You are on page 1of 11

Journal of Fluids and Structures 27 (2011) 648–658

Contents lists available at ScienceDirect

Journal of Fluids and Structures


journal homepage: www.elsevier.com/locate/jfs

Circular cylinder wakes and vortex-induced vibrations


P.W. Bearman n
Department of Aeronautics, Imperial College London, SW7 2AZ, UK

a r t i c l e in f o abstract

Article history: This paper presents a selective review of recent research on vortex-induced vibrations
Received 27 October 2010 of isolated circular cylinders and the flow and vibration of circular cylinders in a tandem
Accepted 30 March 2011 arrangement; a common thread being that the topics raised are of particular interest to
Available online 10 May 2011
the author. The influence of Reynolds number on the response of isolated cylinders is
Keywords: presented and recent developments using forced vibration are discussed. The response
Vortex-induced vibration of a cylinder free to respond in the in-line and transverse directions is contrasted with
Circular cylinder wakes that of a cylinder responding in only one direction. The interference between two
Vortex shedding circular cylinders is discussed and prominence given to the case of cylinders in a
Tandem cylinders
tandem arrangement. The origin of the time–mean lift force on the downstream
cylinder is considered together with the cause of the large amplitude transverse
vibration experienced by the cylinder above vortex resonance. This wake-induced
vibration is shown to be a form of vortex-induced vibration.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Vortex-induced vibration (VIV) of circular cylinders is a long established field of study that can be traced back to the
ancient world. In modern times it has been the subject of concentrated research, addressing questions of fundamental
interest in fluid dynamics and driven by applications to structures in air and denser fluids, particularly water. Progress has
been summarised in a series of review papers and books that have appeared at irregular intervals over the past 40 years.
The reviews include: Parkinson (1974), Sarpkaya (1979), Bearman (1984), Sarpkaya (2004), Williamson and Govardhan
(2004), Gabbai and Benaroya (2005). While this is by no means an exhaustive list, the papers illustrate the diversity of
approach to VIV and provide a valuable compilation and assessment of findings. With the present state of development it
would be an immense task to review all aspects of VIV in a single journal paper and there is no intention to attempt this
here. Rather the aim is to be selective and concentrate on a number of facets that interest the author and hopefully have
some relevance to current and future research on VIV of circular cylinders. But why confine our attention to circular
cylinders when other body shapes are susceptible to VIV? An obvious reason for the interest in the circular cylinder is its
extensive use as a structural form but, as Morkovin (1964) observed, it also presents a ‘‘kaleidoscope of fluid phenomena’’,
not least of which is the influence of Reynolds number. Hence it in no way lessens the challenge to restrict attention to
circular cylinders.
It is well known that the flow around a fixed circular cylinder is affected by a number of parameters, including Reynolds
number, surface roughness, free stream turbulence level, etc., and, although not comprehensively demonstrated, it seems
likely that the response of a flexible cylinder is sensitive to these same parameters. However, from this list we will

n
Tel.: þ 44 2075945055; fax: þ 44 2075941974.
E-mail address: p.bearman@imperial.ac.uk

0889-9746/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfluidstructs.2011.03.021
P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658 649

consider only the influence of Reynolds number on cylinder response in this paper. A further consideration to take into
account for flexible cylinders is in what way is the cylinder flexible? There are rigid cylinders that are elastically mounted,
either at both ends or at one end with the other end free, and there are flexible cylinders, such as cables and riser pipes,
where a range of modes can be excited. Much of the fundamental research on VIV has been performed with a rigid cylinder
elastically mounted at both ends such that the response is uniform along its length while in the majority of applications
the response varies along the cylinder axis. Clearly there is a need for an appreciation of the similarities and differences in
the responses for these two cases. However, while acknowledging that there are many challenges in understanding and
predicting the response of fully flexible cylinders, this review will concentrate on the basic building block of VIV research,
the response of an elastically mounted, rigid cylinder.
When applying dimensional analysis to elastically mounted cylinders there are important variables to consider related
to their structural mass/unit length, m, damping and stiffness. These may be incorporated into the following parameters:
mass ratio mn, the ratio of structural mass to the mass of displaced fluid, fraction of critical damping, x, reduced velocity,
U/N0D, where U is flow velocity, D cylinder diameter and N0 the structure natural frequency in the absence of any fluid
effects. It should be noted that this selection of parameters is not unique and alternative choices can be found in the
literature. In determining maximum VIV amplitude, in addition to damping, an important parameter is the mass ratio and
for structures in water this can be two to three orders of magnitude smaller than that typically found in air. For a given
damping level, as mass ratio reduces, the interaction between the fluid and the structure is more pronounced leading to
larger amplitudes of vibration over a wider range of flow speeds. As extensively studied in the work of Williamson and
Roshko (1988), we now know that for large amplitudes, and by large amplitudes we mean amplitudes greater than
about half a cylinder diameter, the mode of vortex shedding is not always the same as that described by the von Kármán
vortex street model with one vortex of each sign shedding into the wake per oscillation cycle of the structure. This
important aspect of VIV will not be discussed further here and readers are referred to the review by Williamson and
Govardhan (2004).
Considerable insight into VIV response can be obtained from a simple analysis that the author first came across in a
presentation by Parkinson (1974). He considered a mass, spring and dashpot system driven by the fluid force resulting
from vortex shedding. Applying this to transverse oscillations Parkinson made two important assumptions, the first being
that the force and response are sinusoidal with the same frequency N and secondly that the fluid force leads the response
by a constant phase angle F. When these assumed solutions are inserted into the equation for a flexible cylinder the
following relations for vibration amplitude and frequency emerge:
ym =D ¼ Cym sin Fð1=4p3 Þð1=m xÞðU=N0 DÞ2 ðN0 =NÞ; ð1Þ

