Journal 

of Fluid Mechanics
http://journals.cambridge.org/FLM

Additional services for Journal of Fluid Mechanics:

Email alerts: Click here
Subscriptions: Click here
Commercial reprints: Click here
Terms of use : Click here

The timing of vortex shedding in a cylinder wake imposed 
by periodic inflow perturbations
E. KONSTANTINIDIS, S. BALABANI and M. YIANNESKIS

Journal of Fluid Mechanics / Volume 543 / November 2005, pp 45 ­ 55
DOI: 10.1017/S0022112005006786, Published online: 07 November 2005

Link to this article: http://journals.cambridge.org/abstract_S0022112005006786

How to cite this article:
E. KONSTANTINIDIS, S. BALABANI and M. YIANNESKIS (2005). The timing of vortex shedding in 
a cylinder wake imposed by periodic inflow perturbations. Journal of Fluid Mechanics,543, pp 
45­55 doi:10.1017/S0022112005006786

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/FLM, IP address: 128.42.202.150 on 19 Aug 2012
J. Fluid Mech. (2005), vol. 543, pp. 45–55.

c 2005 Cambridge University Press 45
doi:10.1017/S0022112005006786 Printed in the United Kingdom

The timing of vortex shedding in a cylinder wake

imposed by periodic inflow perturbations
By E. K O N S T A N T I N I D I S†, S. B A L A B A N I
A N D M. Y I A N N E S K I S
Experimental and Computational Laboratory for the Analysis of Turbulence, Department
of Mechanical Engineering, King’s College London, Strand, London WC2R 2LS, UK

(Received 11 May 2005 and in revised form 17 August 2005)

The interaction of vortex shedding from a circular cylinder with an inflow which has
low-amplitude periodic velocity oscillations (perturbations) superimposed upon it, was
investigated experimentally by means of particle image velocimetry. The experiments
were made at three perturbation frequencies across the lock-on range in which the
vortex shedding frequency is synchronized with the subharmonic of the imposed
frequency. The basic wake pattern in this range is antisymmetric vortex shedding, i.e.
the familiar 2S mode. The timing of vortex shedding is defined with respect to the
cross-flow oscillation of the wake which is found to play a critical role. Quantitative
analysis of the phase-referenced patterns of vorticity distribution in the wake shows
that a vortex is actually shed from the cylinder when the cross-flow oscillation of the
wake is strongest, marked by a sudden drop in the computed vortex strength. At the
middle of the lock-on range, shedding occurs near the minimum inflow velocity in
the cycle or, equivalently, during the forward stroke of a cylinder oscillating in-line
with the flow. It is argued that the imposed timing of vortex shedding relative to
the cylinder motion induces a negative excitation from the fluid, which might explain
why the in-line response of a freely vibrating cylinder exhibits two positive excitation
regions separated by the lock-on region found in forced oscillations.

1. Introduction
Over the past few decades the study of so-called ‘bluff’ body wakes, and in particular
that of the circular cylinder, has been driven by an urge to understand vortex-induced
vibration, which is of concern in the design of many engineering structures. The
fundamental research in the field has proceeded using two complementary approaches:
some investigators consider the problem where a rigid cylinder is forced to oscillate
in a predefined direction simulating the vortex-induced motion, while others consider
the case where the cylinder is flexibly mounted at the ends and free to vibrate with a
single and, more recently, with two degrees of freedom. Both transverse to the flow
and streamwise oscillations have been considered in numerous studies but the former
case has attracted most of the interest. Recently, Carberry et al. (2004) explored the
relationship between free and forced transverse oscillations of a rigid cylinder. A
strong similarity was observed between the response branches of the freely oscillating
cylinder and the wake modes of the forced cylinder in terms of the patterns of vorticity

† Present address: Department of Engineering and Management of Energy Resources, University

of Western Macedonia, Kozani 50100, Greece.
46 E. Konstantinidis, S. Balabani and M. Yianneskis

U(φ) = Um + ∆U sin(φ)

