Académique Documents
Professionnel Documents
Culture Documents
, V. Seshadri
Department of Applied Mechanics, IIT Delhi, Hauz Khas, New Delhi 110016, India
a r t i c l e i n f o
Article history:
Received 24 May 2007
Received in revised form
14 September 2007
Accepted 2 December 2008
Keywords:
Computational fluid dynamics
Cone flowmeter
Discharge coefficient
Reynolds number
Inlet swirl
Cone vertex angle
a b s t r a c t
Computational Fluid Dynamics (CFD) has emerged as a revolutionary tool for optimizing the design of
any flowmeter for given conditions. The flow features obtained with CFD are more extensive compared
to experiments. In the present study, CFD code FLUENT after validation has been used to investigate
the effect of cone vertex angle and upstream swirl on the performance of cone flowmeter. The values
of discharge coefficient (C
d
) evaluated for different vertex angles shows that the value of discharge
coefficient is independent of Reynolds number and its value decreases with increase in vertex angle.
In the presence of upstream disturbance in the form of swirl, the value of discharge coefficient is also
independent of Reynolds number and its value is only marginally affected by the magnitude of swirl. The
flow in a longitudinal plane shows the presence of a pair of contra-rotating vortices in the recirculation
region just downstream of the cone. The velocity profile downstream becomes stable after a distance of
about 5D.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
Flowmeters used in industries are often subject to highly
disturbed upstream flow conditions due to constraints on the
space available for laying the pipeline network. The disturbances
may be caused due to the presence of valve, elbow, pipe fitting,
bend etc at the upstream of the flowmeter. Conventional flow
measuring devices like Orifice meter, Rotameter, Flow nozzle etc.,
require minimum upstream and downstream straight lengths and
hence can not be used under these conditions. Over the years,
cone flowmeter has emerged as one of the best alternatives for
flow measurement under highly disturbed flow conditions [1]. It
is rather insensitive to outside vibration, cone configuration and
pressure tap location. It also provides flowmeasurement with high
accuracy over for a turn down ratio as high as 30:1 with a much
higher repeatability as compared to other flow measuring devices.
Besides these advantages, cone flowmeter has high durability
and high resistance to abrasion due to its tapered design. Its
taper design minimizes wear (erosion) by reduction in contact of
primary element with high velocity [2]. Liptak [3] has reported
that velocity profile due to presence of cone element tends to
flatten in the center resulting in uniform velocity profile in the
transverse plane. This could be the reason that cone flowmeter
does not require long straight pipe lengths upstream of the
m
Density of manometric fluid
f
Density of working fluid
p Static pressure
C
d
Discharge coefficient
Q Discharge rate (kg/s)
C
1
, C
2
, C
k,
Velocity perturbation
u
b
Bulk velocity
V Cell volume
V
f
Mass flux (velocity) through the face
X Longitudinal coordinate
Under relaxation factor
Turbulence dissipation rate
Dynamic viscosity
t
Eddy viscosity (turbulence viscosity)
ij
kroneker delta
around the body and recirculation zone above the float. Erdal and
Anderson [9] have shown that standard k model is inefficient for
flow prediction for orifice meters. They have also recommended
the use of more advanced turbulence models to improve the flow
prediction in these devices. Puri et al. [10] have evaluated the
drag force on the conical body of variable area orifice meter using
CFD and have also attempted to optimize the shape of the conical
body. The combination of frustum of cone and parabolic apex at
the top gives the streamlined shape of the conical body. A new
design for the variable area orifice meter has been proposed by
Singh et al. [11]. Drag force and pressure drop characteristics at
various sections were determined for the design of the flowmeter.