N=N0 ¼ ½1Cym cos Fð1=2p3 Þð1=m ÞðU=N0 DÞ2 ðym =DÞ11=2 ; ð2Þ
where ym is the amplitude of transverse vibration and Cym the amplitude of the transverse fluid force coefficient.
Eqs. (1) and (2) illustrate very clearly the importance of the phase angle and the role played by mass and damping. For
there to be VIV, F must be between 01 and 1801; on the other hand it will be suppressed if the product mnx is sufficiently
high. From Eq. (2) we see that if F is between 01 and 901 then NoN0 and if F is between 901 and 1801 then N 4N0. We can
also observe that if mnc1 then N EN0, i.e., the cylinder will oscillate at or very close to its natural frequency. In
applications found in ocean engineering mn is often of order 1 and it is in this range of mass ratio where many interesting
phenomena have been found.
While it is important to consider isolated cylinders it is essential to understand how circular cylinders respond when
placed in arrays since there are a number of practical situations involving multiple cylinders. Approaches range from
studying large regular arrays of closely spaced cylinders to studying just two cylinders with varying centre to centre
separations and in a range of orientations from side-by-side to tandem. Here we will consider the flow and response of two
cylinder combinations, concentrating primarily on the in-line or tandem configuration.

2. Influence of Reynolds number on VIV response

From Eq. (1) we see that the product mnx plays a crucial role in determining vibration amplitude. As described by Griffin
et al. (1975), a collapse of data can be obtained if maximum VIV amplitude is plotted against this product to form the
so-called ‘Griffin plot’. There was a popular belief at the time that Reynolds number plays a minor role and that the flow
around a cylinder undergoing large vortex-induced vibrations is insensitive to Reynolds number changes. However, Klamo
et al. (2005) and Govardhan and Williamson (2006) have both since demonstrated the strong influence of Reynolds
number on the maximum response of a cylinder as the Reynolds number range of available data has increased. Fig. 1,
reproduced from Govardhan & Williamson, shows maximum amplitude plotted against Reynolds number for a number of
data sets, both experimental and computational, where mnx is either zero or very small. It should be noted that in
experiments zero damping was achieved by applying external negative damping equal in magnitude to the inherent
structural damping. Referring again to Eq. (1) we see that the Reynolds number dependency can arise due to changes in
Cym and F. The low Reynolds number data in the laminar regime is obtained from CFD studies and these are in good
agreement with the experimental result of Anagnostopoulos and Bearman (1992). In their experiment mnx E0.2 and they
observed an amplitude of 0.55D at a Reynolds number of 110.
650 P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658

Fig. 1. Maximum transverse amplitudes of vibration for very low mass-damping versus Reynolds number (reproduced from Govardhan and Williamson
(2006) where the symbols are defined).

There is more recent data at Reynolds numbers above those plotted in Fig. 1 indicating that the response amplitude
continues to rise. Raghavan and Bernitsas (2011) have published experimental results showing a maximum amplitude of 1.9D
at a Reynolds number of 8.7  104 for a value of mnx around 0.2. This result is in agreement with measurements of Ding et al.
(2004) at slightly higher Reynolds numbers. These results fall well above an extrapolation of the straight line fit shown in Fig. 1
and the reason for this needs to be clearly understood. Presumably the amplitude cannot continue to rise at this rate as the
critical regime is entered and some form of amplitude crisis may occur. Since it is in this general range of Reynolds number
where many practical applications are found it is important that this regime is subjected to further study.