Figure 1. Schematic description of the flow configuration.

and the variation of the phase of the total and vortex lift forces with respect to the
cylinder motion. Central to their analysis is the decomposition of the total lift force
into two components: the apparent mass force due to acceleration of the cylinder and
the vortex lift force due to the vorticity distribution in the wake. It turns out that the
phase of the vortex lift force determines the phase of the total force which, in turn,
determines whether the energy transfer is from the fluid to the cylinder or vice versa.
In contrast, the understanding gained from studies of cylinders oscillating in-line
with the flow, either freely or forcibly, has remained largely unexplored although
a considerable amount of force data exists in the published literature. What
appears to be missing is the phase-referenced patterns of vorticity in the wake
across the ‘lock-on’ range in which vortex shedding synchronizes with the cylinder
oscillation. Only qualitative information from flow visualization studies can be found
(Griffin & Ramberg 1976; Ongoren & Rockwell 1988). In this paper, we report
on particle image velocimetry measurements obtained in the forced wake of a
circular cylinder in an incident flow with low-amplitude periodic velocity oscillations
(perturbations) superimposed upon it. This case is equivalent to that of a cylinder
forced to oscillate in-line with a steady incident flow as long as the perturbation
wavelength is large compared to the cylinder diameter, a condition which is satisfied
in the present experiments. An attempt is made to identify the underlying mechanics
of wake excitation across the synchronization range in terms of the patterns and
the timing of the vortex shedding process. In the light of the present findings, some
aspects of the in-line free-vibration response of a circular cylinder are discussed in
the paper. It is hoped that this study will complement other contributions dealing
with bluff-body wakes, in particular those concerned with bodies in unsteady flows,
and will improve understanding of vortex-induced vibrations for this class of flows.

2. Experimental system and conditions

A brief description of the experimental facility and measurement techniques is
given below. For further details, the reader may refer to previous publications by
the authors (Konstantinidis, Balabani & Yianneskis 2003, 2004, 2005). A schematic
description of the flow configuration is shown in figure 1. The experiments were
carried out in a recirculating-type water tunnel having a 72 mm-square cross-section.
A circular rod of diameter D = 7.2 mm was centrally mounted in the test section.
The incident flow was made uniform by a contraction (area ratio 9:1) which preceded
the test section and had a background turbulence level of 3.3 %. Periodic velocity
oscillations were superimposed onto the mean incident flow via the action of a
rotating valve. Flow conditions were monitored by laser-Doppler velocimetry (LDV)
measurements upstream of the test cylinder. The inflow velocity can be approximated
 The timing of vortex shedding in a perturbed flow 47

Um (m s−1 ) fe (Hz)
U (m s−1 ) Re fe /fo A/D

0.301 15.4 0.019 2170 1.74 0.053
0.299 17.1 0.018 2160 2.00 0.044
0.298 19.5 0.018 2150 2.20 0.042
Table 1. Flow parameters employed in the experiments.

as a function of the phase-angle φ by U (φ) = Um +
U sin(φ); Um is the mean inflow

