Ocean Engineering 151 (2018) 298–307

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Wake and suppression of flow-induced vibration of a circular cylinder

Sangil Kim a, Md Mahbub Alam b, * , Dilip Kumar Maiti c
a
Department of Mechanical Engineering, Kangwon National University, Republic of Korea
b
Institute for Turbulence-Noise-Vibration Interaction and Control, Shenzhen Graduate School, Harbin Institute of Technology, Shenzhen 518055, China
c
Department of Applied Mathematics with Oceanology and Computer Programming, Vidyasagar University, Midnapur WB, India

A R T I C L E I N F O
This experimental study examines the wake flow and the flow-induced vibration of an elastically supported
circular cylinder (referred to as test cylinder, of diameter D), and then the suppression of the flow-induced vi-
Keywords:
Cross flow vibration bration by a control cylinder (of diameter 0.5D) installed behind the circular test cylinder is investigated at
Vortex efflux angle Reynolds number Re ¼ 1.4  104–3.2  104 corresponding to the reduced velocity Ur ¼ 0.4–22. The streamwise
Circular cylinder and lateral separations of the control cylinder from the test cylinder center are varied as 0.2D - 1.1D and 0 - 1.7D,
Suppression respectively. The vortex trail encompassing the vortex generation to the disappearance in the wake of the test
cylinder is identified by means of wake energy distributions obtained from power spectra of fluctuating
streamwise velocities. Based on the energy distributions, the control cylinder positions are determined to suppress
the flow-induced vibration. The process of vortex ‘Generation → Growth → Disappearance’ in the wake of a
vibrating cylinder occurs closer to the cylinder compared to that of a non-vibrating cylinder. The lateral sepa-
ration between the two rows of vortices for a vibrating cylinder is found to be smaller. The controlled vibration
responses of the cylinder are classified into four patterns (I, II, III, and IV) depending on the control cylinder
position. Patterns I and II, both self-limited amplitude vibrations, are characterized by the occurrence of vibra-
tions at a smaller and a larger Ur, respectively, compared to the non-controlled counterpart. Pattern III represents
an increasing vibration amplitude with an increase in Ur, resembling a galloping vibration. Pattern IV features
vibration suppression resulting from the suppression of the Karman street in the wake. This regime is divided into
two subregimes, IV-A and IV-B. While the interaction between the gap flow and freestream side shear layer of the
test cylinder succeeds in the suppression of the Karman street in pattern IV-A, that between the gap shear layer
and the control cylinder gives rise to the suppression of the Karman street in pattern IV-B.

1. Introduction tandem, staggered and side-by-side arrangements are studied by Kim et

Circular cylindrical structures are commonly seen in our daily lives
such as risers, bridge piers, pipelines, most of which stand against
flowing fluids. The alternate vortex shedding from the structures may
produce a large fluctuating pressure on the structures, causing structural
vibrations, acoustic noise, and even resonance, which can trigger struc-
tural failure. Numerous failures in the practical applications of cylindrical
structures in cross flow are illustrated in Chen (1987), Paıdoussis (1993)
and Blevins (1990). The cost associated with a typical engineering
structural failure can easily reach the order of one million dollars and
even billions of dollars. Naturally, there is a pressing need to study and to
understand the fluid dynamics associated with multiple cylindrical
structures in cross flow. Accordingly, many studies have delved into the
interaction between flowing fluid and structures.
Flow-induced vibrations of two identical diameter cylinders in
al. (2009a), Alam and Kim (2009), and Kim and Alam (2015), respec-
tively, where they found that flow-induced vibration characteristics of
the cylinders are highly contingent on the cylinder arrangement and
spacing between the cylinders. Kim and Alam (2015) for two side-by-side
identical cylinders identified four vibration response patterns, namely (i)
the two cylinders vibrating at a large amplitude in the same range of Ur at
0.1  T* < 0.2, (ii) no vibration generated for either cylinder at
0.2  T*  0.9, (iii) the two cylinders vibrating at different ranges of Ur at
0.9 < T* < 2.1, and (iv) each cylinder response resembling an isolated
cylinder response at 2.1  T*  3.2, where T* is the cylinder gap spacing
ratio. Lam and To (2003) performed an experimental study where the
downstream cylinder was flexible and half in diameter than the upstream
cylinder. No vibration was observed. The cause of no vibration was
attributed to the fact that the upstream cylinder of being larger diameter
shelters the downstream cylinder. Tu et al. (2014) numerically studied

* Corresponding author.
E-mail address: alam@hit.edu.cn (M.M. Alam).

https://doi.org/10.1016/j.oceaneng.2018.01.043
Received 3 September 2016; Received in revised form 2 June 2017; Accepted 8 January 2018

0029-8018/© 2018 Elsevier Ltd. All rights reserved.