3. Free vibration versus forced vibration

If we accept the assumptions underlying Eqs. (1) and (2) then this leads to the conclusion that VIV can be equally well
studied by forcing a cylinder to oscillate in sinusoidal motion. Ignoring for the moment Reynolds number, the parameters
that need to be controlled in a forced vibration study, say to obtain Cym and F, are the amplitude ratio, ym/D, and the
reduced velocity, U/ND, defined using the frequency of the forced motion. Experiments employing forced transverse
vibration have been carried out by a number of investigators, including Bishop and Hassan (1964), Sarpkaya (1978), Staubli
(1983) and Hover et al. (1998). In addition, a host of CFD studies has simulated forced motion. However, it is not known a
priori precisely what ranges of amplitude ratio and reduced velocity will be encountered when forced data is applied to
predict the response of a cylinder with given mass, damping and stiffness. This is particularly the case when mass ratio and
damping are low and so a substantial amount of data is required. Since the early work of Bishop and Hassan (1964) the
resolution in amplitude and frequency space has steadily increased. In the latest forced vibration experiments, Morse and
Williamson (2009) performed 5680 runs in a water flume, varying cylinder amplitude and frequency, at a constant
Reynolds number of 4000 and then they repeated the runs for a Reynolds number of 12,000. The acquisition of such
extensive sets of data represents a total of 2000 h of running time. The free and forced-vibration methods have their
advantages and disadvantages, but the forced-vibration technique has the important benefit that the frequency and
amplitude of vibration can be tightly controlled. With free vibration a small increase in flow speed can result in large
changes in cylinder oscillation amplitude, possibly accompanied by a change in the mode of vortex shedding. Such jumps,
or transitions, are difficult to study under conditions of free vibration. On the other hand, with forced vibration a large
number of runs have to be carried out in order to map the conditions under which energy is transfered from the fluid to the
cylinder.
There have been debates in the literature about whether the flows generated by freely oscillating and forced oscillated
cylinders can be the same, and questions regarding the importance of flow history and the role of harmonics in the
displacement trace of a freely vibrating cylinder have been raised. Also the influence of possible non-linearities in
structural quantities such as stiffness and damping cannot be ruled out. However, Morse and Williamson (2009) have
demonstrated that under carefully controlled conditions there is a very close correspondence between these flows if the
parameters are correctly matched. Fig. 2, taken from their work, shows measured and predicted amplitude responses for a
freely vibrating circular cylinder with mn ¼10.49 and x ¼ 0. The forced cylinder data is for a constant Reynolds number of
4000 whereas the Reynolds number of the freely vibrating cylinder varies but the Reynolds numbers are matched at the
peak amplitude.
This work opens the way for highly resolved, forced-vibration data sets to be used to investigate further the fluid/
structure interaction processes that occur during VIV. How might his line of research develop from here? An obvious
direction is to explore further the influence of Reynolds number on the unsteady fluid forces and on the modes of vortex
P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658 651

Fig. 2. Measured and predicted amplitude response for a freely vibrating circular cylinder: K, measured free-vibration response from Govardhan and
Williamson (2006); J, predicted response from forced-vibration data (reproduced from Morse and Williamson (2009)).

Fig. 3. In-line and transverse responses for a frequency ratio of 2, mn ¼2 and x ¼ 0.003 (Assi, 2009).

shedding. It should be noted that in a free-vibration experiment it is normal to fix the structural parameters and vary the
flow speed, and hence the Reynolds number. For a freely vibrating cylinder with low mass and damping the Reynolds
number can change appreciably as the flow speed is varied through the range where VIV occurs.