velocity and
U is the amplitude of velocity oscillation. The phase-angle φ relates
to ‘virtual’ time t via φ = 2πfe t with fe being the frequency of the imposed inflow
perturbations. The independent parameters employed to describe the problem are:
the Reynolds number Re = Um D/ν (where ν is the kinematic viscosity of water),
the frequency ratio, fe /fo , and the peak-to-peak amplitude of the equivalent cylinder
oscillation A/D = 2
U/2πfe D. Herein, fo is the natural vortex shedding frequency
in unforced flow which was computed from a constant-Strouhal-number relationship
S = fo D/Um = 0.215. The results reported in this paper were obtained for the condi-
tions shown in table 1, i.e. at three perturbation frequencies with the mean inflow
velocity and the amplitude of the velocity oscillations maintained nearly constant. The
equivalent amplitude of cylinder oscillation A/D varies slightly with fe /fo . These con-
ditions correspond to the primary lock-on (or synchronization) range for in-line cylin-
der or equivalent flow oscillations, in which the vortex shedding frequency locks on to
the subharmonic of the imposed perturbation frequency (Konstantinidis et al. 2003).
The instantaneous velocity field of the near wake was examined using digital
particle image velocimetry (DPIV). For this purpose the flow was seeded with 10 µm
silvered micro-spheres and the wake was illuminated by a laser sheet perpendicular
to the cylinder axis at midspan. Pairs of images were captured on a digital camera
with a resolution of 1280 × 1024 pixels. The image pairs were analysed by employing
adaptive cross-correlation on a 32×32-pixel interrogation window and a 50 % overlap
ratio, giving a measurement-grid resolution of approximately 0.1D. All images taken
were phase-referenced with respect to the phase of the inflow oscillation. Phase
information was provided by a high-precision optical encoder attached to the shaft of
the perturbation-generating rotating valve. This enabled segregation of repetitive from
non-repetitive incoherent flow structures by averaging at constant phase. The phase-
averaged vorticity fields were calculated from thirty or more instantaneous fields.

3. Results
The instantaneous vorticity fields in the wake of the cylinder revealed two different
vortex patterns which correspond loosely to the known 2S and 2P modes observed in
transverse cylinder oscillations where two single and two pairs of vortices, respectively,
are shed during each cycle of oscillation (Williamson & Roshko 1988). These patterns
have also been observed in flow visualizations of a cylinder wake during in-line
oscillations at Re = 190 by Griffin & Ramberg (1976). In the present study, the two
patterns coexist at fe /fo = 1.74 whereas only the 2S mode was observed at fe /fo >
2. Examination of the instantaneous vorticity fields indicated that the interchange
between the two modes was random and did not involve a change in the timing of
vortex shedding with respect to the inflow oscillation. Previous analysis by means
of proper orthogonal decomposition has shown that the most energetic wake mode
has a 2S structure while the 2P pattern appears as a secondary effect (Konstantinidis
48 E. Konstantinidis, S. Balabani and M. Yianneskis
et al. 2004). In the following, the results only for the 2S mode are shown in order
to facilitate direct comparisons at different frequencies. However, this does not limit
the generality of the results since the wake characteristics discussed below were quite
similar for both patterns.
The interaction between the inflow perturbations and the vortex shedding process
in the wake is examined and an attempt is made to identify critical events which
determine the timing of vortex shedding, i.e. the phase at which a vortex pinches
off from the cylinder with respect to the inflow velocity oscillation. Descriptions of
the mechanics of vortex formation and shedding in the unforced cylinder wake at
subcritical Reynolds numbers can be found in Gerrard (1966) who has demonstrated
the critical role of the flow oscillations across the wake axis and within the vortex
formation region. This characteristic will prove useful also in the present context.
An obstacle in the study of wake flows is that it is difficult to find a consistent way
to compute the circulation of the vortices, or vortex strength, especially when these
are attached to the cylinder. To circumvent this problem, the following computational
procedure was employed. Initially, individual vortices were identified in the phase-
averaged vorticity fields ω by the peaks in the absolute vorticity magnitude, denoted
ωp . The circulation around the peak locations was then computed by a closed-
pathline integral for increasing radii and the maximum circulation was found; rmax is
the corresponding radius. Generally, the behaviour observed was that the circulation
initially increases, reaches a plateau and then drops off again with increasing radii.
Subsequently, the vortex strength Γ and centroid {xc , yc } were computed using surface
integration (quadrature) of the vorticity distribution within the area enclosed by rmax
as described, for example, in Cantwell & Coles (1983). The integration was performed
only for values such that ω > 0.15ωp in order to minimize the influence of noisy
data. The above procedure enabled computation of the vortex strength in a well-
defined and consistent way. Furthermore, the results provide physical insight into the
shedding process as it is described below.
Figure 2 shows the vorticity distribution in the near wake at five phases during
an imposed oscillation cycle (corresponding to half a vortex shedding cycle). In this
figure, attention is directed to the clockwise-rotating (negative) vortex enclosed by
the velocity vectors in subplots (a–e) and the corresponding variations in its strength
along the wake, Γ , the imposed velocity oscillation, U (φ), and the velocity oscillation
across the wake axis at the location in which its root-mean-square value is maximum,
V (φ), in subplots (f–h), respectively. The tails of the velocity vectors shown delineate
a circle which corresponds to the circular path of integration where the maximum
circulation around the vortex was computed. However, the vortex strength estimates
shown in figure 2(f ) were computed from a quadrature as explained above. The
following events occur in label sequence: (a) The vortex, fed with circulation from
the cylinder, attains its maximum strength. At this phase, the inflow decelerates
(equivalent to a cylinder moving forward in a steady inflow). (b) V (φ) is about to
become strongest as the lower-side shear layer is entrained across the wake axis and
cuts off the supply of vorticity when the inflow velocity attains its minimum value
(cylinder moving forward at maximum speed). (c) The shed vortex picks up strength
from the shear layer feeding in vorticity as the ‘potential’ flow outside the wake
accelerates (cylinder at forwardmost position). (d ) As the inflow velocity approaches
its maximum value the vortex strength reaches a secondary maximum (cylinder in
backward stroke). (e) The inflow starts to decelerate and the shed vortex moves
downstream with gradually diminishing strength (cylinder towards its backwardmost
position).
 The timing of vortex shedding in a perturbed flow 49