Seshadri et al. [12] have established the effect of body shape on
the annubar factor using CFD and have concluded that annubar
factor reduces with increase in blockage factor. Annubar factor
for elliptical shape with high slenderness ratio has the highest
annubar factor and minimum permanent pressure loss. Gandhi
et al. [13] have numerically investigated the flow behaviour of
Vortex shedding meter and have shown that triangular shape
having 60
1
_
1
4
4
_
D
2
d
2
_
_
2 P (1)
where, =
_
D
2
d
2
D
2
. (2)
3. Mathematical formulation
Basically, CFD is the numerical solution of governing equations
of motion, which describe the flow behavior of fluids. In the
modelling, mass, momentum, energy conservation equations (if
necessary) must be satisfied. The governing equations for steady
incompressible flows are:
x
i
(u
i
) = S
m
. (3)
The mass conservation equation is valid for both incompressible
and compressible flows. The source term S
m
is the mass added to
the continuous phase from the dispersed phase.
The momentum conservation equation for steady flow is
written as,
x
j
_
u
i
u
j
_
=
P
i
x
i
+
i,j
x
j
+g
i
+F
i
(4)
where, P
i
is the static pressure, g
i
is the gravitational body force
and F
i
is the external body force.
ij
is the stress tensor and is
computed as:
ij
=
_
u
i
x
j
+
u
j
x
i
_
2
3
ij
u
k
x
k
. (5)
3.1. Turbulence model
Most of the practical flows are turbulent and are characterized
by the fluctuation of velocity around its time average value, u =
u + u
x
j
_
u
i
u
j
_
=
P
i
x
i
+
i,j
x
j
+g
i
+F
i
+
x
j
_
u
i
u
j
_
. (6)
The additional terms in the above equation are the Reynolds
stresses and these need to be modeled for closure solutions. The
Boussinesq hypothesis [14] is used to relate the Reynolds stresses
to the mean velocity gradient as
u
i
u
j
=
t
_
u
i
x
j
+
u
j
x
i
_
2
3
_
k +
t
u
i
x
i
_
ij
. (7)
In the present investigation two-equation turbulence model (RNG
k [15]) has been used for closure solution of the set of equations.
The additional two equations in this model are
u
i
k
x
i
=
x
i
_
eff
k
x
i
_
+G
k
(8)
u
i
x
i
=
x
i
_
eff
x
i
_
+C
1
k
(G
k
) C
2
2
k
R (9)
R.K. Singh et al. / Flow Measurement and Instrumentation 20 (2009) 6974 71
Fig. 1. Design and drawing details of V-cone flowmeter used for validation [18].
where G
k
is the generation of turbulent kinetic energy due to the
mean velocity gradient and is calculated as
G
k
=
t
S
2
(10)
and S is the mean rate of shear stress tensor defined as
S =
_
2S
ij
S
ij
and S
ij
=
1
2
_
u
i
x
j
+
u
j
x
i
_
. (11)
The effective viscosity is modeled in the RNG theory using scale
elimination procedure resulting in a differential equation for
turbulent viscosity as
d
_
2
_
= 1.72
_
1 +c
d
where =
eff
and c
= 100. (12)
In the high Reynolds number limit, the above equation gives
t
=
C
k
2
with C
3
(1 /
0
)
1 +
3
2
k
(13)
where, = Sk/.
The values of constants in the turbulence model used are
the standard values reported in literature [C
1
= 1.42, C
2
=
1.68, C
= 0.0845,
k
=
= 0.7179,
0
= 4.38 and =
0.012].
4. Solution scheme
Flow Investigations have been carried out using CFD Code
FLUENT [16] which is based on cell centered finite volume
approach. Second order discretization scheme was used for all
governing equations since the grid which consists of tetrahedral
cells are usually not collinear to the flow direction. Under
relaxation factor has been used for all parameters to satisfy
Scarborough condition for convergence. Coupling between the
pressure and velocity field was established using PISO[17] scheme
which is based on the higher degree of the approximate relation
between the corrections for pressure and velocity as it is more
appropriate for swirling flows. All discretized equations have been
solved using segregated solver. An implicit solution scheme with
conjunction of Algebraic Multigrid (AMG) has been used for faster
convergence. Double precision was used in the computation and
solutions were convergeduntil the sumof the all the residual terms
was less than 10
6
.
