NUMERICAL ANALYSIS OF TURBULENT FLOW

CHARACTERISTICS OF AN INCOMPRESSIBLE FLUID IN A

VENTURI-TYPE DEVICE
Natha M. Caetano and Luiz E. M. Lima
Mechanical Academic Department, Federal Technological University of Paran, 84016-210, Ponta
Grossa, PR, Brazil, lelima@utfpr.edu.br, http:// www.utfpr.edu.br/ pontagrossa
Keywords: Venturi, Numerical analysis, Turbulence.
Abstract. The Venturi is a device widely used in various industrial applications: automotive, food,
agricultural, oil, among others. The Venturi has circular cross section or polygonal, depending on your
application and it consists of three sequential parts: the convergent, the throat and the divergent. The
industrial applications of the Venturi can vary fromminor adjustments in the mechanismto more complex
engineering adaptations. The development of a computational model for ow in a Venturi constitutes an
additional tool for the analysis and determination of ow variables involved in the equipments which
apply the Venturi principle. This paper presents a numerical study of the characteristics of turbulent ow
of incompressible uids in Venturi-type devices, through the use of different turbulence models. The
Reynolds decomposition process usually applied to incompressible ow was used as a mathematical tool
in the formulation of the turbulent ow. This decomposition provides the Reynolds Averaged Navier-
Stokes (RANS) equations. For the closing of the mean equations it is necessary to model the unknown
terms (Reynolds stresses) that arise from nonlinearities in the instantaneous mean ow equations. In
the present study, the Reynolds stresses were modeled using different versions of the k-epsilon model,
among others. For this study, Ansys/Fluent commercial software for Computational Fluid Dynamics
(CFD), which uses a nite volume method in the discretization of the governing equations, will be used.
To validate this computational model, their results will be compared against pressure experimental data.
The study of the ow eld, the pressure distribution and the turbulent stresses is of great importance in
understanding the ows in these equipments, as well as in the improvement of some of its applications
and developing new applications of the Venturi principle.
1 INTRODUCTION
The Venturi is a device that was rst used to solve simple problems and improve the hy-
draulics knowledge, years after its development it started to be used in some industrial appli-
cations, among them the ow measurement. Currently the Venturi principle is used for various
technological purposes in different applications, we highlight the air puriers, gas-solid injec-
tors and jet pumps. The industrial applications of the Venturi vary from minor adjustments in
the mechanism to more complex engineering adaptations.
The Venturi has circular cross section or polygonal, depending on your application and it
consists of three sequential parts: the convergent, the throat and the divergent. These geometric
characteristics allows the pressure recovery occurs from the inlet to the outlet of the Venturi.
Due to the gradual reduction of area are not present certain phenomena, such as the vena con-
cracta (Fox et al., 2010), thus the ow is regular throughout the course of the Venturi when the
ow rate are relatively low. For higher ow rates, the separation points can occur along the
Venturi, which causes the occurrence of eddies, or counter ow, so that the pressure does not
fully recover (McDonald and Fox, 1966).
Equipments based on the Venturi principle are capable of performing the transformation of
pressure energy into velocity energy. This application is used extensively by industry to improve
air quality and also dene its moisture, receiving the name of Venturi aeration (Baylar et al.,
2009). The air puriers are generally used in the food industry, with the aim of removing
impurities and regulating the temperature, by a uid that is injected in the throat region and
subsequently dispersed in the form of droplets in the environment in which it is desired to
purify (Puentes et al., 2012). For gas-solid injectors the same principle of the air puriers is
used. The air circulating in the Venturi takes solid particles that are inserted in the throat region
carrying them to the desired locale (Domingues, 2006).
In the lift jet pumps technology the Venturi design requires a higher level of engineering. The
Venturi principle associated with this equipment is used in lift and ow of submerse uids such
as water or oil. The operation of a lift jet pump is relatively simple. A motor uid is injected
at the entrance of the Venturi-type device, which causes the uid surrounding it is sucked into a
chamber, where the mixing occurs. The two uids are mixed and then transported to the surface
through the duct using the thrust resulting from the motor uid (Oliveira et al., 1996).
Ghassemi and Fasih (2011) study cavitating Venturis with three different throat diameters to
investigate the effect of Venturi size on its mass ow rate. They conducted three different sets of
experiments to investigate the performance of the Venturis. In the experiments, the mass ow
rates were examined under different downstream and upstream pressure conditions and time
varying downstream pressure. It is found that the Venturi size has no effect on its expecting
function to keep mass ow rate constant. Also, it is shown that by applying a discharge coef-
cient and using only upstream pressure, the cavitating Venturi can be used as a ow-meter with
a high degree of accuracy in a wide range of mass ow rate.
Blocken et al. (2011) used Computational Fluid Dynamics (CFD) to gain insight in the aero-
dynamic performance of a Venturi-shaped roof (called VENTEC roof). The simulations are
performed with the 3D steady Reynolds-averaged Navier-Stokes (RANS) equations and the
Renormalization Group (RNG) k model. A detailed analysis is conducted of the inuence
of the so-called Venturi-effect and the wind-blocking effect on the aerodynamic performance
of the VENTEC roof. The results indicate that because of the wind-blocking effect, the highest
contraction ratio does not provide the best aerodynamic performance and the largest negative
pressure, which is a counter-intuitive result.
This works presents a numerical study of the characteristics of turbulent ow of incompress-
ible uids in Venturi-type devices, through the use of different turbulence models. The aim of
this study is to obtain a computational model of the ow in a Venturi that allows the analysis of
this device in several applications. For this study, Ansys/Fluent commercial software for Com-
putational Fluid Dynamics (CFD), which uses a nite volume method in the discretization of
the governing equations, will be used. To validate this computational model, their results will
be compared against pressure experimental data.
2 GOVERNING EQUATIONS AND NUMERICAL MODEL
The present numerical study of turbulent ows of incompressible uids in a Venturi-type
device was done using the commercial software for Computational Fluid Dynamics (CFD) An-
sys/Fluent version 14.5. The numerical model is based on the solution of the Reynolds averaged
Navier-Stokes (RANS) equations for incompressible ows using a nite volume method on
non-structured non-uniform grid. The turbulent stress terms were closed with a two equations
k turbulence model, that will be presented below.
2.1 Reynolds Averaged Navier-Stokes (RANS) equations
The RANS equations are time averaged Navier-Stokes equations written for the instanta-
neous velocity decomposed in a mean value and a uctuation. For an incompressible Newtonian
uid, these equations can be written as:
u
i
x
i
= 0, (1)
u
i
t
+