4. Simultaneous in-line and transverse vibrations

It is unusual in practical situations for cylinders only to have flexibility in the cross-flow direction and normally a
cylinder is free to vibrate due to vortex shedding in all directions normal to its axis. For an elastically mounted rigid
cylinder, free to respond only in the in-line direction, it may vibrate due to vortex shedding, albeit at a significantly lower
amplitude, at roughly half the flow speed required for transverse vibration since a fluctuation in drag is produced every
time a vortex is shed whereas a fluctuation in the transverse force requires vortices to be shed alternately from each side of
the cylinder. A long flexible cylinder has many modes of vibration and at a given flow speed it is possible to have
simultaneous VIV in the in-line and transverse directions, but with different structural modes excited. Depending on the
ratio of natural frequencies, an elastically mounted rigid cylinder may also vibrate simultaneously in the in-line and
transverse directions. Employing the forced-vibration technique discussed above it should be possible in theory to study
this aspect of VIV in more depth. Difficulties arise of course because there are now two amplitude and two frequency
variables, as well as a phase angle between the motions. Dahl et al. (2010) have recently published data on the VIV
response of a cylinder free to respond in the transverse and in-line directions and find, in common with several others, that
the cylinder vibrates in the in-line direction at precisely twice the frequency of the transverse response, for a range of
nominal natural frequency ratios. Such an observation significantly reduces the number of runs required to study forced
vibration for this case, but the task is still a daunting one.
Fig. 3, taken from the work of Assi (2009), shows the in-line, Ax, and transverse response, Ay, of an elastically mounted
rigid cylinder plotted against reduced velocity, based on the natural frequency of the cylinder in the transverse direction.
The ratio of in-line to transverse natural frequencies is 2, with mn ¼2 and x ¼0.003. We see that the maximum in-line
response is equal to roughly a third of the maximum transverse response and its magnitude is far higher than might have
652 P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658

been expected from work on cylinders free to respond in only the in-line direction. Although the parameters are not quite
the same as those for the purely transverse response plotted in Fig. 2, it is clear that the way the response varies with
reduced velocity is distinctly different.
Jauvtis and Williamson (2003) investigated the case where a rigid cylinder is allowed to respond in the in-line and
transverse directions with a frequency ratio of unity. A comprehensive study of the effect of the natural frequency ratio on
the response of an elastically mounted cylinder has been conducted by Dahl et al. (2006). Their results for in-line and
transverse amplitudes and frequencies versus reduced velocity are reproduced in Fig. 4. It can be seen that there are
systematic changes in the amplitude profiles as the natural frequency ratio is increased from 1 to 2. Although the
parameters are not identical in the two experiments there is reasonably close correspondence between the data plotted in
Fig. 3 for a frequency ratio of 2 and the results in Fig. 4 for a frequency ratio of 1.9. Measurements from an experiment on a
multi-mode riser model, reported by Chaplin et al. (2005), show a number of similar characteristics to those found using
two degrees of freedom, sectional models, particularly the amplitude and oscillation frequency ratios between in-line and

Fig. 4. Amplitude of cylinder motions and oscillation frequencies for different frequency ratios: K, transverse amplitude;  , in-line amplitude;
þ, transverse frequency; n, in-line frequency (reproduced from Dahl et al. (2006)).
P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658 653

transverse responses. Although the in-line amplitudes are typically only about a third of the transverse amplitudes they
occur at double the frequency and so their contribution to restricting fatigue life can be substantial. It seems clear that
there is a need for a fuller understanding of how in-line and transverse VIV responses interact and why oscillation
frequencies adopt a frequency ratio of 2.