S. Kim et al.
Fig. 1. Free-vibration experimental setup and coordinate system.
Fig. 2. Position of velocity measurements using a one-dimensional hot-
wire probe.
Fig. 3. Visualization test equipment using water channel.
flow-induced vibrations of an elastically mounted circular cylinder sub-
jected to a planar shear flow with the single- (only cross-flow direction)
and two-degree of freedoms (both in-line and cross-flow directions) in
the laminar flow (Re ¼ 150). The effects of shear rate (k ¼ 0.0–0.1),
reduced velocity (Ur ¼ 3.0–12.0) and natural frequency ratio
(r ¼ 1.0–2.0) on the characteristics of vortex-induced vibration (VIV)
responses were studied. Chen et al. (2013) investigated the VIV of a
circular cylinder under a suction flow, reporting an optimal suction flow
299
Ocean Engineering 151 (2018) 298–307
Fig. 4. Flow-induced vibration response of the test cylinder (no control):
dependence of amplitude ratio a/D on reduced velocity Ur.
rate at which VIV was completely suppressed. The rotary oscillation
control can be implemented to suppress the response amplitude of VIV by
locking the vortex shedding frequency at the forcing frequency. Using a
traveling wave wall method, Xu et al. (2014) focused on the suppression
of the VIV of an elastically supported circular cylinder with a two-degree
of freedom at a low Re ¼ 200.
To investigate the feasibility for the flow control around a cylinder,
Strykowski and Sreenivasan (1990) used a control cylinder of the
diameter ranging from 0.05D to 0.33D and observed that a placement of a
smaller cylinder in the near wake of the cylinder can alter the vortex
shedding. The temporal growth rate of disturbances was weakened and
drag was reduced. Zhao et al. (2007) performed a similar numerical
investigation using a control cylinder of diameter 0.5D. Sakamoto et al.
(1991) examined the suppression of the fluid forces by adding a control
cylinder in a shear layer of a square cylinder. The reduction was, how-
ever, more for the fluctuating forces than for the time-mean forces, 95%
for fluctuating lift, 75% for fluctuating drag, and only 30% for time-mean
drag. Wu et al. (2012) conducted experiments with a focus on vibration
suppression of a deepwater riser by using four control cylinders, each of
diameter of 0.25D. The vibration was suppressed for certain configura-
tions. The mechanism of vibration suppression, however, remained un-
clear. Igarashi and Terachi (2002) examined the wake of a flat plate
adding a control cylinder of diameter (0.04–0.4)D and found a larger
reduction in drag for a larger diameter control cylinder.
Wu et al. (2014) numerically investigated VIV characteristics of a
cylinder controlled by a hinged flat plate. The addition of the hinged
plate efficiently suppressed the VIV of and force fluctuations on the
cylinder. An active control method for suppressing VIV response of an
elastically mounted cylinder by forcing rotary oscillation was presented
by Du and Sun (2015). Kim et al. (2009b) used tripwires to suppress VIV
and galloping vibrations of a single and two tandem cylinders. Tripping
wire positions α measured from the leading stagnation line of the cylin-
der were changed from α ¼ 20 to 60 to determine the optimum
range of α for suppressing structural vibrations. The vibrations on both
cylinders were completely suppressed for α ¼ 20 –30 . Lu et al. (2014) in
a laminar flow investigated the effect of rod-to-cylinder spacing ratio, rod
and cylinder diameter ratio, Re, and angle of attack on the main circular
cylinder. It was observed that the range of the spacing ratio where sig-
nificant force suppression is achieved becomes narrower as the Re in-
creases in the laminar regime but was insensitive to the diameter ratio.
Among the studies delving into the topics associated with circular
cylinders, the studies that examined the suppression of vibration and
fluid forces provide some basic techniques to prevent failure or to extend
the service life of cylindrical structures. This study aims (i) to identify the
behavior of vortex flow created by a free-vibrating circular cylinder and
to investigate some aspects, like the vortex generation, growth, decay,
S. Kim et al.
Fig. 5. Time-mean streamwise velocity distribution in the wake of the test
cylinder at different x for (a) Ur ¼ 4.03, (b) Ur ¼ 5.73, and (c) Ur ¼ 9.42.
efflux angle, etc, and (ii) hence to search optimum positions of the con-
trol cylinder in order to suppress the flow-induced vibration of a cylinder.
In addition, the results obtained from a wind-tunnel test are demon-
strated through a visualization test in a water channel.
2. Experimental details
The free vibration test was conducted in an open-loop wind tunnel at
Kangwon National University, Korea. The test section of the tunnel was
300
Ocean Engineering 151 (2018) 298–307
Fig. 6. Energy distribution in the wake of the test cylinder at different x for
(a) Ur ¼ 4.03, (b) Ur ¼ 5.73, and (c) Ur ¼ 9.42.
0.4 m in width, 1.0 m in height and 1.2 m in length. The supporting
structure for the free vibration test was installed outside of the working
section. As illustrated in Fig. 1, the test cylinder was mounted on an
aluminum frame structure. The test cylinder used in the free vibration
test was 66 mm in diameter (D) and had a 10-mm-diameter aluminum
shaft at the cylinder center. Three plastic disks, two of them at the two
ends and one at the midsection of the cylinder, were fitted to support the
shaft at the cylinder axis. The shaft at the cylinder was extended to the
outsides of the tunnel to anchor support assemblies, as shown in Fig. 1.
Each support assembly consisted of two leaf springs each 0.3 mm thick
(phosphor bronze) and two coil springs, allowing the cylinder to vibrate
in the vertical direction (cross-flow) only. The cylinders spanned the
0.4 m dimension of the tunnel. The cylinder to be as light as possible was
a paper tube, wrapped by a vinyl film of 0.1 mm in thickness. The tube
had a wall thickness of 1 mm and a weight of 0.21 kg/m.
The cylinder including the shaft, disk, etc, had a mass m ¼ 0.267 kg,
S. Kim et al.
Fig. 7. Flow visualization results in the water channel for the cases of (a) no
vibration and (b) vibration. Re ¼ 700.
Fig. 8. The control cylinder placement marked by ‘ ’.
yielding a mass ratio m* ¼ m/mf ¼ 226.5, where mf is the mass of the fluid
displaced by the cylinder. The resultant natural frequency fn of the cyl-
inder system in still air was 11.1 Hz. The cylinder in still air was given an
initial displacement and then released to vibrate. Due to structural
damping, the vibration amplitude dampens with time. When the cylinder
was about to stop, it was again displaced and released. This was done a
few times. Simultaneously, the cylinder vibration was recorded using a