x
j
(u
j
u
i
) = f
i
+

x
j

ij
+

u
i
x
j
+
u
j
x
i

i
u

, (2)
where u is the velocity, x is the space coordinate, t is the time, f
i
is the volume force, p is the
pressure, is the specic mass,
ij
is the delta of Kronecker and is the kinematic viscosity. The
left side of Eq. (2) represents the time variation of momentum due to the transient regime and
the convection in the mean ow. The transient term is kept for numerical integration purposes.
This variation is balanced by the mean volume force, the average pressure gradient, the viscous
stresses and the apparent viscous stresses (u

i
u

j
) due to uctuations in the velocities eld,
generally known as the Reynolds stresses.
2.2 Turbulence models
The turbulence models used for the simulation of the ow in Venturi in this work are: the
standard k , the RNG k and the realizable k .
In the standard k model the turbulent kinetic energy, k, is derived from the exact equa-
tion, while the dissipation rate of the turbulent kinetic energy, , is obtained by reasoning about
physical phenomena. The RNG k model is derived using statistical techniques called renor-
malization group theory and has an extra term in the dissipation rate that optimizes the accuracy
of the results for uids subjected to tension quickly. The realizable k model differs from
the standard model to contain an alternative formulation for the turbulence model and present
an motion equation for the dissipation rate modied, derived from the exact equation for the
motion to the main area of vorticity change. The transport equations for standard k model
are presented below (Launder and Spalding, 1974).
The turbulence kinetic energy, k, and its dissipation rate, , are obtained from the following
transport equations:

t
(k) +

x
i
(ku
i
) =

x
j

+

t

k
x
j

+ P
k
+ P
b
Y
M
+ S
k
, (3)
and

t
() +

x
i
(u
i
) =

x
j

+

t

x
j

+ C
1

k
(P
k
+ C
3
P
b
) C
2

2
k
+ S

. (4)
In Eqs. (3) and (4), P
k
represents the production of turbulence kinetic energy due to the mean
velocity gradients. P
b
is the production of turbulence kinetic energy due to buoyancy. Y
M
represents the contribution of the uctuating dilatation in compressible turbulence to the overall
dissipation rate. C
1
, C
2
, and C
3
are model constants.
k
and

are the turbulent Prandtl

numbers for k and , respectively. S
k
and S

are user-dened source terms.