5. Validation of the computer code
Any prediction made using CFD code is accepted only after
the validation of the code. Validation of the CFD code establishes
the extent of accuracy and reliability of the turbulence model. In
the present investigation, CFD code FLUENT has been validated
against the experimental data of Singh et al. [18]. The geometry
of the cone flowmeter used for validation of the CFD code
is given in Fig. 1. Flow prediction using different turbulence
models was carried out using different turbulence models with
water as working fluid. It was found that RNG k model gives
best matching with experimental results. Similar observation
has also been made by Erdal and Andersson [9] who have also
concluded that standard k model fails to describe flow features
of flowmeter. For sake of brevity, comparison of RNG k model
results with experimental data is only presented. Fig. 2a shows
the comparison of predicted C
d
with corresponding experimental
values for = 0.64. The deviation of between the experimental
values and the predicted values are of the same order as that of
the experimental uncertainties with the computed values. Further,
validation is carried out by predicting the flow for cone flowmeter
having = 0.77 and similar trends are observed (Fig. 2b). The
maximum difference between the experimental and predicted
results for both configurations is of the order of 4%, which is
within acceptable limits for validation. These deviations could be
attributed to the limitations of the turbulence model, uncertainty
in the experimental results, roughness of the pipe etc. Having
demonstrated the validation of the CFD code, it has been used for
parametric investigations.
72 R.K. Singh et al. / Flow Measurement and Instrumentation 20 (2009) 6974
Fig. 2a. Comparison of predicted C
d
with experimental values for a V-cone
flowmeter having = 0.64.
Fig. 2b. Comparison of predicted C
d
with experimental values for a V-cone
flowmeter having = 0.77.
6. Geometry details, boundary conditions and parameters
investigated
Flow predictions have been carried out for a cone flowmeter
consisting of a concentric cone placed inside a 52.10 mm diameter
pipe. The cone has maximum diameter of 40 mm and minimum
diameter of 18 mm for = 0.64. The tube housing the pressure
taps is sufficiently strong to support the cone centrally inside
the pipe and hence radial support struts have been deleted for
the parametric investigation. Details of the geometry of the cone
flowmeter analysed are given in Fig. 1. The upstream length of the
pipe was 5D whereas 20D downstream length was provided for
flow simulation. The longer downstream length has been chosen
in order to ensure that boundary condition at the exit has no effect
on the computational flow field close to the cone. Geometry of
the flow domain has been modelled using bottom up approach in
GAMBIT, the preprocessor of CFD code FLUENT [16]. 3-D model
of the flow domain which includes the cone is shown in the
Fig. 3. The process of grid generation is very crucial for accuracy,
stability and economy of the prediction. A fine grid leads to better
accuracy and hence it is necessary to generate reasonably fine grid
in the region of high velocity gradients whereas regions having
smooth flow condition can be meshed with coarser grids. In the
latest version of GAMBIT, tetrahedral and hexagonal core mesh
can be generated in the same flow domain, which is helpful in
the discretization of the complex flow domains with adequate
number of grids in computational domain. In the flow simulation,
the flowdomainwas meshedwithbothstructure andunstructured
mesh. For efficient discretization, geometry was divided into three
Fig. 3. 3-D model of central region of the flowmeter including cone.
Table 1
Geometry details for different cone vertex angles studied ( = 0.64).
Cone vertex angle (
, 80
and 100
over a range
of Reynolds numbers. To study the effect of upstream swirl on the
performance of cone flowmeter, flow simulations have also been
carried out for a 10
, 20
and 30
, 80
and 100
cone
vertex angle has a value of discharge coefficient of 0.78 whereas
for 100
angle whereas
60
cone angle has 0.06% standard deviation. The mean value and
standard variation of C
d
for 80
, 20
and 30
is 1.22%whereas for 20
and30
swirl,
the change is 2.56% and 2.90% respectively. This phenomenon can
be possibly attributed to the fact that the presence of swirl affects
both the pressure field and the structure of wake behind the cone.