laser displacement meter of 40 mm range with a resolution of 10 μm.
The fn was determined from the Fourier power spectrum of the vibration
301
Ocean Engineering 151 (2018) 298–307
signal. The structural damping ratio ζ ¼ 0.0033, estimated from the
logarithmic decrement of the amplitude.
Streamwise flow velocities (u) at a series of locations in the cylinder
wake were measured using a hot wire probe as illustrated in Figs. 1 and
2, providing time-mean and fluctuating velocity fields. The hotwire sig-
nals were offset, low-pass filtered (cut off frequency 1.0 kHz), amplified,
and the digitized using an A/D board at a sampling frequency of 500 Hz.
The sampling duration was 20 s for each record. The velocity measure-
ment positions, denoted by dots, in the wake of the cylinder are displayed
in Fig. 2. The measurements were done at six streamwise distances
x ¼ 0D, 1D, 2D, 3D, 4D and 5D measured from the rear stagnation point,
with lateral distances of y ¼ 2D ~ 3D depending on x.
A propeller-type velocity meter was used to measure freestream ve-
locity U∞, having a resolution of 0.01 m/s in a range of 0.6–40 m/s. The
reduced velocity Ur (¼ U∞/fnD) was varied from 1.5 to 22 by changing
U∞ from 1 to 15 m/s, corresponding to Re ¼ 4 365–65470, with the
freestream turbulent intensity less than 0.55%. To avoid the damage to
the cylinder, Ur was not further increased when the vibration amplitude
reached 0.25D, given m*ζ ¼ 0.75. Measurement uncertainties in Ur, vi-
bration amplitude, fn, and ζ were estimated to be 2%, 1.6%, 2%, and
2.5%, respectively.
Furthermore, to observe the flow structure around the vibrating cyl-
inder, a flow visualization test was performed in a water channel
(width  depth  length ¼ 0.3  0.4  2.0 m test section) at Kitami
Institute of Technology, Japan. The vibration was given using a forced
vibration setup shown in Fig. 3. A pair of slide rails was placed on the
water channel equipment, and a transparent acryl plate was attached on
it to enable one-dimensional reciprocal movement. The acryl plate was
connected to the crank-link shaft driven by the motor so that the rotation
of the motor generates reciprocating motion of the acryl plate, hence of
the cylinder, in the crossflow direction. The same setup has also been
subsequently used for two cylinders by Alam and Kim (2009). The cyl-
inder model was of D ¼ 20 mm and length L ¼ 400 mm, while a length of
250 mm was submerged in water. The vibration frequency of the cylinder
could be controlled by the number of rotation of the motor, and the
amplitude could be controlled by an adjustment of the sliding stage of the
X axis. To visualize the wake structure, Hydrogen bubble technique was
deployed. The hydrogen bubbles were generated from the water elec-
trolyzed by the DC current transferred to the cathode connected to the
tungsten wire installed in the upstream of the cylinder. A camcorder was
installed above the setup to record video images of the flow. The visu-
alization test involves both no-vibration and vibration of the cylinder at
Re ¼ 700. To reproduce the free vibration in the water channel, the
oscillation frequency of the cylinder was adjusted to achieve the same Ur
corresponding to the wind tunnel test.
Takai and Sakamoto (2006) examined flow-induced torsional re-
sponses of a rectangular prism in a wind tunnel and the flow structure
was clarified using forced vibration test in a water channel. Morse and
Williamson (2009) made use of forced-vibration experiments to predict
free-vibration response and wake modes obtained by Govardhan and
Williamson (2000). They observed a close agreement between forced-
and free-vibration results, concluding that the forced vibration is
reasonable to predict free-vibration response. Carberry et al. (2004)
observed a strong similarity in the variation of wake mode, force, and
phase for both forced and free vibrations.
3. Flow-induced vibration and wake flow field of the cylinder
3.1. Vibration response
Before conducting an extensive measurement of vibration control, the
flow-induced vibration response and the wake of the test cylinder were
examined. Fig. 4 shows vibration amplitude ratio a/D against Ur, where a
is called harmonic amplitude motion, calculated as the rms of the
pffiffiffi
displacement signal multiplied by 2. The results from Feng (1968) for
S. Kim et al.
Fig. 9. Flow-induced vibration characteristics of the test cylinder at (a) y ¼ 0 - 9D, x ¼ 0–1.1D and (b) y ¼ 1.0D – 1.7D, x ¼ 0–1.1D. In each of the inset graphs, x-
and y-axis represent reduced velocity Ur (¼ U∞/(fn⋅D)) and vibration amplitude ratio a/D, respectively. The solid/dashed lines indicate with/without con-
trol cylinder.
almost the same m*ζ is also included for a comparison purpose. The
maximum amplitude obtained by Feng is somewhat higher than that
obtained presently, because of a smaller m*ζ in Feng (1968). Overall, the
vibration responses measured presently agree well with those from Feng
(1968), given the slight difference in m*ζ. The vibration in the present
measurements starts at Ur ¼ 5.32 and stops at Ur ¼ 7.5, with a maximum
amplitude of 0.14 occurring at Ur ¼ 5.73. Therefore, Ur ¼ 5.32–7.5 can be
defined as the vibration regime. This response curve will be used as the
baseline for vibration suppression in the next section. For the detailed
wake measurements, three reduced velocities Ur ¼ 4.03 before the vi-
bration regime, Ur ¼ 5.73 corresponding to the maximum a/D and
Ur ¼ 9.43 beyond the vibration regime were chosen.
3.2. Wake flow field
Fig. 5 presents the time-mean velocity distribution along the y-
302
Ocean Engineering 151 (2018) 298–307
direction in the wake of the test cylinder at different x for all three
Ur ¼ 4.03, 5.73 and 9.43 considered. The starting point of an arrow in
Fig. 5 is the velocity measurement point, and the length of the arrow is
proportional to the flow velocity. The flow velocities at all x and Ur are
given at the same scale for the sake of a better interpretation. All velocity
distributions (Fig. 5) showed qualitatively similar behavior with the
change in Ur at their respective locations (x ¼ 0D, 1D, 2D, 3D, 4D and 5D).
Apparently, the velocity distribution, being symmetric about the wake
centerline, is largely affected at smaller x and approaches to that (U∞)
without the cylinder when the measurement point moves far from the
wake centerline (y ¼ 0) as well as from the cylinder. In Fig. 5(a–c), the
points at which the velocity of the wake flow becomes U∞ are marked
with “●”. The trace of ‘●’ on each side of the wake is presented by a
straight line, which can be regarded as the outer edge of the wake. The
vortices generated from the cylinder thus propagate within a ‘V’ shape
region enclosed by the two straight lines. The slope of these lines about
S. Kim et al. Ocean Engineering 151 (2018) 298–307

