You are on page 1of 4

CHIN. PHYS. LETT. Vol. 28, No.

12 (2011) 124704

Wake Oscillator Model Proposed for the Stream-Wise Vortex-Induced Vibration


of a Circular Cylinder in the Second Excitation Region *
XU Wan-Hai()** , DU Jie(), YU Jian-Xing(), LI Jing-Cheng()
Key Laboratory of Port and Ocean Engineering, Tianjin University, Tianjin 300072

(Received 30 August 2011)


A wake oscillator model is presented for the stream-wise vortex-induced vibration of a circular cylinder in the
second excitation region. The near wake dynamics related to the fluctuating nature of alternate vortex shedding
is modeled based on the classical van der Pol equation. An appropriate approach used in cross-flow VIV is
developed to estimate the model empirical parameters. The comparison between our calculations and experiments
is carried out to validate the proposed model. It is found that the present model results agree fairly well with
the experimental data.

PACS: 47.32.Cc, 47.85.Dh, 47.11.+j

DOI:10.1088/0256-307X/28/12/124704

Vortex-induced vibrations (VIVs) of a circular


cylinder occur in many industrial applications, such as
thermowells, heat exchangers, underwater cables and
risers. They may cause significant fatigue damage.
Thus this phenomenon must be addressed in design
of such structures. Some studies on VIVs have been
comprehensively reviewed by Sarpkaya,[1] both experimental and theoretical investigations of the fundamental aspects of vortex-induced vibration of circular
cylinders are discussed in some detail.
Many of the works were, however, related to crossflow oscillations, and few studies were reported on
stream-wise (or in-line) oscillations. Flow-induced
stream-wise vibrations of various kinds of cylindrical
and axisymmetric bodies were reviewed in depth by
Naudascher.[2] There are two different modes of excitation within in-line VIVs, the first excitation region originates from symmetric vortex shedding in the
lower reduced velocity region of 1.0< < 2.32.5,
while the second excitation region from alternating
vortex shedding in the higher reduced velocity region
of 2.32.5< < 3.8 ( is the reduced velocity, defined by = /( ), where is the incident flow
velocity, is the natural frequency of the cylinder
and is the cylinder diameter).[3] Flow-induced inline oscillation of a two-dimensional circular cylinder
model was experimentally investigated in a wind tunnel using the free-oscillation method in order to understand some of the fundamental characteristics of
the system by Matsuda et al.[4] Recently, Okajima et
al.[5] conducted free oscillation tests in a water tunnel, instead of a wind tunnel, the rigid cylinder models were elastically supported at both ends, the massdamping parameter (= 2; is the mass ratio,
is the structural damping factor) was varied over a

wide range, in order to evaluate the critical value at


which the in-line oscillation is suppressed.
A large number of in-line VIV experimental results
have been reported. However, only a few analytical
models have been proposed for the stream-wise oscillations of structures. One mathematical model has been
developed by Currie and Turnbull,[6] attempting to
represent a cylinder vibrating in the in-line direction.
This mathematical model is based on the Van der Pol
equation and is similar to the lift wake-oscillator for
cross-flow oscillations. They thought that the cylinder
oscillations in the second excitation region were a simple harmonic of the velocity driven transverse cylinder
oscillations, while those in the first excitation region
might be amplitude driven.
For in-line VIVs, very few theoretical studies have
proposed a model governing the drag coefficient, and
a practical tool for predicting in-line VIVs is still unavailable in industry. The objective of this study is to
propose a simple model for the near wake dynamics of
a cylinder with the second excitation region of streamwise oscillations, a wake oscillator, which is based on
the Van der Pol equation, is used to model the alternating vortex shedding behind the cylinder. The
empirical parameters in the wake oscillator model are
calibrated and used. Finally, we compare our calculation results with the experimental data for validating
the proposed model.
A cylinder subjected to vortex-induced oscillations
in the stream-wise direction (Fig. 1) is described as a
damped mass-spring oscillator,[6]
(