5. Interference between circular cylinders

There are numerous situations where circular cylinders are found in arrays such that they can no longer be treated as
isolated when considering the flows around them and their susceptibility to flow-induced vibration. An obvious difficulty
in studying such problems from a fundamental point of view is the large range of possible independent variables; but we
know from past work that this is fertile ground for research and that there are many fascinating flow phenomena involved.
A very substantial body of research exists on the response of cylinders in large arrays, such as are found in cross-flow heat
exchangers. Rather than move directly to an array of multiple cylinders, here we will consider just two circular cylinders
and assume their diameters to be equal. If the cylinders are fixed then only two additional parameters are introduced: the
ratio of centre to centre spacing, s, to cylinder diameter and the angle, a, between the free stream direction and the line
between the centres; with a ¼01 being the case when the cylinders are in-line in a tandem arrangement. An
understandable question to ask is how much of what is known about isolated cylinders can be carried over to cylinders
in arrays?
The author’s first encounter with two circular cylinders was in a study (Bearman and Wadcock, 1973) of the flow
around a pair of fixed cylinders in a side-by-side arrangement, i.e. a ¼901. It was observed that when s/D was in the range
1.1–2 an instability occurred in which the gap flow became intermittently biased towards the near wake of one cylinder or
the other resulting in a marked difference in the drag forces on the two cylinders. Flow visualisation for s/D ¼1.85 is shown
in Fig. 5 and here the flow between the cylinders is biased towards the upper cylinder, resulting in a higher drag on the
upper cylinder. Later work has shown that a ¼901 represents a critical condition and that for values of a slightly less or
slightly greater than 901 the gap flow becomes permanently biased towards the wake of the cylinder that is marginally
upstream.
An interesting feature of the flow about side-by-side cylinders is that they experience a mean repulsive force whereas
ideal flow theory predicts they will be attracted. Zdravkovich (1977) compiled a very comprehensive set of data for the
mean forces acting on two circular cylinders for a wide range of separations and flow angles. The regions of flow
interference for two circular cylinders proposed by Zdravkovich (1988) and his map of interference regimes are shown in
Fig. 6. The small shaded areas indicate instability regions and the one on the y0/D, or side-by-side axis, has already been
mentioned. A second instability region can be found in the vicinity of the tandem cylinder axis and relates to whether or
not the shear layers from the upstream cylinder reattach on the downstream one. For a detailed assessment of the flow
around a pair of fixed circular cylinders, highlighting research carried out in the past 20 years, the reader is referred to the
review by Sumner (2010).
We will now concentrate on configurations with a between about 7351 and primarily in the region in Fig. 6 described
by Zdravkovich as ‘wake interference’. In this range, apart from the case where the downstream cylinder is positioned on
the x0/D axis and the mean lift is zero, Zdravkovich (1977) shows that on the downstream cylinder there is a mean lift, or
transverse force, acting towards the x0/D axis . Hence between a ¼01 and 901 the mean lift force on the cylinder designated
as the downstream cylinder changes sign. Assi (2009) presents measurements, for a number of centre to centre spacings
and Reynolds numbers, of the time–mean lift and drag coefficients on the rear cylinder for a range of a values spanning the
tandem arrangement. His measurements for a centre to centre spacing of 4D and Reynolds numbers around 104 are shown
in Fig. 7 plotted against the transverse centre to centre spacing of the cylinders, y0/D. In this case he found that the lift force
coefficient was zero on the line of centres, reached a maximum magnitude of about 0.65 at a transverse spacing of just
over 1D and returned to zero by about 3D. There are a number of theories proposed in the literature for the origin of the

Fig. 5. Side-by-side cylinders with a centre to centre spacing of 1.85D showing the flow biased towards the upper cylinder (Bearman and Wadcock,
1973).
654 P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658

Fig. 6. Interference regions for two circular cylinders (Zdravkovich, 1988).

Fig. 7. Time–mean lift and drag coefficients on the downstream cylinder of a cylinder pair for a centre to centre separation of 4D (Assi, 2009).

transverse force seen in Fig. 7 but, as Price (1976) pointed out, none of them provide a satisfactory explanation for its
magnitude although most predict the correct sign.
On the other hand the drag coefficient behaves very much as one might expect for a cylinder in the wake of another
cylinder. The finding of Price (1976) is important because tandem cylinders, i.e. a ¼01 when the cylinders are at rest, have
become the focus for much research on two fixed and two elastically mounted cylinders. In order to restrict the range of
parameters in flow-induced vibration studies a number of investigators have fixed the upstream cylinder and allowed the
rear one to oscillate only in the transverse direction. This is the case that will be discussed here simply because it offers the
possibility of being easier to understand compared to the situation when both cylinders are free to respond in any
direction.
When the cylinders of a tandem pair, with the upstream one fixed and the downstream one free to oscillate in the
transverse direction, are very far apart the rear cylinder behaves very much like an isolated cylinder and the response is
similar to the familiar VIV type. For very close separations the separating shear layers from the upstream cylinder may
attach steadily or unsteadily on to the rear cylinder and this has a significant influence on the form of the flow-induced
P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658 655