Fig. 9. (continued).

the x-axis is found to be 16.3 , 17.3 and 16.6 for Ur ¼ 4.03, 5.73 and dashed line is thus can be regarded as the trajectory of the outer edges of
9.42, respectively. the vortices. The lateral separation between the two trajectories in the
wake is almost constant for x < 3D but augments for x > 3D. Neverthe-
less, there is a difference in the lateral separation between vibration and
3.3. Spectral analysis and wake structure
no-vibration cases; the lateral separation is smaller for the vibration case
(Ur ¼ 5.73). Interestingly, for Ur ¼ 4.03 and 9.42 both falling in no vi-
The power spectrum of a fluctuating velocity provides the content of
bration regime, the energy at a given y grows with an increase in x up to
energy at different frequencies. It is calculated using a fast Fourier
x ¼ 2D, and declines for a further increase in x. The vortices thus go
transform (FFT) algorithm. In a cylinder wake, as the fluctuation of ve-
through a process of Generation (x < 0D) → Growth (x ¼ 0D – 2D) →
locity largely stems from the vortex shedding, the power spectrum of the
Decay (x > 2D). However, for Ur ¼ 5.73, the process comprising the
velocity fluctuation thus features a peak at the shedding frequency. The
generation, growth, and decay appears very close to the cylinder with the
magnitude of the peak represents the energy at the shedding frequency.
largest energy magnitude at x ¼ 0D. The difference is attributed to the
Fig. 6 shows the distribution of energy at the vortex shedding frequency
vibration generated at Ur ¼ 5.73. The cylinder vibration results in the
obtained from the power spectra of fluctuating velocities measured at
vortex shedding closer to the cylinder as will be shown later. Schaefer
different locations. The thick horizontal lines scale the energy intensities
and Eskinazi (1958) for a stationary cylinder at a low Re (¼ 118)
at different measurement locations. The energy varies with y at each x
observed a declination in the lateral separation of the outer edges of
between 0D to 5D. Symmetrical about the wake centerline, the energy
vortices for x < 2.5 and an increase for x ¼ 2.5–6, which is consistent
distribution at a given x displays a peak (maximum energy) on each side
with the present results except that the declination is not identified
of the wake centerline. A red dashed line traces the peak energy points.
presently. The difference might be due to the coarse resolution in y in the
Schaefer and Eskinazi (1958) for a stationary cylinder showed that, at a
present experiment and/or due to the difference in Re between the
given x, the velocity fluctuation increases with y and reaches a maximum
present and their experiments.
before declining. They confirmed that the position of maximum velocity
It is found that the fluid dynamics at Ur ¼ 4.03 and 9.42 bears
fluctuation corresponds to the outer edge of the vortex core. The red