)
(1)
+ 2 + + 2 = ,

where is the coordinate of the cylinder axis in

* Supported by the Specialized Research Fund for the Doctoral Program of Higher Education of China (SRFDP)
(20100032120047), the Independent Innovation Fund of Tianjin University (2010XJ-0098), and the National Natural Science Foundation of China (10902112).
** Correspondence author. Email: wanhaixu@hotmail.com; xuwanhai@tju.edu.cn
c 2011 Chinese Physical Society and IOP Publishing Ltd

124704-1

CHIN. PHYS. LETT. Vol. 28, No. 12 (2011) 124704

the stream-wise direction, noting that the overdot denotes the derivative with respect to the dimensional
time ; is the structure reduced damping, is
the structure angular frequency and the vortex shedding angular frequency = 2/, where
is the Strouhal number, is a stall parameter, defined as = /4,[7] is the mean drag coefficient. The dimensionless mass ratio = /2
with being the cylinder mass, the fluid density,
the cylinder diameter. Both the mass of the structure and the fluid-added mass are taken into account.
= 2 /2 is the forcing term induced by vortex
shedding with is the fluctuating drag coefficient.
Z
U
Y

wake structure, in dimensionless variables, and it can


be expressed as
= ,

0
.
16 2 2

(5)

The action of the structure on the fluid wake oscillator has several choices, named as displacement coupling = , velocity coupling = and acceleration coupling =
,[7] where and are empirical
coefficients to be determined. Because a static in-line
displacement of the structure in a uniform flow does
not modify the fluctuating nature of the near wake,
an in-line displacement of the structure at a constant
velocity only changes the mean drag force. Therefore,
it is advisable that the coupling term of structure and
wake is acceleration.

Cylinder amplitude,

0.15

Fig. 1. Model of coupled cylinder structure and wake for


in-line vortex-induced vibrations.

In the present study, the second excitation region


is defined by the reduced velocity range from 2.3 to
3.8. The mean drag force term does not contribute
to the cylinder oscillations, the instantaneous drag in
this region is assumed to be associated with a vortex shedding frequency corresponding to two times the
Strouhal number, and thus the instantaneous drag coefficient satisfies the Van der Pol equation[8]
+ ( 2 1) + 42 = .

(2)

The dimensionless wake variable may be associated


with the fluctuating drag coefficient on the cylinder,
it is defined as = 2 /0 , where 0 could be
interpreted to the fluctuating drag coefficient of a stationary cylinder. is the forcing term, which models
the effects of the cylinder motion on the near wake.
Through the forcing terms to the Van der Pol equation for the instantaneous fluctuating drag coefficient
and the forcing term induced by vortex shedding to
a damped mass-spring oscillator, the wake and structure are coupled. Introducing the dimensionless time
= and space coordinate = /, Eqs. (1) and
(2) can be rewritten in dimensionless forms:
(
)
+ 2 = ,
(3)
+ 2 +

+ ( 2 1) + 4 = ,
(4)
where = / is the reduced angular frequency
of the structure, is the action of the fluid near the

0.10

0.05

0.00

1.2

1.5

1.8

2.1

2.4

2.7

3.0

Angular frequency

Fig. 2. Lock-in bands in the (, 0 ) plane for synchronization of vortex shedding with in-line cylinder vibration.
Diamond: numerical result for = 300.[12] Plus: experimental data for = 200.[13] Triangle: for = 190.[14]
Solid line: for model parameter = 12. Dashed line:
= 8.