Swirl also sets up a radial pressure gradient which increases the
upstreampressure thereby reducing the pressure difference across
the cone. This results in a slight increase in the value of discharge
coefficient.
8. Conclusions
Flow predictions and parametric investigations have been car-
ried out using CFDcode FLUENT to investigate the characteristics
of cone flowmeter. The major conclusions drawn from computa-
tional study are summarized as follows.
1. Comparison of CFD predictions with experimental results
shows that RNG k model is capable of predicting the
performance of cone flowmeter fairly accurately.
2. Velocity profiles downstream of the cone are free from wake
effect beyond a distance of 5D.
3. Discharge coefficient of cone flowmeter reduces marginally
with increase of cone vertex angle. Thus the cone flowmeter
having 60
swirl.
References
[1] Anon. Flowmeasurement solvedwithVenturi-cone meter. InTech1989;36(2):
302.
[2] Anon. Saudi plants install cone meters. Oil and Gas Journal 2000;98(30):73.
[3] Liptak BG. Flowmeasurement. Pennsylvania, America: ChiltonBook Company;
1993.
[4] Genisi, Donald, Chowdhury, Jayadev. Choosing the best flowmeter. Chemical
Engineering 1991;98(3):88100.
[5] Ifft, SA. Installation effects on the cone flowmeter in gas american society
of mechanical engineers. Fluids Engineering Division (Publication) FED. 211;
1997. p. 638.
[6] Peters B, Steven R, George D, Nored M. Tests on the V-cone flow meter
at southwest research institute and the utah state university in accordance
with the new API chapter 5.7 test protocol. In: North sea flow measurement
workshop. 2004.
[7] Prabhu SV, Mascomani R, Balakrishanan K, Konnur MS. Effect of upstreampipe
fittings on performance of orifice and conical Flowmeters. FlowMeasurement
Instrumentation 1996;7(1):4954.
[8] Buckle U, Drust F, Howe B, Milling A. Investigation of floating element
flowmeter. Flow Measurement and Instrumentation 1992;3(4):21525.
[9] Erdal A, Anderson HI. Numerical aspect of flow computation through Orifices.
Flow Measurement and Instrumentation 1997;8(1):2737.
[10] Puri J S, Singh S N, Seshadri V, Sanghi S. Performance of an adjustable orifice
meter with conical body. In: Proceeding of 12th ISME conference. 2001.
p. 4906.
[11] Singh SN, Gandhi BK, Seshadri V, Chauhan VS. Design of a bluff body
for development of variable area orifice meter. Flow Measurement and
Instrumentation 2004;15:97103.
[12] Seshadri V, Gandhi BK, Singh SN, Pandey RK. Analysis of the effect of bluff body
shape on annubar factor using CFD. Measurements 2004;35:2532.
[13] Gandhi BK, Singh SN, Seshadri V, Singh J. Effect of bluff body shape on Vortex
flowmeter performance. Indian Journal for Engineering and Material Sciences
2004;11:37884.
[14] Hinze JO. Turbulence. 6th edition. NewYork: McGraw-Hill Publications; 1975.
[15] Yakhot V, Orszag SS. Renormalization group analysis of turbulence: I Basic
theory. Journal of Scientific Computing 1996;1(1):151.
[16] Fluent 6.2 User Guide, Vol. 14, Fluent Inc., Lebanon, NH 03766. 2002.
[17] Issa RI. Solution of implicitly discretized fluid flow equations by operator
splitting. Journal of Computational Physics 1986;62:4065.
[18] Singh SN, Seshadri V, Singh RK, Gawhade RK. Effect of upstream flow
disturbances on the performance characteristics of cone flowmeter. Flow
Measurement and Instrumentation 2006;17:2917.
[19] Joshi SM. Eurekone flowmeter international conference on hydrocarbon flow
measurement. 2003. p. 427.
[20] Shapiro AH. Pressure field and Fluid accelerations. In: Illustrated experiments
in fluid mechanics. National Committee for Fluid Mechanics Films. Cambridge
(MA): MIT Press.