303
S. Kim et al.
Fig. 10. Classification of flow-induced vibration characteristics of the test
cylinder when the control cylinder is attached behind the test cylinder.
similarity. Visualized flow structures in the same scale for Ur ¼ 4.03 (no-
vibration) and 5.73 (vibration) are presented in Fig. 7. As seen in the
figures, Karman vortex street is observed in the wake irrespective of the
presence/absence of the vibration. While the lateral separation between
the outer edges of the wake is found wider for Ur ¼ 5.73 than for
Ur ¼ 4.03, the lateral separation between the two rows of vortices is
smaller for Ur ¼ 5.73. The observation is consistent with that of Figs. 5
and 6.
4. Suppression of flow-induced vibration of the test cylinder by a
control cylinder
In the previous section, it has been reported that the test cylinder
vibrates at a large amplitude at Ur ¼ 5.73, the amplitude being a/
D ¼ 0.14. An experiment was performed to suppress the flow-induced
vibration of the test cylinder. In order to suppress the flow-induced vi-
bration, another circular cylinder of 0.5D (hereafter, referred to as ‘the
control cylinder’) was placed in the wake of the test cylinder. As shown in
Fig. 8, the control cylinder position, denoted by ‘ ’, is varied as x ¼ 0D,
0.2D, 0.4D, 0.6D, 0.7D, 0.8D, 1.0D and 1.1D and y ¼ 0–1.7D with
Δy ¼ 0.1D. Thus, a total of 136 control cylinder positions are considered.
For each control cylinder position, the vibration response of the test
cylinder is investigated.
The vibration responses of the test cylinder for the different positions
of the control cylinder are shown in Fig. 9. Each inset graph in Fig. 9
illustrates the variation in a/D with Ur. The streamwise and lateral dis-
tances of the control cylinder from the rear stagnation point are shown on
the x- and y-axis, respectively. While the response curve for y ¼ 0 - 0.9D is
presented in Fig. 9(a) and that for the rest (y ¼ 1.0D – 1.7D) is in Fig.
9(b). In the inset graphs, the response of the test cylinder in the absence
304
Ocean Engineering 151 (2018) 298–307
Fig. 11. Contour plot of maximum vibration amplitude due to the vortex
excitation for different positions of the control cylinder.
of the control cylinder, dashed line, is also shown in order to see how the
controlled response (solid line) is different from the non-controlled
response (dashed line). As seen in the figure, the vibration response is
highly sensitive to the control cylinder position. A scrupulous observa-
tion of the responses one by one facilitated a classification of the re-
sponses into four patterns, namely I, II, III, and IV, each pattern having
distinct characteristics. Zoom-in views of representative patterns and
occurrence regimes are shown in Fig. 10.
Pattern I is characterized by the occurrence of vibration at smaller Ur
compared with the non-controlled counterpart (Figs. 9 and 10). The
maximum amplitude Ur shifts toward the left. This pattern appears for
x ¼ 0D - 1.1D and y ¼ (0.9–1.5)D - 1.7D, i.e., when the control cylinder is
away from the wake centerline, being nearly side-by-side. The regime at
which pattern I prevails is marked with ‘ ’ in Figs. 9 and 10.
Pattern II (surrounded by ‘ ’) appearing for x ¼ (0.6–1.0)D – 1.1D
and y ¼ 0D - 0.9D features the vibration occurrence for a longer range of
Ur, mostly extending toward the higher Ur. This pattern is found opposite
to pattern I in the view of the amplitude peak shifting left or right
compared to the non-controlled counterpart. The vibration response in
patterns I and II is self-limited.
Pattern III (enclosed by ‘ ’ for x ¼ 0.4D - 0.8D and y ¼ (0–0.3)D –
0.4D) is connected to the vibration amplitude escalating with an increase
in Ur, resembling a galloping vibration. Since the initiation of the vi-
bration in pattern III takes place almost at the same Ur as that for the non-
controlled case, it can be stated that the vibration is initially generated
due to the vortex excitation and then transmutes into a galloping vibra-
tion with a sufficient increase in Ur.
Pattern IV (bordered with ) represents the absence of vibration in
most of the regime or a very small (<20% of non-controlled case, at
x ¼ 0.5D, y ¼ 0.6D - 1.0D) amplitude vibration. This is the most inter-
esting pattern or regime where vibration is suppressed. Based on the
S. Kim et al. Ocean Engineering 151 (2018) 298–307

