Você está na página 1de 6

Proceedings of the ASME 2016 35th International Conference on Ocean, Offshore and Arctic Engineering

OMAE2016
June 19-24, 2016, Busan, South Korea

OMAE2016-54592
VORTEX SHEDDING CONTROL USING JETS: A COMPUTATIONAL STUDY WITH
LATTICE BOLTZMANN METHOD

Guoqiang Fu
Deepwater Engineering & Technology
Research Center
Harbin Engineering University
Harbin, Heilongjiang, China

Bassam A. Younis
University of California, Davis
Davis, California, USA

Liping Sun
Deepwater Engineering & Technology
Research Center
Harbin Engineering University
Harbin, Heilongjiang, China

Shaoshi Dai
Deepwater Engineering & Technology
Research Center
Harbin Engineering University
Harbin, Heilongjiang, China
induced vibration (VIV) and vortex-induced motion (VIM). In
some cases, resonance may also occur with the potential for
structural failure and disruption of operations. It is therefore not
surprising that a great deal of effort has been directed towards
finding a reliable and effective means for the reduction or
altogether suppression of vortex shedding from bluff bodies of
various shapes and, in particular, from smooth circular cylinders
which are taken to represent commonly-found structures in
offshore operations such as risers, spar platforms, and the
columns of tension-leg platforms. Zdravkovich [4] reviewed and
classified various methods for vortex-shedding control of which
a large number has been reported in the literature. Our interest in
this study is in a novel method that has not hitherto been the
subject of study. The method involves the use of jets to modify
the overall behavior of the flow in such a way as to reduce the
size of the shed vortices and with that, the magnitude of
oscillations in the lift and drag forces. The jets are introduced
from discrete holes located at the stagnation point of a twodimensional cylinder, and along the stagnation line of a threedimensional cylinder, and are directed into the approach flow.
The use of jets as a tool for flow control is by no means novel [57] the novelty here is in the mode of application.
The computations were performed using the Lattice
Boltzmann Method (LBM). The use of this method for fluidflow simulations is far less common than e.g. finite-volume
methods however there is a significant body of findings in the
literature that suggests that this method is more accurate and less
computationally intensive compared to others. Because of the
need to accurately resolve the initial region of the jets
development while also computing the flow over much greater
spatial and temporal scale, it was thought worthwhile to evaluate

ABSTRACT
The use of the Lattice Boltzmann Method (LBM) for fluidflow simulation has been the subject of several recent studies
where it was reported that the method offers many advantages
such as high accuracy coupled with computational efficiency. In
this paper, we report on the use of this method, in conjunction
with Large-Eddy Simulations, to study an interesting
phenomenon related to the suppression of vortex shedding from
circular cylinders. Specifically, it has been observed in
experiments that vortex shedding from a cylinder can be
drastically reduced by the injection of a fluid jet into the
approach flow. We first present results for a cylinder without jet
injection in order to quantify the suitability of the LBM for use
in such flows. Thereafter, we present results for the case with jet
injection where we consider the case where Re=55,440.
Preliminary results conclusively demonstrate that the presence
of jet injection does indeed lead to substantial reduction in the
magnitude of lift forces on the cylinder. This strongly argues in
favor of further research to understand the dynamics of this
phenomenon, and the range of parameters needed to maximize
its beneficial effects.
INTRODUCTION
The occurrence and consequences of vortex shedding from
bluff bodies immersed in turbulent flows are well documented in
experiments and computations [1-3]. In offshore engineering,
the occurrence of vortex shedding is invariably an undesirable
event leading, for example, to potentially damaging vortex-

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

the LBM in this flow to determine whether its performance


warrants further evaluation in a later study.

where

relaxation characteristic time which is related to the macroscopic


viscosity via:

NOMENCLATURE
L
Re
St
U
Uj

CD

Drag coefficient

(3)

Usually, the equilibrium distribution function takes the


following form:

Strouhal number ( St fL / U )
Velocity of incoming flow
Velocity of jet flow
Frequency of vortex shedding

FD
FL

CL

1
2

cs2 ( )t

Characteristic length (diameter for circular cylinder,


height for square cylinder)
Reynolds number ( Re UL / )

Dimensionless time (t* Ut / L)