–5 –4 –3 –2 –1 0 1 2 3 4

(a) (b)
1 1

y y
0
D D 0

–1 –1

0 1 2 3 4 0 1 2 3 4
x/D x/D

(c) (d)
1 1

y y
0
D D 0

–1 –1

0 1 2 3 4 0 1 2 3 4
x/D x/D
3.5
(e) (f)
1
(a)
3.0 (d)(e)
y Γ (c)
0 (b)
D UmD
2.5
–1
2.0
0 1 2 3 4 1 2 3 4 5 6
x/D xc /D
1.1 (c)
(g) (a) (b) (h)
(d) 1
(e) (d)

U(φ) (c) V(φ) 0

1.0 (a) (e)
Um Um
(b)
–1
0.9
0 180 360 540 720 0 180 360 540 720
φ φ

Figure 2. (a–e) Phase-averaged vorticity distributions in the wake showing the vortex shedding
process at fe /fo = 2. Contour colour coding shown on top. Velocity vectors delineate a circular
area for which the circulation around the peak vorticity is maximised and within which the
vortex strength was computed by a surface quadrature. (f ) Vortex strength along the wake.
(g) Inflow velocity oscillation upstream of the cylinder. (h) Cross-flow oscillation of the wake
at the location of maximum intensity {x/D, y/D} = {1.2, 0}.

A feature that stands out in figure 2(f ) is the sudden drop in vortex strength.
Whether this feature is real or artificial is debatable. A possible physical mechanism
is the cross-annihilation of oppositely signed vorticity in the near wake due the
flapping motion of the shear layer. The negative vortex appears to be attached to the
50 E. Konstantinidis, S. Balabani and M. Yianneskis

(b) (c)
1 2.2

2.1
V(φ)
0
Um
2.0
fe
–1 fo
1.9

1.1
1.8
U(φ)
Um 1.0
(a) 1.7
0.9
0 180 360 540 720 0 60 120
φ ∆φ

Figure 3. (a) Inflow velocity oscillation; (b) cross-flow oscillation of the wake;
(c) phase-difference between U (φ) and V (φ). – – fe /fo = 1.74; — fe /fo = 2.00; – . – fe /fo = 2.20.