occurrence of galloping vibration. In order to distinguish the galloping

vibration regime, the vibration amplitude at Ur ¼ 11 is presented in Fig.
12. Apparently, galloping vibration occurs in pattern III. The amplitude
at the same Ur (¼ 11) is maximum at x ¼ 0.4D, y ¼ (0–0.1)D, decreasing
with increasing y. Now we can connect the regime of pattern IV to the
energy distribution in Fig. 6(b). At x ¼ 0, the energy at the vortex shed-
ding frequency was distributed over y ¼ (0.5–1.3)D (Fig. 6(b)). The range
of y matches with the lateral width of the pattern IV regime.
For free vibration of a cylinder at a small m*ζ (<0.05), vibration
frequency fosc may vary with Ur in the lock-in or galloping regime, fosc
being different from fn (e.g., Assi, 2014). The detuning between the fosc
and fn has also been identified by Mittal & Kumar (2001) for tandem
cylinders at m*ζ ¼ 0.00051. At a large m*ζ > 0.25, one can expect
fosc/fn  1 (Feng, 1968; Govardhan and Williamson, 2000). In the present
experiment with m*ζ ¼ 0.75, the value fosc/fn registers around 1 when the
cylinder vibrates. The distribution of fosc/fn as a function of Ur is thus not
presented.

4.1. Mechanism of vibration suppression

The most interesting phenomenon, the suppression of vibration

(pattern IV), can be narrated here. It is worth investigating the flow
structures associated with the vibration suppression. Flow visualization
test results for patterns IV-A and IV-B are shown in Fig. 13. Interestingly,
for pattern IV-A (Fig. 13(a)), several features can be pointed out, such as
(i) a narrow wake behind the test cylinder and a wide wake behind the
control cylinder, (ii) the interaction of the gap flow with the freestream
side shear layer of the test cylinder, leading to impaired vortex shedding
from the test cylinder, (iii) the narrow wake merging with the wide wake
where only the wide wake survives, and (iv) hence, finally the resultant
wake lacking an alternating Karman vortex street. The suppression of
vibration of the test cylinder thus results from the absence of definite
Fig. 12. Contour plot of vibration amplitude due to galloping at Ur ¼ 11 for
Karman vortex street in the wake. The observation is consistent with the
different positions of the control cylinder.
results obtained by Kim and Alam (2015) for two equal diameter cylin-
ders in side-by-side arrangements. They found no vibration for either
mechanism of the vibration suppression, this regime can further be
cylinder at gap spacings of 0.2D - 0.9D. For this range of spacing, the gap
divided into two subregimes, IV-A (x ¼ 0D - 0.6D, y ¼ 0.8D 1.4D) and
flow biased toward a side acts as a base bleed, postpones interaction and
IV-B (x ¼ 0D – 1.0D, y ¼ 0.3–0.5D - 0.7D), see Figs. 9 and 10.
rolling of the shear layers, and hence weakens vortex sheddings. Vibra-
The regimes of the four patterns can also be clarified from the contour
tion was thus not generated in either of the two cylinders. On the other
plot of the maximum vibration amplitude due to the vortex excitation as
hand, in pattern IV-B, the upper shear layer separating from the test
shown in Fig. 11. Interestingly, the maximum amplitude also distin-
cylinder reattaches on the control cylinder and swerves toward the other
guishes the regimes of the vibration classifications. While the vibration
side shear layer of the test cylinder (Fig. 13(b)). Eventually, a very wide
amplitude is large in the most of the regime of pattern II, it is suppressed
wake persists behind the two combined cylinders. The wake again lacks
in the entire regime of pattern IV. An intermediate amplitude prevails in
alternating Karman vortices. The physics behind the suppression in this
the regimes of patterns I and III. They are however different by the
pattern is that the control cylinder position is on the upper shear layer or

Fig. 13. Flow visualization test result and corresponding

schematic diagram for pattern IV where the vibration of the
test cylinder is suppressed: (a) pattern IV-A and (b) pattern
IV-B. Re ¼ 700.