vibration. Of interest here, and in a number of practical applications, is the intermediate range of separations beyond about
2 or 3 cylinder diameters but less than say 20. Several investigators, including Bokaian and Geoola (1984) and Hover and
Triantafyllou (2001), have investigated this regime and observed that, depending on the value of reduced velocity, the
response can be of the VIV type or, at higher reduced velocity, it appears to be similar to galloping. Measurements by
Hover and Triantafyllou (2001) and Assi et al. (2006) of the transverse response amplitude of the downstream cylinder
versus reduced velocity are shown in Fig. 8. The parameters for the two sets of measurements are close but not identical
with s/D equal to 4 in one case and 4.75 in the other. It can be seen that the response continues to rise beyond the reduced
velocity expected for resonance with vortex shedding and that the amplitude becomes considerably greater than that
expected for an isolated cylinder at a similar Reynolds number.
An important question that arises from Fig. 8 is what is the mechanism causing the oscillations to continue to increase with
increasing reduced velocity in a way that is similar to a galloping response, such as that experienced by a square section cylinder?
If it is due to galloping it cannot be of the classic kind described by Den Hartog (1956), where a fluid dynamic instability occurs,
related to the cross-sectional shape of the body, and its motion generates forces that increase the amplitude of vibration. This is
because in the tandem cylinder case when the rear cylinder is displaced in the transverse direction away from the line of centres
there is a hydrodynamic restoring force, as shown in Fig. 7, which is acting to return the cylinder to its original position. Hence
quasi-steady arguments, as used in galloping theory, suggest stability of the rear cylinder rather than instability.
Assi et al. (2010) studied the transverse response of the rear cylinder for a range of separations, as shown in Fig. 9 for
centre to centre separations between 4 and 20 diameters. The response peak at a reduced velocity of about 5 is clear and is
a direct result of vortex shedding from the rear cylinder. The response at a given reduced velocity above the VIV peak
decreases steadily as the separation is increased from 4 to 20 diameters and appears to level out at the highest reduced
velocities tested. Further, Assi et al. (2010) proposed that the origin of the steady transverse force on the rear cylinder of a
tandem cylinder pair and the excitation responsible for the response above vortex resonance is a result of vortex
interactions between vortices shed from the front and rear cylinders. Hence if vortex shedding is totally suppressed from
the upstream cylinder the response of the downstream cylinder for reduced velocities beyond its own VIV peak should also
be suppressed. Convincing evidence for this is obtained by taking a separation of 4D and replacing the unsteady wake of
the upstream cylinder by a steady wake, achieved by using suitable resistance screens, with a near identical time–mean
velocity profile. The time–mean lift and drag coefficients of the downstream cylinder in this steady wake flow are shown in
Fig. 10 and, comparing with the results in Fig. 7, we see that removing vortex shedding reduces the maximum amplitude of
the transverse force from 0.65 to 0.2. This residual lift could be due to one or more of the mechanisms discussed by Price
(1976). It can be seen that the variation of the drag coefficient with transverse position is very much the same as that
obtained with an upstream cylinder thus confirming that the momentum deficit across the steady wake is very close to
that experienced in the mean by the downstream cylinder of a tandem pair.
Turning to the response of a cylinder in a wake without the presence of regular discrete vortices, we observe in Fig. 11
that the transverse response is very similar to that of a single cylinder in a uniform approaching flow. The small shift in the
peak response to a higher reduced velocity, where reduced velocity is defined using the outer free stream velocity, is
consistent with the cylinder being in a wake. The high level of response found in the tandem arrangement at reduced
velocities above vortex resonance disappears. For a tandem pair Assi et al. (2010) describe how vortices shed from the
upstream cylinder interact with vortices formed at the downstream cylinder to enhance or decrease the transverse force
experienced by the downstream cylinder, depending on the phasing of the arrival of a vortex from upstream. Sketches of
the forms of the interactions involved are reproduced in Fig. 12. It is the time average effect of these interactions, which
gives rise to the main part of the time average transverse force plotted in Fig. 7.

Fig. 8. Transverse response amplitude versus reduced velocity for the downstream circular cylinder of a pair of cylinders in a tandem arrangement:
K, s/D ¼ 4 (Assi et al. 2006); B, s/D ¼ 4.75 (Hover and Triantafyllou, 2001).
656 P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658

Fig. 9. Influence of centre to centre separation distance and reduced velocity on the response of the downstream cylinder (Assi et al., 2010).

Fig. 10. Time–mean force coefficients acting on a cylinder placed in a steady wake without discrete vortices (Assi et al., 2010).

The combined effect of the vortex interactions produces a transverse response of the downstream cylinder at its
frequency of vibration but with variable amplitude. This frequency is greater than its true natural frequency due to the
influence of the time average transverse force that acts as an additional stiffness. Having found the cause of the vibration of
the downstream cylinder how do we classify it? Assi et al. (2010) describe it as a wake-induced vibration but it is not
simple buffet because the unsteady transverse force driving the oscillation arises from an interaction between vortices
shed from the upstream cylinder and those shed from the downstream cylinder. At high reduced velocities the frequency
P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658 657

Fig. 11. Effect of a steady approaching wake flow on the transverse response of a circular cylinder (Assi et al., 2010). O, isolated cylinder in a steady
wake flow; B, downstream cylinder of a tandem arrangement with s/D¼ 4; K, isolated cylinder in a uniform free stream.