cylinder by a strand of vorticity extending all the way to its generator in figure 2(a–d ).
A careful look at the velocity vectors in figure 2(b) shows that the vectors within the
strand of vorticity connection are nearly normal to the tangent of the circle delineated
by the vectors at this particular phase. It might be argued that the vortex is no longer
fed with circulation from the cylinder and is, thus, shed. The sharp decrease in vortex
strength coincides with the occurrence of the maximum cross-flow oscillation, i.e.
V (φ) = Vmax , which is caused by the entrainment of the shear layer across the wake
axis cutting off the supply of vorticity and eventually determining the timing of
vortex shedding. The same feature was observed at other perturbation frequencies
for which the wake exhibits a slightly different morphology as is shown further
below.
According to the analysis presented above, it is deemed appropriate to use the
relative phase between U (φ) and V (φ) in order to examine the variation in the timing
of vortex shedding as fe /fo is varied across the synchronization range (2S mode).
Figure 3 shows the response of the V (φ) oscillation across the wake axis to the imposed
inflow perturbation U (φ) at the three different frequencies. All lines shown are sine
fits to the actual velocity data; U (φ) = Um +
U sin(φ) curves are based on phase-
averaged LDV data obtained upstream of the test cylinder; V (φ) = Vmax sin(φ/2+
φ)
curves are based on phase-averaged DPIV data obtained in the wake and correspond
to the location of maximum cross-flow oscillation along the wake axis; x/D = 1.2, 1.2
and 1.3 at fe /fo = 1.74, 2.00 and 2.20, respectively. It should be noted that the sign
of V (φ) determines whether a vortex is shed from the upper or the lower side once
per oscillation and that the flow field is a mirror image with respect to the wake axis
for a phase-shift of 360◦ . As fe /fo is varied there is a gradual change in the phase
of V (φ) with respect to the phase of U (φ). At fe /fo = 1.74, V (φ) leads U (φ) by

φ = −28◦ whereas the opposite occurs at fe /fo = 2.20 (
φ = 104◦ ). As discussed
above, the timing of vortex shedding occurs near the phase of V (φ) = Vmax . This
implies that a vortex is shed in each forward stroke of the oscillating cylinder within
the synchronization range which corresponds to phase-angles between φ = 180◦
and 360◦ . The implications of this assertion for the in-line free vibration response
of a flexibly mounted circular cylinder will be discussed in § 4. When the imposed
wake frequency is lower than the natural one, the period of vortex shedding is
 The timing of vortex shedding in a perturbed flow 51
increased compared to the natural. The wake responds to this effect by shifting
the timing of vortex shedding to occur earlier in the cycle and at the beginning
of the forward stroke, e.g. Vmax = V (244◦ ) at fe /fo = 1.74. This is like trying to
adhere to a universal formation process (Jeon & Gharib 2004). The inverse occurs
at frequencies greater than the natural wake frequency; the wake adjusts the timing
of vortex shedding to occur later in the cycle and towards the end of the forward
stroke. In fact, Vmax = V (376◦ ) at fe /fo = 2.2 which corresponds to the beginning of
the backward stroke (see § 4). At the middle of the lock-on range, i.e. fe /fo = 2, the
perturbations interact with the natural vortex formation process which brings about
a remarkable intensification of the wake fluctuations and the fluid forces exerted on
the cylinder as has been consistently observed in previous related studies (Tanida,
Okajima & Watanabe 1973; Barbi et al. 1986, Armstrong, Barnes & Grant 1987;
Konstantinidis et al. 2003).
Figure 4 shows the phase-referenced vorticity distributions in the wake at the
three perturbation frequencies employed. The phase shown in each case was selected
to give the best possible match among the different frequencies based on the
φ
estimates found above. A close correspondence of the wake patterns can be observed.
The vorticity contours show a positive (rotating counter-clockwise) vortex on the
lower side at the timing of shedding, even though the vortex appears connected
to the cylinder by a vorticity strand. Upon examining the vorticity distributions in
figure 4, the fact that the longitudinal spacing between the vortices varies inversely
proportionally with the frequency ratio fe /fo should be taken into account (Griffin &
Ramberg 1976; Konstantinidis et al. 2003). As a result of this effect, the wake pattern
appears expanded and contracted for frequencies below and above, respectively, the
natural wake frequency. This effect should not be confused with the variation of the
vortex formation region. If the formation length lf is defined as the location along
the wake axis where the streamwise velocity fluctuation becomes strongest, then
lf /D = 1.0, 1.1 and 1.3 at fe /fo = 1.74, 2.00 and 2.20, respectively (cf. to the value of
2.3 in unforced flow). Previous detailed LDV measurements along the wake axis have
clearly shown that the formation length is minimized in the middle of the lock-on
range (Konstantinidis et al. 2003). It should be noted that the perturbation frequency
has a remarkable effect on the morphology of the shed vortex. Referring to the
positive vortex on the lower side of the cylinder, a gradual modification from a rather
circular shape at fe /fo = 1.74 to an elliptical one with its minor axis nearly aligned
with the flow at fe /fo = 2.20 is observed. This feature demonstrates the complexity of
the fluid mechanics and of the interaction between the natural wake instability and
the imposed perturbation. Even these subtle differences in the vorticity distribution
can have a pronounced effect on the fluid forces imparted to the cylinder as shown
by related force measurements (Tanida et al. 1973; Nishihara, Kaneko & Watanabe
2005).