305
S. Kim et al.
vortex trajectory of the non-control cylinder (see Fig. 6). Therefore, the
control cylinder distorts or disorganizes the upper shear layer of the test
cylinder (Alam et al., 2003). Here the interaction between the shear layer
and control cylinder succeeds in the suppression of the alternating Kar-
man vortices from the cylinders. On the other hand, in pattern IV-A the
interaction between the gap flow and freestream side shear layer leads to
the suppression of alternating Karman vortices.
Bokaian and Geoola (1984) observed galloping vibration of a flexible
cylinder neighbored by another rigid cylinder of the same diameter with
spacing x ¼ 0.59D – 0.75D, y ¼ 0-0.5D. Interestingly, the regime III to
some extent agrees with ranges of x and y for identical cylinders, given
the diameters of the two cylinders are different presently. The gap flow
switch has been suggested as an excitation mechanism for a flexible
cylinder placed behind another of identical diameter (Zdravkovich,
1988; Ruscheweyh and Dielen, 1992; Dielen and Ruscheweyh, 1995).
The flexible cylinder oscillation was not excited when it was inline with
the upstream cylinder, but excited only when it was with an initial
displacement greater than 0.2D (Zdravkovich, 1974) or with incoming
flow incidence angle being larger than about 10 . The oscillations gen-
eration was explained by the existence of two flow modes during the
vibration; one appearing for a longer time was with an intense flow
through the gap, generating a large inward lift force, and the other pre-
vailing for a shorter time was with no gap flow, leading to very small or
negligible lift force. This hysteresis in the flow maintained the large
amplitude oscillation. It is believed that the gap flow switch is respon-
sible for the galloping occurring in regime II.
Regime IV is identified as suppressed vibration regime, while regime
II is the galloping vibration regime. For a fixed cylinder, it has been
demonstrated that placing a control cylinder around a cylinder can
suppress time-mean drag and fluctuating forces (e.g., Igarashi and Tsut-
sui, 1989; Strykowski and Sreenivasan, 1990; Sakamoto et al., 1991). The
present research warns that if the cylinder elasticity is considered, the
force control technique might be useful for certain regimes but not for
some other regimes.
4. Conclusions
Firstly, the wake of a cylinder subjected to vortex excited vibration is
investigated in terms of time-mean velocity and energy intensity that was
obtained through the power spectral analysis of fluctuating streamwise
velocity. Secondly, the effectiveness of a control cylinder placed at x ¼ 0D
1.1D and y ¼ 0D- 1.7D is examined to suppress the flow-induced vi-
bration of the test cylinder for different positions of the control cylinder.
Finally, a forced vibration experiment in a water tunnel is conducted to
visualize flow structures, reproducing the free vibration results.
When a cylinder undergoes a vortex excited vibration, the process of
the ‘Generation → Growth → Disappearance’ of vortices appears closer to
the cylinder compared to that for no-vibration. The efflux angle
embraced the process of Generation → Growth → Disappearance is
within the range of 16 –17 and appears to be constant regardless of the
presence/absence of vibration or changes in Ur. The vortex trajectory
identified from the power spectral energy in the wake is substantiated by
flow visualization test. The lateral separation between the two vortex
rows in a wake does not change for x < 3D but widens for x > 3D.
Compared to that for a non-vibrating cylinder, the lateral separation for a
vibrating cylinder undergoing vortex excitation is smaller.
A control cylinder placed in the wake of the test cylinder alters the
vibration response of the test cylinder into four patterns, namely I, II, III,
and IV depending on the control cylinder position. The regimes of four
patterns are also clarified by presenting contour plots of the maximum
vibration amplitude due to the vortex excitation. Pattern I is charac-
terized by the occurrence of vibration at a smaller Ur compared to the
non-controlled counterpart. Unlike pattern I, the pattern II exemplifies
the vibration occurrence for a longer range of Ur, mostly extending
toward the higher Ur. Pattern III features the vibration amplitude
306
Ocean Engineering 151 (2018) 298–307
increasing with an increase in Ur, resembling a galloping vibration. On
the other hand, pattern IV represents the suppression of vibration. The
suppression of vibration results from the suppression of Karman street in
the wake. Based on how the Karman street is suppressed, this regime
can further be divided into two subregimes, IV-A and IV-B. While the
interaction between the gap flow and freestream side shear layer of the
test cylinder succeeds in the suppression of the Karman street in pattern
IV-A, that between the gap shear layer and the control cylinder gives
rise to the suppression of the Karman street in pattern IV-B.
Acknowledgement
Kim wishes to acknowledge the support from 2016 Research Grant
from Kangwon National University (No. 620160129). Alam wishes to
acknowledge the support from National Natural Science Foundation of
China through Grants 11672096 and from Research Grant Council of
Shenzhen Government through grant JCYJ20160531191442288.
References
Alam, M.M., Moriya, M., Takai, K., Sakamoto, H., 2003. Fluctuating fluid forces acting on
two circular cylinders in a tandem arrangement at a subcritical Reynolds number.