The turbulent (or eddy) viscosity,
t
, is computed by combining k and as follows:

t
= C

k
2

, (5)
where C

is a constant.
The production of kinetic energy P
k
is given by:
P
k
= u

i
u

j
u
j
x
i

t
S
2
, (6)
where S

2
t
S
ij
S
ij
is the modulus of the mean rate-of-strain tensor considering the
Boussinesq hypothesis.
The effect of buoyancy P
b
is given by:
P
b
= g
i

t
Pr
t
T
x
i
, (7)
where T is the temperature, Pr
t
is the turbulent Prandtl number for energy and g
i
is the compo-
nent of the gravitational vector in the i
th
direction. For the standard and realizable models, the
default value of Pr
t
is 0.85. = (1/)(/T)
p
is the coefcient of thermal expansion. The
model constants are dened as: C
1
= 1.44, C
2
= 1.92, C
3
= 0.33, C

= 0.09,
k
= 1.0
and

= 1.3.
3 EXPERIMENTAL APPARATUS AND DATA
The experimental apparatus used in this work was made by Edibon Technical Teaching
Equipment with purposes of study and measurement of head loss in pipes and accessories
(Edibon, 2012). The components of this experimental facility are schematically shown in Fig. 1.
A centrifugal pump supplies the water to the experimental circuit. After pass by the experimen-
tal circuit the water ows downward to a collecting tank of 165 l. The water exits the collecting
tank to feed the centrifugal pump in a closed loop. The tests were conducted at nearly atmo-
spheric pressure of 1021 mbar and ambient temperature of 18