4. Discussion
The response of a flexibly mounted circular cylinder to vortex-induced vibration
depends on a number of flow and structural parameters, see e.g. Bearman (1984)
for a review on the subject. Given the structural parameters, the amplitude of
vibration can be described in terms of a main parameter: the reduced velocity
U ∗ = Um /fn D where fn is the natural vibration frequency of the structure. On the
other hand, in forced oscillation studies the frequency and amplitude of oscillation,
usually encountered in terms of the dimensionless parameters fe /fo and A/D, can
52 E. Konstantinidis, S. Balabani and M. Yianneskis
1.1 (a)
2
U(φ)
1.0
Um 1

y
__
0.9 0
D
1
–1
V(φ)
0
Um –2
–1
0 180 360 540 720 0 1 2 3 4 5 6 7
φ x/ D
1.1 (b)
2
U(φ)
1.0
Um 1

0.9 y
__ 0
D
1
–1
V(φ)
0
Um –2
–1
0 180 360 540 720 0 1 2 3 4 5 6 7
φ x/D

1.1 (c)
2
U(φ)
1.0
Um 1

0.9 y
__ 0
D
1
–1
V(φ)
0
Um –2
–1
0 180 360 540 720 0 1 2 3 4 5 6 7
φ x/D

Figure 4. Phase-averaged vorticity distribution in the forced wake at the phase that a positive
vortex is shed from the lower side of the cylinder together with the corresponding points
(red symbols) in the inflow and cross-flow oscillation. (a) fe /fo = 1.74; (b) fe /fo = 2.00;
(c) fe /fo = 2.20. Contour colour coding as in figure 2.

be varied independently. Correlating free and forced cylinder studies may be viewed
as a transformation of the independent variables fe ↔ fn . The structural frequency
fn acts as an excitation frequency in free vibrations. This transformation appears
straightforward but one needs to be careful because the frequency at which the
cylinder responds can be different from fn , depending on the combined flow and
structural parameters (Williamson & Govardhan 2004). In practice, however, the
cylinder vibration and the wake synchronize at a frequency sufficiently close to fn .
 The timing of vortex shedding in a perturbed flow 53
0.20
Lock-on
region
0.15

A 0.10
D

st
0.05 1 response
2nd response
branch
branch

0
2.5 2.0 1.5
fe/fo

2.0 2.5 3.0 3.5

U*

Figure 5. The lock-on region found in forced cylinder oscillations in-line with the flow
juxtaposed with the response branches of a flexibly mounted cylinder free to vibrate in-line
with the flow. Lock-on envelope produced from the data in Konstantinidis et al. (2003).
Positive excitation envelope reproduced from published data in Okajima et al. (2004). Bull’s
eye symbols indicate the present experimental data.