Next, we calibrate all the parameters described


above. The fluctuating drag coefficient of the stationary cylinder 0 taken is equal to 0.2.[9] The Strouhal
number depends on the Reynolds number, for the
sake of simplicity, assuming = 0.17,[10] the mean
drag coefficient = 1.2, the value of as well
as that of is determined from the experimental results on forced and free VIVs, we have used a similar approach in the determination of a wake oscillator
models empirical coefficients in cross-flow VIVs.[11]
In-line oscillation of the cylinder and wake is defined
as
= 0 cos(),

(6)

= 0 cos( ),

(7)

where 0 and 0 are the dimensionless amplitudes


of the cylinder and wake, respectively, is angular
frequency, is phase. The action of the structure on the fluid wake oscillator could be written as
= 2 cos(). A reduced velocity = 1/()

124704-2

CHIN. PHYS. LETT. Vol. 28, No. 12 (2011) 124704

is defined based on the angular frequency,[11] enforcing


the hypothesis of harmonicity and frequency synchronization. Substituting Eqs. (6) and (7) in the wake
oscillator, Eq. (4), only the main harmonic contribution of the nonlinearities is considered, the amplitude
of the transfer function of the wake oscillator is yielded
using elementary algebra, it reads
[
( 2 4 )2 ]
( 2 )2
0
06 804 + 16 1 +
02 = 16
. (8)

0.15

= 2, the mass-damping parameter is used in


the stream-wise direction instead of (= 2 3 2 )
in cross-flow direction. The maximum structure displacement amplitude reads

0
max =
1+
.
16 2 2 (2 + )
16 2 2 (2 +)
(11)
Experimental measurements of the modally normalized maximum amplitude versus the response parameter is plotted in Fig. 3. It can be found that the
proposed curve-fit formula could be written as

nd

_ max

0.12

2nd_max = 0.1720.949 .

0.09

Assuming that the maximum structure displacement


amplitude is the same as the experimental results presented by Eq. (12), according to Eqs. (9), (11) and
(12), we can obtain the specification of the empirical
parameter .

0.06
nd _ max=0.172

-0.949

0.03

0.00

(12)

0.06

Present model
Experiment

0.05

Fig. 3. The normalized maximum amplitude versus the


mass-damping parameter and the proposed curve fitting. Diamond: cantilevered cylinder response amplitude
in a water tunnel.[15] Circle: two-dimensional circular
cylinder response amplitude in a wind tunnel.[16] Triangle: two-dimensional circular cylinder response amplitude
in a water tunnel. [17]

The free wake oscillator response 0 = 2 is supposed to prevail on the forced response, a lock-out
state is defined, looking for the boundary from polynomial (8) in the (, 0 ) plane. In Fig. 2, the points of
numerical simulation[12] and bounds of the lock-in region, adapted from the experimental results[13,14] are
plotted. The parameter may be chosen by matching
the model response (8) to experimental and numerical
data on lock-in extension in previous literature. It can
be expressed as

rms

0.01

0.00
2.2

Considering the appearance of the lock-in phenomenon, the basic resonance state displayed =

2.4

2.6

2.8

3.0

3.2

3.4

3.6

3.8

Fig. 4. The response amplitude of the circular cylinder


with = 16.0, = 0.77.
0.06

0.05

Present model
Experiment

0.04

(9)

For a given set of model parameters, the amplitude


of cylinder vibration is a function of the structural
damping factor , the mass ratio and the ratio of
the nominal vortex shedding frequency to the natural frequency of the cylinder structure. Substituting
Eqs. (6) and (7) in the dimensionless structure model,
Eq. (3), the amplitude y0 is found in the form:
[(
)2 (
)2 2 ]1/2

0 = 2 2 2 + 2 +

[
2 ( 2 2 )( 2 4) ]1/2
(
)
1+
.
2 2 + /
(10)

0.03

0.02

rms

= 12 for 0 2, = 8 for > 2.

0.04

0.03

0.02

0.01

0.00
2.2

2.4

2.6

2.8

3.0

3.2

3.4

3.6

Fig. 5. The response amplitude of the circular cylinder


with = 10.5, = 1.58.