C. The water ow rate measure-

ment uses a rotameter calibrated at 100 l/h of uncertainty operating within the range of 600 l/h
to 6000 l/h. The Venturi is made of acrylic and your dimensions are shown in the Fig. 1.
Q
P3
Centrifugal
pump
Rotameter
(flow meter)
Valve
Venturi
(test section)
Water
reservoir
Bourdon
pressure gauge
60 mm 50 mm 20 mm
32 mm 32 mm 20 mm
P2 P1
U-tube
manometers
Water recycling circuit
Inlet Outlet
Throat Convergent Divergent
Water
circuit
Figure 1: Schematic representation of the experimental facility.
The pressure taps are localized at three axial positions of the Venturi: inlet, throat and outlet,
see Fig. 1. Two U-tube manometers are used to measuring the pressure differences between
inlet and throat and between throat and outlet. A Bourdon pressure gauge measures the pres-
sure at outlet relative to atmosphere. The pressure uncertainty was estimated at approximately
20 mbar. The atmospheric pressure reading comes from the weather service provided by
Simepar Technological Institute (available at http://www.simepar.br/site/).
3.1 Experimental data
The experimental test grid consists of one sets with 14 values of water ow rates Q measured
by the rotameter shown in Fig. 1, see Table 1. The ow rates range encompasses the region of
turbulent ow occurrence. In Table 1 the variables P
1
, P
2
and P
3
correspond to the absolute
pressures at the inlet, the throat and the outlet of the Venturi.
Table 1: Experimental test grid data.
Q Re P
1
(abs.) P
2
(abs.) P
3
(abs.)
[l/h] [] [Pa] [Pa] [Pa]
600 6552 108149 108051 108100
800 8737 108198 108012 108100
1000 10921 110208 109943 110100
1200 13105 112237 111865 112100
1400 15289 116257 115747 116100
1600 17473 120306 119639 120100
1800 19657 124031 123551 124100
2000 21842 128433 127374 128100
2200 24026 132522 131237 132100
2400 26210 138590 137080 138100
2600 28394 144561 142874 144100
2800 30578 150747 148678 150100
3000 32762 156885 154462 156100
3200 34947 164943 162247 164100
4 NUMERICAL RESULTS AND ANALYSIS
The computational domain used in the simulations corresponds to the internal region of the
Venturi geometry dened in Fig. 1, using boundary condition of symmetry around the revolution
z axis. The boundary conditions used as input parameters in the simulation are the inlet velocity,
based on Q, and the outlet absolute pressure, P
3
, obtained experimentally, see Table 1.
The computational mesh used was rened near the surface where the gradients are more
signicant and the rest of the eld was rened enough to provide a good convergence of the
results, see Fig. 2. All simulations were performed for permanent regime. The pressure-velocity
coupling was performed using the Simple method (Patankar and Spalding, 1972).
Figure 2: Computational mesh used.
The Fig. 3 shows a comparison of the numerical (num.) results against the experimental
(exp.) results for absolute pressure at the inlet, P
1
, and the throat, P
2
, as a function of ow rate
Q. These numerical results were obtained using the standard k model.
Figure 3: Comparison of numerical (num.) and experimental (exp.) results for absolute pressure values at inlet, P
1
(left), and at throat, P
2
(right), as a function of ow rate Q.
The results presented in the Fig. 3 show a good agreement between numerical simulation and
experimental data. The three turbulence models used presented a smaller relative deviation than
1% compared with the experimental data. This is due to the fact that the Venturi geometry is rel-
atively straightforward and hence the turbulence models used tends to show good performance.
Therefore, the results obtained in this work by these three turbulence models for simulation of
turbulent ows in the Venturi-type device are similar.
The Fig. 4 shows the distribution of the absolute pressure and the velocity through the Ven-
turi at ow rate of 3200 l/h. These results show the reduction of pressure in the throat, and
subsequent recovery of pressure in the outlet of the Venturi, as expected in this type of device.
Figure 4: Distribution of the absolute pressure (left) and velocity (right) through the Venturi at ow rate of 3200 l/h.
The values of the absolute pressure and the velocity are in Pa and m/s, respectively.
Near the Venturi outlet is possible to observe the existence of a counterow, see Fig. 5. This
phenomenon occurs due to vortices formation caused by non-slip condition on the wall in the
region downstream of the throat where the velocity magnitude is reduced. For the lower values
of ow rate this counterow is negligible, thus the ow is regular throughout the Venturi.
Figure 5: Detail of the counterow occurrence in the velocity distribution.
The Fig. 6 shows the distribution of the turbulent kinetic energy and the turbulent viscosity
through the Venturi at ow rate of 3200 l/h. These results show an increase in the turbulent
kinetic energy after the throat. This increase in turbulent kinetic energy explains the vortices
formation in the divergent section after the throat. This can also be observed by an increase in
the turbulent viscosity at this divergent section.
For all other values of ow rate presented in Table 1, the numerical results are similar to
those obtained with the ow rate of 3200 l/h with proportional values for each of the ow
characteristics shown in Figs. 4 and 6.
Figure 6: Distributions of turbulent kinetic energy (left) and turbulent viscosity (right) through the Venturi at ow
rate of 3200 l/h. The values of the turbulent kinetic energy and the turbulent viscosity are in m
2
/s
2
and kg/(m.s),
respectively.
5 CONCLUSIONS
This work presented a numerical analysis of turbulent ow in a Venturi-type device. The
results were obtained with three different versions of the k turbulence model: standard,
RNG and realizable.
The three turbulence models presented good results for the Venturi topology once this geom-
etry is relatively straightforward. The validation of the computational model presented a good
level of agreement with experimental results.
The divergent section in the Venturi presents higher values of turbulent kinetic energy and
turbulent viscosity. This characteristics can be used to maximize the mixture level after the
throat in some industrial applications of this device.
This study allowed a better understand for the behavior of turbulent ow in a Venturi device.
Next steps will be introducing complex phenomena in this analysis of Venturi-type devices, for
example: multiphase ow, particle injection, cavitation, etc.
REFERENCES
Baylar A., Aydin M.C., Unsal M., and Ozkan F. Numerical modeling of Venturi ows for deter-
mining air injection rates using Fluent v6.2. Mathematical and Computational Applications,
14(2):97, 2009.
Blocken B., van Hooff T., Aanen L., and Bronsema B. Computational analysis of the per-
formance of a Venturi-shaped roof for natural ventilation: Venturi-effect versus wind-
blocking effect. Computers & Fluids, 48(1):202213, 2011. ISSN 0045-7930. doi:
http://dx.doi.org/10.1016/j.compuid.2011.04.012.
Domingues R.B. Injection of FCC particles through a Venturi injector feeder. Masters Thesis,
State University of Campinas, 2006.
Edibon. Practical Exercises Manual of the AFT Unit. Edibon Technical Teaching Equipment,
Madrid, Spain, 1st edition, 2012. Manual of the experimental unit for study and measurement
of head loss in pipes and accessories.
Fox R.W., McDonald A.T., and Pritchard P.J. Introduction to uid mechanics, volume 7. John
Wiley & Sons New York, 2010.
Ghassemi H. and Fasih H.F. Application of small size cavitating Venturi as ow controller and
ow meter. Flow Measurement and Instrumentation, 22(5):406412, 2011. ISSN 0955-5986.
doi:http://dx.doi.org/10.1016/j.owmeasinst.2011.05.001.
Launder B.E. and Spalding D.B. The numerical computation of turbulent ows. Computer
Methods in Applied Mechanics and Engineering, 3(2):269289, 1974. ISSN 0045-7825.
doi:http://dx.doi.org/10.1016/0045-7825(74)90029-2.
McDonald A.T. and Fox R.W. An experimental investigation of incompressible ow in conical
diffusers. International Journal of Mechanical Sciences, 8(2):125139, 1966. ISSN 0020-
7403. doi:http://dx.doi.org/10.1016/0020-7403(66)90071-3.
Oliveira E.F., Botrel T.A., Frizzone J.A., and PAZ V.P.S. Hydraulic analysis of hydro-ejectors.
Scientia Agricola, 53:241248, 1996. ISSN 0103-9016.
Patankar S.V. and Spalding D.B. A calculation procedure for heat, mass and momentum trans-
fer in three-dimensional parabolic ows. International Journal of Heat and Mass Transfer,
15(10):17871806, 1972.
Puentes N.A.G., Guerra V.G., Coury J.R., and Gonalves J.A.S. Droplet dispersion angle mea-
surements on a Pease-Antony Venturi scrubber. Brazilian Journal of Chemical Engineering,
29:99106, 2012. ISSN 0104-6632.