A flexibly mounted cylinder restrained to vibrate only in-line with the flow direction
exhibits two positive excitation regions, or response branches, as U ∗ is varied.
Paradoxically, these two branches lie on either side of the lock-on region found
in forced oscillation studies, which occurs around U ∗ ≈ 1/2S ≈ 2.5. This is illustrated
in figure 5 where the lock-on region for forced oscillations and the positive excitation
regions for free vibrations in-line with the flow are juxtaposed. The transformation
fe ↔ fn has been employed so that U ∗ = S −1 (fe /fo )−1 . When U ∗ < 1/2S, symmetric
vortex formation on both sides of the cylinder occurs while for U ∗ > 1/2S the
antisymmetric 2S vortex pattern is observed (Naudascher 1987; Okajima et al. 2004).
It should be noted that the response depends on the combined flow and structural
parameters. The positive excitation regions shown are after the recent measurements
of Okajima et al. with the lowest combined mass damping. In this condition the highest
amplitude response was observed so that the hatched area in figure 5 encompasses
all the measured response curves.
The behaviour observed in figure 5 implies that over the lock-on range in which
the forces acting on a forcibly oscillating cylinder are magnified (Tanida et al. 1973;
Barbi et al. 1986; Nishihara et al. 2005), the in-line response of a freely vibrating
cylinder is negligible. This seemingly paradoxical behaviour may be explained by
extending the concept of the vortex lift force to the drag direction (Carberry et al.
2004). The fluctuating drag force may be regarded as the sum of two components:
an apparent mass force due to the displaced fluid and a vortex drag force due to the
vorticity distribution in the wake. Since the apparent mass force is always in-phase
with the relative cylinder motion, the phase of the total drag force is determined by
the phase of the vortex force, i.e. by the vorticity distribution in the wake relative
to the cylinder motion. The lack of cylinder excitation inside the lock-on range, i.e.
fe /fo ≈ 2, implies that the vortex drag force opposes the cylinder motion (damping
force). Interestingly, measurements of the phase of the total drag force predict this
54 E. Konstantinidis, S. Balabani and M. Yianneskis
behaviour correctly; the out-of-phase component of the drag force is negative near
the middle of the lock-on range (Tanida et al. 1973; Nishihara et al. 2005). Similarly,
when the cylinder is free to vibrate, the vortex drag force does not excite vibrations
for frequencies in the lock-on range. The present study links this behaviour to the
timing of a shedding vortex during the forward stroke of the cylinder that induces a
negative excitation from the fluid.
As discussed in § 3, the timing of vortex shedding is gradually shifted towards the
backward stroke of the cylinder as the reduced velocity increases or decreases from
U ∗ ≈ 2.5 (fe /fo ≈ 2). This enables a positive coupling between the vortex force and
the cylinder motion to be established as the frequency limits of the lock-on range
are approached and, therefore, vortex-induced vibrations ensue. Although the present
analysis correctly suggests a positive coupling at fe /fo = 2.2 which is inside the first
response branch in figure 5, it should be noted that the wake pattern observed in
the present forced-oscillation study corresponds to antisymmetric vortex shedding
(2S mode), whereas the first excitation region is associated with symmetric vortex
shedding. Naudascher (1987) classified the latter as motion-induced excitation or
‘wake-breathing’. This is a remarkable feature that highlights the intrinsic differences
between forced and free vibrations. The self-excited vibration of flexibly mounted
cylinders in the two excitation regions and the corresponding wake patterns is beyond
the scope of the present study but it is hoped that the discussion presented will
improve understanding of this problem and stimulate further the research interest in
it. It seems plausible that a phase ‘jump’ or transition might exist near the end of the
first response branch to account for the change in the wake pattern. Furthermore,
the overlap between the lock-on and positive excitation regions on either side might
indicate hysteresis effects.
Finally, it is intriguing to ask if inflow perturbations may be used to damp the free
vibrations of flexibly mounted rigid cylinders or even those of flexible cylinders. Based
on the discussion presented above, it is conjectured that periodic inflow perturbations
or any means that can control the timing of vortex shedding in the cylinder wake
may, in principle at least, be used to suppress the free vibrations. This approach
departs from the established methodologies used for the reduction of vortex-induced
vibrations of cylindrical structures which have focused on the destruction of vortex
shedding, e.g. by three-dimensional geometric perturbations.