Kinematic viscosity
Fluid density
In-line component of total force
Transverse component of total force

t*

f i eq is the local equilibrium function and is the

e u u u e e

(4)
fi eq i 1 i 2 2 i 2i
c
2
c
c
s
s
s

where cs is the speed of sound, u is the macroscopic


velocity, is the macroscopic density, is the Kronecker
delta and i are weighting constants built to preserve the

isotropy. The and indices denote the different spatial


components of the vectors appearing in the equation. The
Einstein summation convention over repeated indices has been
used.
The discrete velocity is evaluated using the D2Q9 model
(see Fig. 1):

2
CD FD / U D
2

2
Lift coefficient CL FL / U D
2

COMPUTATIONAL MODEL
Lattice Boltzmann Method
In the application of the Lattice Boltzmann Method (LBM)
to fluids in motion, the fluid is no longer treated as a continuum
but rather as being composed of particles which can flow along
some directions in space. This particle-based method uses
microparticle movement according to simple rules to take the
place of the complex macro-phenomena. The particle movement
at every time-step consist of two parts: streaming operation and
collision operation. Qian et al. presented the single relaxation
time (SRT) model of the LBM in what became known as the
LBGK (Lattice Bhatnagar-Gross-Krook) model [8~10]. The
LBGK model is the most widely used model in the LBM. It was
originally developed as an improved modification of the Lattice
Gas Automata to satisfy the Galilean invariance and remove
statistical noise. The Maxwell equilibrium distribution and BGK
collision theory are employed in the formulation.
The Boltzmann transport equation can be written as
follows:

Figure 1: The D2Q9 lattice

0



ei c cos i 1 ,sin i 1
4
4



2c cos i 1 ,sin i 1
4
4


Accordingly weighted parameters
4
9

1
i
9
1
36

fi
ei fi i , i 1,..., b,
(1)
t
where f i is the particle distribution function in the i-direction,
ei is the corresponding discrete velocity and i is the
collision operator which, in the BGK approximation, is:

1
iBGK ( fi eq fi )

for i=0
for i =1~4

(5)

otherwise.

are given as:

for i =0

(6)

for i =1~4
otherwise.

(2)

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

RESULTS AND DISCUSSION

Large Eddy Simulation


We use Large-Eddy Simulations (LES) to represent the
effects of turbulence on the flow. In this approach, the large-scale
motions that dominate the flow are resolved directly while the
small-scale motions are modeled. This is achieved by applying a
low-pass filter to the governing equation by defining a filter
function that separates a flow variable f (x, t ) into a large-

To check the ability of the LBM-LES method to accurately


simulate the phenomenon of vortex shedding, we apply the
method to the well-documented case of uniform flow past a fixed
square cylinder at Re=20000 (based on length side L = 1m, and
inlet velocity U=0.02m/s).

f (x, t ) and a small-scale quantity f (x, t ) :


(7)
f (x, t ) f (x, t ) f (x, t )

scale quantity

The computational domain is shown in Figure 2.

where f (x, t ) G(x, x) f (x)dx


and

G(x, x) is the filter function.

LBM-LES
The filtered Boltzmann equation can be now be obtained
from the differential form of the Boltzmann equation:

fi
ei fi i
t
where i is the collision operator.

Fig. 2

(8)

Table1: Square cylinder. Results at Re=20000

In the LBGK model,

1
fi (x ei , t t ) fi (x, t ) fi eq (x, t ) fi (x, t )

The total viscosity

consists of kinematic viscosity

0 t

and

(10)

The Wall-Adapting Local Eddy viscosity model, which


provides a consistent local eddy-viscosity and near wall
behavior, is used here. The actual implementation is formulated
as follows:

G G
S S G
d

5/2

g g
2

1 2
1
2
g g
g2

2
3
u

x

d
G

g
where

d
G
d

5/4

St

CD

281,492/0.080s

0.153

2.11

205,568/0.2s

0.153

2.044

152,258/0.233s

0.157

2.07

Target values

0.14

2.2

In order to test the sensitivity of the LBM to the number of


lattices employed, and the times-step size chosen, computations
were performed where both these parameters were varied over a
wide range. The results are presented in Table 1. The target
values listed there represent an average of results, both
experimental and computational, that have been reported in the
literature. It is clear from Table 1 that the LBM method produces
results that are in line with expectations bearing in mind that no
effort was made in order to improve the correlation with the
target value e.g. by making adjustments to the value of the
constant C .