J. Wind Eng. Ind. Aerod. 91 (1–2), 139–154.
Alam, M.M., Kim, S., 2009. Free vibration of two identical circular cylinders in staggered
arrangement. Fluid Dynam. Res. 41 (3), 035507.
Assi, G.R., 2014. Wake-induced vibration of tandem cylinders of different diameters.
J. Fluid Struct. 50, 329–339.
Blevins, R.D., 1990. Flow-induced Vibration, second ed. Van Nostrand Reinhold, New
York, USA.
Bokaian, A., Geoola, F., 1984. Wake-induced galloping of two interfering circular
cylinders. J. Fluid Mech. 146, 383–415.
Carberry, J., Govardhan, R., Sheridan, J., Rockwell, D., Williamson, C.H.K., 2004. Wake
states and response branches of forced and freely oscillating cylinders. Eur. J. Mech. B
23, 89–97.
Chen, S.S., 1987. Flow-induced Vibration of Circular Cylindrical Structures. Hemisphere
Publishing, Washington.
Chen, W.L., Xin, D.B., Xu, F., Li, H., Ou, J.P., Hu, H., 2013. Suppression of vortex-induced
vibration of a circular cylinder using suction-based flow control. J. Fluid Struct. 42,
25–39.
Dielen, B., Ruscheweyh, H., 1995. Mechanism of interference galloping of two identical
circular cylinders in cross flow. J. Wind Eng. Ind. Aerod. 54, 289–300.
Du, L., Sun, X., 2015. Suppression of vortex-induced vibration using the rotary oscillation
of a cylinder. Phys. Fluids 27, 023603.
Feng, C.C., 1968. The Measurements of Vortex-induced Effects in Flow Past a Stationary
and Oscillating Circular and D-section Cylinders. Master’s thesis. University of British
Columbia, Vancouver, Canada.
Govardhan, R., Williamson, C.H.K., 2000. Modes of vortex formation and frequency
response of a freely vibrating cylinder. J. Fluid Mech. 420, 85–130.
Igarashi, T., Terachi, N., 2002. Drag reduction of flat plate normal to airstream by flow
control using a rod. J. Wind Eng. Ind. Aerod. 90, 359–376.
Igarashi, T., Tsutsui, T., 1989. Flow control around a circular cylinder by a new method.
Trans. JWRI 55, 701–713.
Kim, S., Alam, M.M., Sakamoto, H., Zhou, Y., 2009a. Flow-induced vibrations of two
circular cylinders in tandem arrangement. Part 1: characteristics of vibration. J. Wind
Eng. Ind. Aerod. 97 (5–6), 304–311.
Kim, S., Alam, M.M., Sakamoto, H., Zhou, Y., 2009b. Flow-induced vibrations of two
circular cylinders in tandem arrangement. Part 2: suppression of vibration. J. Wind
Eng. Ind. Aerod. 97 (5–6), 312–319.
Kim, S., Alam, M.M., 2015. Characteristics and suppression of flow-induced vibrations of
two side-by-side circular cylinders. J. Fluid Struct. 54 (4), 629–642.
Lam, K.M., To, A.P., 2003. Interference effect of an upstream larger cylinder on the lock-
in vibration of a flexibly mounted circular cylinder. J. Fluid Struct. 17 (8),
1059–1078.
Lu, L., Liu, M., Teng, B., Cui, Z., Tang, G., Zhao, M., Cheng, L., 2014. Numerical
investigation of fluid flow past circular cylinder with multiple control rods at low
Reynolds number. J. Fluid Struct. 48, 235–259.
Mittal, S., Kumar, V., 2001. Flow-induced oscillations of two cylinders in tandem and
staggered arrangements. J. Fluid Struct. 15, 717–736.
Morse, T.L., Williamson, C.H.K., 2009. Prediction of vortex-induced vibration response by
employing controlled motion. J. Fluid Mech. 634, 5–39.
Paıdoussis, M.P., 1993. Some Curiosity-driven Research in Fluid-structure Interactions and
its Current Applications. Calvin Rice Lecture. ASME Trans. J. Press. Vessel 115, 2-14.
Ruscheweyh, H., Dielen, B., 1992. Interference galloping-investigations concerning the
phase lag of the flow switching. J. Wind Eng. Ind. Aerod. 43, 2047–2056.
Sakamoto, H., Tan, K., Haniu, H., 1991. An optimum suppression of fluid forces by
controlling a shear layer separated from a square prism. J. Fluid Eng. 113, 183–189.
Schaefer, J.W., Eskinazi, S., 1958. An analysis of the vortex street generated in a viscous
fluid. J. Fluid Mech. 6, 241–260.
S. Kim et al.
Strykowski, P.J., Sreenivasan, K.R., 1990. On the formation and suppression of vortex
shedding at low Reynolds numbers. J. Fluid Mech. 218, 71–107.
Takai, K., Sakamoto, H., 2006. Response characteristics and suppression of torsional
vibration of rectangular prisms with various width-to-depth ratios. Wind Struct. 9,
1–22.
Tu, J., Zhou, D., Bao, Y., Fang, C., Zhang, K., Li, C., Han, Z., 2014. Flow-induced vibration
on a circular cylinder in planar shear flow. Comput. Fluids 105, 138–154.
Wu, H., Sun, D.P., Lu, L., Teng, B., Tang, G.Q., Song, J.N., 2012. Experimental
investigation on the suppression of vortex-induced vibration of long flexible riser by
multiple control rods. J. Fluid Struct. 30, 115–132.
Wu, J., Shu, C., Zhao, N., 2014. Numerical investigation of vortex-induced vibration of a
circular cylinder with a hinged flat plate. Phys. Fluids 26, 063601.
307
Ocean Engineering 151 (2018) 298–307
Xu, F., Chen, W.L., Xiao, Y.Q., Li, H., Ou, J.P., 2014. Numerical study on the suppression
of the vortex-induced vibration of an elastically mounted cylinder by a traveling
wave wall. J. Fluid Struct. 44, 145–165.
Zdravkovich, M.M., 1974. Flow-induced vibration of two cylinders in tandem
arrangements, and their suppression. In: Proceedings of the International Symposium
on Flow Induced Structural Vibrations, Karlsruhe 1972. Springer, pp. 631–639.
Zdravkovich, M.M., 1988. Review of interference-induced oscillations in flow past two
parallel circular cylinders in various arrangements. J. Wind Eng. Ind. Aerod. 28,
183–199.
Zhao, M., Cheng, L., Teng, B., Dong, G., 2007. Hydrodynamic forces on dual cylinders of
different diameters in steady currents. J. Fluid Struct. 23, 59–83.