Fig. 12. Sketches showing the generation of an unsteady transverse force on the downstream cylinder for a reduced velocity above vortex resonance
(Assi et al., 2010).

of vortex shedding is higher than the oscillation frequency of the cylinder so not every vortex shed from the upstream
cylinder causes the types of interaction seen in Fig. 12. However, their effect is sufficient to sustain oscillations and while
this is certainly a wake-induced vibration it also appears to be a hitherto unrecognised form of vortex-induced vibration.
A further conclusion from this work is that the understanding of the flows and responses of two or more cylinders presents
a continuing challenge and there may be yet more modes of shedding and mechanisms of response to be uncovered.

6. Conclusions

Some conclusions that can be drawn from this very selective review of circular cylinder wakes and vortex-induced
vibrations are listed here. The strong dependency on Reynolds number of the flow around a fixed circular cylinder is well
known but it has only been appreciated relatively recently that the response of a flexible cylinder with a low product of
mass ratio and damping is also highly susceptible to Reynolds number changes. While its influence is well documented at
low Reynolds numbers more research is required to understand its effect at high subcritical values and through the critical
range since many practical applications are found in these regimes. Using current experimental techniques much more
finely resolved forced-vibration studies can now be carried out and this poses the question as to what more can be learnt
about VIV using forced vibration? While much of the basic study of VIV has been conducted on cylinders free to respond
only in the transverse direction the variation of transverse response with reduced velocity is markedly different when a
658 P.W. Bearman / Journal of Fluids and Structures 27 (2011) 648–658

cylinder is free to move with two degrees of freedom. Also the in-line response is significantly greater than when a
cylinder is constrained to only move in the in-line direction. This poses a number of fundamental questions about how the
flow responds when a cylinder is free to oscillate transverse and in-line, including how the frequencies of oscillation adopt
a ratio equal or very close to 1:2. When two circular cylinders are arranged such that their flow fields can interfere with
each other a number of interesting phenomena emerge, as documented by previous researchers. Here we have
concentrated on two cylinders in a tandem arrangement where the rear cylinder is free to respond in the transverse
direction. In addition to a VIV response, the downstream cylinder experiences a wake-induced vibration caused by
interactions between vortices shed from both cylinders. It is proposed that this is another type of vortex-induced vibration.
However, this may present less than half the story because to study this problem fully both cylinders need to be flexible.
Are there new modes of shedding and response to be uncovered for two flexible circular cylinders with interacting flows?

Acknowledgements

The author wishes to acknowledge the following for stimulating discussions on topics raised in this review: Gustavo
Assi and Bruno Carmo, both at the University of Sa~ o Paulo, Brazil and Fran Huera Huarte, University Rovira i Virgili,
Tarragona, Spain.