We validated the simulation results using the wake


oscillator model described in the previous section by
comparing them with some experimental data. The
reduced velocity is varied from 2.3 to 3.8. The
readers can refer to Refs. [5,15,17] for details of the

124704-3

CHIN. PHYS. LETT. Vol. 28, No. 12 (2011) 124704

experiments. Okajima et al.[5,15,17] studied the flowinduced in-line oscillation of a circular cylinder; the
experiments were carried out by free-oscillation tests
in a water tunnel at subcritical Reynolds numbers.
The relationship between movement of a cylinder and
flow of a near field during an in-line oscillation was
investigated. We simulate the two cases of the experiments, = 16.0, = 0.77 and = 10.5,
= 1.58. The rms displacement response amplitude rms is calculated. The comparison between our
calculation and experiment with = 16.0, = 0.77
is shown in Fig. 4. It can be seen that the simulation
response amplitude of the in-line oscillation increases
from = 2.3 to 3.0, and decreases from = 3.0 to
3.8, with the maximum value obtained at = 3.0.
We plot the numerical results and the experimental
data with = 10.5, = 1.58 in Fig. 5. It can be
shown that there is an increase in the stream-wise
amplitude from = 2.3 to 2.9, and a decrease from
= 2.9 to 3.8, with the maximum value obtained
at = 2.9. The maximum response amplitude obtained by the present model and experiment is nearly
the same. From Figs. 4 and 5, we can find that the
present model results agree fairly well with the experimental data, some aspect of the dynamics observed
with experiments can be reproduced.
In summary, we have proposed a wake oscillator
model for the stream-wise VIV of a circular cylinder
in the second synchronization region. The predicted
results are compared with the experimental data to
validate the present model. Good agreement is obtained, and some aspects of the cylinder in-line VIV
could be reproduced. These comparisons show the applicability and usefulness of the present model for predicting the in-line VIV of engineering structures. Such

a model becomes really useful when computational


limits arise for flow-field numerical simulations, particularly at high Reynolds numbers. Moreover, phenomenological models based on wake oscillators allow
accessible analytical considerations and thus help the
understanding of the physics of the stream-wise VIV.

References
[1] Sarpkaya T 2004 J. Fluids Structures 19 389
[2] Naudascher E 1987 J. Fluids Structures 1 265
[3] Okajima A, Kosugi T and Nakamura A 2002 ASME J. Pressure Vessel Technol. 124 89
[4] Matsuda K, Uejima H and Sugimoto T 2003 J. Wind Engin. Industrial Aerodynamics 91 83
[5] Okajima A, Nakamura A, Kosugi T, Uchida H and Tamaki
R 2004 Eur. J. Mech. B Fluids 23 115
[6] Currie I G and Turnbull D H 1987 J. Fluids Structures 1
185
[7] Facchinetti M L, de Langre E and Biolley F 2004 J. Fluids
Structures 19 123
[8] Furnes G K and Sorensen K 2007 Proceedings of the 17th
International Offshore and Polar Engineering Conference
(Lisbon, Portugal: ISOPE) 2781
[9] King R 1977 Ocean Engin. 4 141
[10] Finn L, Lambrakos K and Maher J 1999 Proceedings of the
Fourth International Conference on Advances (Aberdeen
Scotland: Riser Technologies)
[11] Xu W H, Wu Y X, Zeng X H, ZHONG X F and YU J X
2010 J. Hydrodyn. 22 381
[12] Nobari M R H and Naderan H 2006 Computers and Fluids
35 393
[13] Hall MS and Griffin OM 1993 Trans ASME, J. Fluids Engin. 115 283
[14] Griffin O M and Ramberg S E 1976 J. Fluid Mech. 75 257
[15] Nakamura A, Okajima A and Kosugi T 2001 JSME Int. J.
B 44 705
[16] Matsuda K, Uejima H and Sugimoto T 2003 J. Wind Engin. Industrial Aerodynamics 91 83
[17] Okajima A, Kosugi T and Nakamura A 2001 JSME Int. J.
B 44 695

124704-4

You might also like