RN, Kinney J, Britton C. Liquid property and diameter effects on
Venturi meters used with wet gas ows. In: 6th International Symposium on
Fluid Flow Measurement. 2006.
[15] Reader-Harris MJ, Hodges D, Gibson J. Venturi tube performance in wet gas:
computation and experiment. In: 6th South East Asia Hydrocarbon Flow
Measurement Workshop. 2007.
[16] Xu Y, Duan Y, Zhao Y. Numerical simulation on high pressure wet gas
owmetering overreading characteristics of Verturi tube. Journal of Tianjin
University 2012;45(3):221227.
[17] Reader-Harris MJ, Gibson J, Rushworth R, Hodges D. Effects of installation,
tapping length and different gases on Venturi tubes of convergent angle 10.5
deg and the derivation of a discharge coefcient equation. NEL report. 2005;
Report No: 2005/225.
[18] Wu J. Numerical research of wafer cone meter on its expansibility factor and
characteristics of wet gas metering. M.S. Thesis. Tianjin University; 2007.
[19] Duan Y. Research of double-throttle device on the high pressure wet fas
metering model CFD based. M.S. Thesis. Tianjin University; 2009.
[20] ISO. Measurement of uid ow by means of pressure differential devices
inserted in circular cross-section conduits running full, Part 4: Venturi Tubes.
BS EN ISO 5167-4 2003. 2003.
[21] Fluent Inc. FLUENT6.3 Users Guide. Lebanon, NH 03766: Fluent Inc; 2006.
[22] Larsen M, Howell J. Least-squares smoothing of direct-exchange areas in
zonal analysis. Sandia National Labs., Albuquerque, NM (USA); Texas Univ.,
Austin (USA). Dept. of Mechanical Engineering; 1984.
[23] Launder BE. Second-moment closure: present y and future? International
Journal of Heat and Fluid Flow 1989;10:282300.
[24] Launder BE, Reece GJ, Rodi W. Progress in the development of a Reynoldsstress turbulence closure. Journal of Fluid Mechanics 1975;68:537566.
[25] Morsi SA, Alexander AJ. An investigation of particle trajectories in two-phase
ow systems. Journal of Fluid Mechanics 1972;55:193208.
[26] Crowe CT. Multiphase Flow Handbook, vol. 8. USA: CRC Press; 2005 p. 42.
[27] Xu L, Zhou W, Li X, Tang S. Wet gas metering using a revised Venturi meter
and soft-computing approximation techniques. IEEE Transactions on Instrumentation and Measurement 2011;60:947956.
[28] Fincke JR, Ronnenkamp C, Kruse D, Krogue J, Householder D. Performance
characteristics of an extended throat ow nozzle for the measurement of
high void fraction multi-phase ows. 4th International Symposium of Fluid
Flow Measurement. Denver, USA 1999.
[29] Wu Z, Xie F. Optimization of Venturi tube design for pipeline pulverized coal
ow measurements. Frontiers of Energy and Power Engineering in China
2008;2:369373.
[30] Xu L, Zhou W, Li X. Wet gas ow modeling for a vertically mounted Venturi
meter. Measurement Science and Technology 2012;23(4):045301.