References

Anagnostopoulos, P., Bearman, P.W., 1992. Response characteristics of a vortex-excited cylinderat low Reynolds numbers. Journal of Fluids and Structures
6, 39–50.
Assi, G.R.S., Meneghini, J.R., Aranha, J.A.P., Bearman, P.W., Casaprima, E., 2006. Experimental investigation of flow-induced vibration interference between
two circular cylinders. Journal of Fluids and Structures 22, 819–827.
Assi,G.R.S., 2009 Mechanisms for Flow-Induced Vibration of Interfering Bluff Bodies. PhD thesis, Imperial College, London.
Assi, G.R.S., Bearman, P.W., Meneghini, J.R., 2010. On the wake-induced vibration of tandem circular cylinders: the vortex interaction excitation
mechanism. Journal of Fluid Mechanics 661, 365–401.
Bearman, P.W., Wadcock, A.J., 1973. The interaction between a pair of circular cylinders normal to a stream. Journal of Fluid Mechanics 61, 499–511.
Bearman, P.W., 1984. Vortex shedding from oscillating bluff bodies. Annual Review of Fluid Mechanics 16, 195–222.
Bishop, R.E.D., Hassan, A.Y., 1964. The lift and drag forces on a circular cylinder oscillating in a flowing fluid. Proceedings of the Royal Society London
Series A 227, 51–75.
Bokaian, A., Geoola, F., 1984. Wake-induced galloping of two interfering circular cylinders. Journal of Fluid Mechanics 146, 383–415.
Chaplin, J.R., Bearman, P.W., Huera Huarte, F.J., Pattenden, R.J., 2005. Laboratory measurements of vortex-induced vibrations of a vertical tension riser in a
stepped current. Journal of Fluids and Structures 21, 3–24.
Dahl, J.M., Hover, F.S., Triantafyllou, M.S., 2006. Two-degree-of-freedom vortex-induced vibrations using a force assisted apparatus. Journal of Fluids and
Structures 22, 807–818.
Dahl, J.M., Hover, F.S., Triantafyllou, M.S., Oakley, O.H., 2010. Dual resonance in vortex-induced vibrations at subcritical and supercritical Reynolds
numbers. Journal of Fluid Mechanics 643, 395–424.
Den Hartog, J.P., 1956. Mechanical Vibrations, 4th ed. McGraw Hill.
Ding, J., Balasubramanian, S., Lokken, R., Yung, T., 2004. Lift and damping characteristics of bare and staked cylinders at riser scale Reynolds numbers.
In: Proceedings of Offshore Technology Conference, Paper No. 16341.
Gabbai, R.D., Benaroya, H., 2005. An overview of modelling and experiment of vortex-Induced vibration of circular cylinders. Journal of Sound and
Vibration 282, 575–616.
Govardhan, R.N., Williamson, C.H.K., 2006. Defining the ‘modified Griffin plot’ in vortex-induced vibration: revealing the effect of Reynolds number using
controlled damping. Journal of Fluid Mechanics 561, 147–180.
Griffin, O.M., Skop, R.A., Ramberg, S.E., 1975 The resonant vortex-excited vibrations ofstructures and cable systems. In: Proceedings of Offshore
Technology Conference, Houston, Texas, OTC Paper 2319.
Hover, F.S., Techet, A.H., Triantafyllou, M.S., 1998. Forces on oscillating uniform and tapered cylinders in cross flow. Journal of Fluid Mechanics 363,
97–114.
Hover, F.S., Triantafyllou, M.S., 2001. Galloping response of a cylinder with upstream wake interference. Journal of Fluids and Structures 15, 503–512.
Jauvtis, N., Williamson, C.H.K., 2003. Vortex-induced vibration of a cylinder with two degrees of freedom. Journal of Fluids and Structures 17, 1035–1042.
Klamo, J.T., Leonard, A., Roshko, A., 2005. On the maximum amplitude for a freely vibrating cylinder in crossflow. Journal of Fluids and Structures
21, 429–434.
Morkovin, M.V., 1964 Flow around circular cylinders: a kaleidoscope of challenging fluid phenomena. In: Proceedings of the Symposium On Fully
Separated Flow, ASME.
Morse, T.L., Williamson, C.H.K., 2009. Prediction of vortex-induced vibration response by employing controlled motion. Journal of Fluid Mechanics
634, 5–39.
Parkinson, G.V., 1974. Mathematical models of flow-induced vibrations. In: Naudascher, E. (Ed.), In Flow Induced Structural Vibrations. Springer, Berlin.
Price, S.J., 1976. The origin and nature of the lift force on the leeward of two bluff bodies. Aeronautical Quarterly 26, 154–168.
Raghavan, K., Bernitsas, M.M., 2011. Experimental investigation of Reynolds number effect on vortex induced vibration of rigid circular cylinder on elastic
supports. Ocean Engineering 38, 719–731.
Sarpkaya, T., 1978. Fluid forces on oscillating cylinders. Journal of Waterway, Port, Coastal and Ocean Engineering 104, 275–290.
Sarpkaya, T., 1979. Vortex-induced oscillations. Journal of Applied Mechanics 46, 241–258.
Sarpkaya, T., 2004. A critical review of the intrinsic nature of vortex-induced vibrations. Journal of Fluids and Structures 19, 389–447.
Staubli, T., 1983. Calculation of the vibration of an elastically mounted cylinder using experimental data from forced vibration. Journal of Fluids
Engineering 105, 225–229.
Sumner, D., 2010. Two circular cylinders in cross-flow: a review. Journal of Fluids and Structures 26, 849–899.
Williamson, C.H.K., Govardhan, R., 2004. Vortex-induced vibrations. Annual Review of Fluid Mechanics 36, 413–455.
Williamson, C.H.K., Roshko, A., 1988. Vortex formation in the wake of an oscillating cylinder. Journal of Fluids and Structures 2, 355–381.
Zdravkovich, M.M., 1977. Review of flow interference between two circular cylinders in various arrangements. Journal of Fluids Engineering, 618–633.
Zdravkovich, M.M., 1988. Review of interference-induced oscillations in flow past two circular cylinders in various arrangements. Journal of Wind
Engineering and Industrial Aerodynamics 38, 197–211.

You might also like