3/2

number of lattices/ t
(9)

eddy viscosity t :

2
f

Square cylinder. Computational domain.

(11)

(12)
(13)

Turning next to the case of a circular cylinder in uniform


flow, computations were performed for a range of Reynolds
number to further test the LBMs performance in a case where
experimental data are available for model validation. The
computational domain was in the same configuration as before
(Fig. 2) and the Reynolds number set to Re=55440.
To test the sensitivity of the LBM to the number of lattices
and the time-step size for this case, computations were
performed for three numbers of lattices, and four different time
steps. The results are presented in Table 2. Shown there are the

(14)

f C x is the filter scale, and the constant C is

assigned its typical value of 0.2.

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

1.8

values of Strouhal number, the mean drag coefficient and the


root-mean-square of the lift and drag coefficients. Taken
altogether, and in the interest of computational efficiency, the
remainder of the results were obtained with the settings of Run
2.
Table 2: Circular cylinder. Results at Re=55440
Number
Run
CD
of
St
t (s)
CD
No.
lattices

1.6
1.4

CL

43606

0.0014

0.244

1.403

0.173

0.878

145696

0.0014

0.265

1.378

0.116

0.847

385348

0.0014

0.265

1.386

0.133

0.823

145696

0.0012

0.258

1.399

0.113

0.813

145696

0.0018

0.244

1.312

0.169

0.930

145696

0.0020

0.249

1.356

0.181

1.032

CD
1.2
1.0
0.8
2.0

The predicted time histories of the lift and drag coefficients


for the case of Re=55,440 are shown in Fig.3. The plots reveal
the establishment of a stable, periodic pattern of behavior in time
with a lift coefficient signal having a mean value of zero, and a
frequency of half that of the drag coefficient. These features are
in accordance with expectation since the cylinder is symmetric,
and since the fluctuations in drag are induced by each vortex that
is shed from the surface while the fluctuations in lift are induced
by every other vortex shed.
The predicted variation of the mean drag coefficient and
the Strouhal number with Reynolds number are compared in
Fig.4. Also shown there are the experimental results of Schewe
(1993), and the computational results of Younis and Przulj
(2006) obtained using a finite-volume method with an eddyviscosity turbulence closure. Neither of the two computational
approaches succeeds in predicting the correct values of St in the
low Re range where the experiments show a constant level of
just under 0.2 which the computations show values that range
from 0.22 to 0.27. It should be mentioned the lower values
obtained in the experiments were not obtained by any of the
other computational studies of this flow. The observed
discrepancy is probably due to the occurrence of transition in
the measurements something that will have depended a great
deal on the details of the test facility and which is not easily
reproducible by computations.

1.0

C L 0.0
-1.0
-2.0
40
Fig.3

60

t * 80

100

Circular cylinder. Time histories of

CL and CD

(Re=55,440).

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

coherent concentration of vortices behind the cylinder.


Significant differences in flow patterns appear at the front of the
cylinders, around the stagnation point. The reason for this quite
spectacular is result is apparent from the plot: the injection of a
jet from the forward stagnation point has energized the boundary
layers that develop over the cylinder walls thereby causing the
flow to remain attached for a greater turning angle. Separation is
that delayed leading to the formation of wakes that are much
reduced in size and strength than is the case for the cylinder with
no jet.

0.5
LBM-LES

0.4

Younis and Przulj(2006)


BL, Deng et al.(1993b)
Exp.,Schewe(1993)

0.3

St
0.2
0.1
0.0
3
10

10

10

10

10

Re
2.0
LBM-LES
Exp.,Schewe(1993)
Younis and Przulj(2006)
Zdravkovich(1990)
3D, Tamura et al.(1990)

1.6
1.2

CD

0.8
Fig.5

0.4
0.0 3
10

10

10

10

10

A further clear demonstration of the effectiveness of this


method in suppressing vortex shedding is evident in Fig. 6 which
shows the time series of the lift coefficient before the
introduction of the jet, and afterwards. Quantitatively, the
reduction in the maximum value of the lift coefficient and in
Strouhal number arising from the jet are presented in presented
in Table 3.

Re
Fig.4

Computed vorticity patterns without (above) and


with jet flow (top).