Partial support for conducting this research was provided by the Engineering and
Physical Sciences Research Council U.K. under grant GR/R29802/01.

REFERENCES
Armstrong, B. J., Barnes, F. H. & Grant, I. 1987 A comparison of the structure of the wake
behind a circular cylinder in steady flow with that in a perturbed flow. Phys. Fluids 30, 19–26.
Barbi, C., Favier, D. P., Maresca, C. A. & Telionis, D. P. 1986 Vortex shedding and lock-on of a
circular cylinder in oscillatory flow. J. Fluid Mech. 170, 527–544.
Bearman, P. W. 1984 Vortex shedding from oscillating bluff bodies. Annu. Rev. Fluid Mech. 16,
195–222.
Cantwell, B. & Coles, D. 1983 An experimental study of entrainment and transport in the
turbulent near wake of a circular cylinder. J. Fluid Mech. 136, 321–374.
Carberry, J., Govardhan, R., Sheridan, J., Rockwell, D. & Williamson, C. 2004 Wake states
and response branches of forced and freely oscillating cylinders. Eur. J. Mech. B Fluids 23,
89–97.
Gerrard, J. H. 1966 The mechanics of the formation region of the vortices behind bluff bodies.
J. Fluid Mech. 25, 401–413.
 The timing of vortex shedding in a perturbed flow 55
Griffin, O. M. & Ramberg, S. E. 1976 Vortex shedding from a cylinder vibrating in line with an
incident uniform flow. J. Fluid Mech. 75, 526–537.
Jeon, D. & Gharib, M. 2004 On the relationship between the vortex formation process and cylinder
wake vortex patterns. J. Fluid Mech. 519, 161–181.
Konstantinidis, E., Balabani, S. & Yianneskis, M. 2003 The effect of flow perturbations on the
near wake characteristics of a circular cylinder. J. Fluid Struct. 18, 367–386.
Konstantinidis, E., Balabani, S. & Yianneskis, M. 2004 A PIV study of the wake of a circular
cylinder subjected to low amplitude flow perturbations. In Proc. 8th Intl Conf. on Flow Induced
Vibration (ed. de Langre & Axisa). Ecole Polytechnique, Paris.
Konstantinidis, E., Balabani, S. & Yianneskis, M. 2005 Conditional averaging of PIV plane wake
data using a cross-correlation approach. Exps. Fluids 39, 38–47.
Naudascher, E. 1987 Flow-induced streamwise vibrations of structures. J. Fluids Struct. 1, 265–298.
Nishihara, T., Kaneko, S. & Watanabe, T. 2005 Characteristics of fluid dynamic forces acting on a
circular cylinder oscillated in the streamwise direction and its wake patterns. J. Fluids Struct.
20, 505–518.
Okajima, A., Nakamura, A., Kosugi, T., Uchida, H. & Tamaki, R. 2004 Flow-induced in-line
oscillation of a circular cylinder. Eur. J. Mech. B Fluids 23, 115–125.
Ongoren, A. & Rockwell, D. 1988 Flow structure from an oscillating cylinder. Part 2. Mode
competition in the near wake. J. Fluid Mech. 191, 225–245.
Tanida, Y., Okajima, A. & Watanabe, Y. 1973 Stability of a circular cylinder oscillating in uniform
flow or in a wake. J. Fluid Mech. 61, 769–784.
Williamson, C. H. K. & Govardhan, R. 2004 Vortex-induced vibrations. Annu. Rev. Fluid Mech.
36, 413–455.
Williamson, C. H. K. & Roshko, A. 1988 Vortex formation in the wake of an oscillating cylinder.
J. Fluids Struct. 2, 355–381.