Circular cylinder. Computed mean drag coefficient


and Strouhal number [12]

The case of uniform flow past a fixed circular cylinder with


a jet injected from the front stagnation point directly into the
approach flow is considered next. In the results presented below,
the ratio of jet velocity to incoming velocity was 3. All
computational details were as for the case of no jet flow and
hence the differences observed below arise solely from the
introduction of the jet.
An overall impression of the changes to the flow arising
from the jet can be seen from Figure 5. Plotted there are the
contours of (absolute) vorticity generated by the cylinder with
and without the jet. The flow pattern that emerges in the absence
of the jet is in line with expectation in that it shows a well
organized pattern of vortices shed alternately from the cylinder.
In contrast, for the case where the jet is present, this pattern is
entirely altered most notably in the total absence now of a

CONCLUSIONS
In this paper, we report on a preliminary study into an
interesting effect namely the near suppression of vortex shedding
from a circular cylinder under the effect of a jet injected from the
front stagnation point directly into the approach flow. The
computations were performed using the Lattice Boltzmann
Method in conjunction with Large Eddy Simulation. The method
was first applied to the prediction of the benchmark case of
turbulent flow around a square cylinder over a range of Reynolds
number with results, in terms of average drag coefficient and
Strouhal number, broadly in accord with those obtained with a
finite-volume method and a turbulence closure. The method was
then applied to the case of a round cylinder at Re = 55,440 first

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

ACKNOWLEDGMENTS

with no jet being applied in order to obtain baseline results, and


then with a jet introduce at an injection velocity ratio (of jet to
approach velocity) of 3. Visualization of the flow, which was
accomplished via contours of vorticity at an instance in time,
showed large-scale modification of the flow pattern around and
behind the cylinder signifying major reduction in the strength of
vortex shedding. This effect, which is also evident in the fivefold decrease in maximum lift, is attributed to the energization
of the near-surface flow that has the effect of keeping the
boundary layer attached for a greater turning angle. Work is now
in progress to explore this phenomenon in more detail and to
determine the range of parameters for which it can be used most
effectively.

The work is sponsored by the National Natural Science


Foundation of China (Grant No. 11472087).
This paper is funded by the International Exchange
Program of Harbin Engineering University for Innovation
Oriented Talents Cultivation.
REFERENCES

2.4

without jet flow

1.6

with jet flow

0.8

CL

0.0

-0.8
-1.6
20
Fig.8

40

60

80

t*

100

120

140

Circular cylinder. Predicted CL with and without jet

Table 3: Circular cylinder. Comparison of flow parameters


without jet

with jet

St

0.265

0.067

Maximum CL

1.199

0.062

[1] von Karman T. Collected works of Theodore von Karman.


London, UK: Butterworths Scientific Publications 1956.
[3] Williamson C H K. Vortex dynamics in the cylinder wake .
Annual Review of Fluid Mechanics, 1996, 28(1): 477-539.
[4] Zdravkovich M M. Flow around Circular Cylinders; Vol. I
Fundamentals. Journal of Fluid Mechanics, 1997, 350(1):
377-378.
[5] Zdravkovich M M. Review and classification of various
aerodynamic and hydrodynamic means for suppressing
vortex shedding. Journal of Wind Engineering and Industrial
Aerodynamics, 1981, 7(2): 145-189.
[6] Tensi J, BouI, PaillF, Modification of the wake behind a
circular cylinder by using synthetic jets. Journal of
Visualization, 2002, 5(1): 37-44.
[7] Amitay M, Honohan A, Trautman M, Modification of the
aerodynamic characteristics of bluff bodies using fluidic
actuators. AIAA paper, 1997, 2004.
[8] Bra J C, Michard M, Sunyach M, Changing lift and drag by
jet oscillation: experiments on a circular cylinder with
turbulent separation. European Journal of MechanicsB/Fluids, 2000, 19(5): 575-595.
[9] Chen S, Chen H, Martnez D, Lattice Boltzmann model for
simulation of magnetohydrodynamics. Physical Review
Letters, 1991, 67(27): 3776.
[10] Koelman J. A simple lattice Boltzmann scheme for NavierStokes fluid flow. Europhysics Letters, 1991, 15(6): 603.
[11] Qian Y H, d'Humires D, Lallemand P. Lattice BGK models
for Navier-Stokes equation. Europhysics Letters, 1992,
17(6): 479.
[12] Younis, BA, and Przulj, V, Computation of Turbulent
VortexShedding, Computational Mechanics, 2006, 37: 408425.

Copyright 2016 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 11/16/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Você também pode gostar