You are on page 1of 18

Technical Editor's Note:

The following review paper was prepared by invitation of the Applied Mechanics Division to acquaint our readers with
some of the mechanics work which has been done and some of the problems which remain in the important field of
fluid-solid interaction. ENGINEERING SOCIETIES

JUN2 61979

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


LIBRARY
T. Sarpkaya
Vortex-Induced Oscillations
Professor,
Department of Mechanical Engineering,
Naval Postgraduate School, A Selective Review
Monterey, Calif. 93940.
Fellow ASME

This paper reviews the vortex-induced oscillations in a few specific fundamental cases.
Research topics discussed are vortex shedding from a stationary bluff body; consequences
of the synchronization phenomenon; wake-oscillator models; added mass, damping, and
dynamic response measurements; flow-field models and the discrete-vortex method;
mechanism of synchronization; and, finally, in-line oscillations. Because of the selective
nature of the review, a fairly comprehensive listing of recent contributions to the litera-
ture on these and related aspects of flow-induced oscillations research is an essential part
of the exposition.

Introduction been incorporated into the experimental investigation of flow-induced


oscillations and into the evolution of conceptually plausible idealized
Experience has shown that elastic structures near linear-resonance models.
conditions can develop flow-induced oscillations by extracting energy In the material to follow emphasis is placed on the understanding
from the flow about them. The oscillations of the structure modify of the transverse oscillations of cylinders in uniform flow. The dis-
the flow and give rise to nonlinear interaction. This is in addition to cussion of the laboratory and numerical experiments occupy the major
any nonlinearity which can arise from the restoring force (variable portion of the review. No attempt has been made to be encyclopedic
support stiffness) and/or from response-dependent structural in coverage as there are recent texts and survey articles, for example,
damping. The understanding of these nonlinear interactions is of Blevins [1], Parkinson [2], King [3], and McCroskey [4], which cover
paramount importance. the field in some detail.
Four general classes cover a wide range of the fluid-structure in-
teraction phenomenon: vortex-induced oscillations, flutter, galloping, Vortex Shedding From a Stationary Bluff Body
and buffeting. The present review will report on only the first of these There are substantial gaps in the understanding of the fundamental
phenomena. mechanisms at work in the forcing of a body by the forces acting on
The problems associated with vortex-induced oscillations and their its afterbody and in the prediction of the lock-in range in which res-
important consequences have proved to be exceedingly difficult both onant oscillations occur. To a degree, the difficulties in obtaining a
theoretically and experimentally. The difficulty is partly due to our solution or in devising a model stem from an inadequate knowledge
incomplete knowledge of the flow field about a stationary body but of the precise mechanism of flow separation and its consequences in
it is also due to fundamental problems associated with the coupling steady flow past a stationary body (see, e.g., Morkovin [5], Marris [6],
of the oscillations and the flow field. It would indeed be gratifying to Mair and Maull [7], and Berger and Wille [8] for comprehensive re-
note that our understanding of the separated flow about a bluff body, views and references).
particularly in the region enclosing the body and its near wake, is
Observations as well as numerical experiments have shown that the
complete through laboratory and numerical experiments. This is,
wake of a bluff body is comprised of an alternating vortex street. This
regrettably, not yet the case, but there have been nevertheless sig-
phenomenon is intrinsic to the flow itself and is a consequence of the
nificant advances since the time attention has shifted from idealized
interaction between the shear layers, base pressure, diffusion, and
vortex streets to the details of separation, three-dimensional flow and
dissipation of vorticity, and the far wake, i.e., a consequence of the
the role played by the afterbody on the formation, growth, and motion
particular vorticity distribution throughout the field.
of vortices. Many of the key developments of the past decade have
As the vortices grow symmetrically, say, in an impulsively started
flow about a cylinder, the shear layer joining the separation point to
Contributed by the Applied Mechanics Division for publication in the
JOURNAL OF APPLIED MECHANICS. one of the vortices begins to develop instabilities (presumably Toll-
Discussion on this paper should be addressed to the Editorial Department, mien-Schlichting instabilities [9]) and is drawn across the wake in
ASME, United Engineering Center, 345 East 47th Street, New York, N. Y. response to the base pressure reduced by the action of the vortex
10017, and will be accepted until September 1, 1979. Readers who need more growing across the wake (see Fig. 1). This nearly corresponds to a time
time to prepare a Discussion should request an extension of the deadline from
the Editorial Department. Manuscript received by ASME Applied Mechanics when the sheet drawn in has least circulation or is most permeable.
Division, January, 1979. The stretching, diffusion, and dissipation of vorticity break up the

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 241


Copyright © 1979 by ASME
-1.0 -0.5 0.0 0.5
70 p i i i i I'* i i i"i" v i

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


Fig. 1 (a) Initiation of vortex shedding [ 10]

8 0 1 - t*
deforming turbulent sheet and thereby the further supply of circu-
lation to the vortex whose rate of growth has already been reduced Fig. 1(b) Alternate shedding of vortices and its relation to the lift force
to its minimum. This corresponds to the shedding of the first vortex. [10]
The vortex across the wake still continues to grow (but at a decreasing
rate) and entrains part of the oppositely signed vorticity left in the
range of values of D and V. This postulated constant is often called
wake by the cut sheet and the irrotational fluid drawn from outside
the Strouhal number and Strouhal's own data suggested a value of
through the opening created by the shedding of the first vortex. The
about 0.185. The characterization of the vortex-shedding process by
shedding process for the second vortex does not commence until the
a simple frequency is a practical simplification. As first noted by Lord
circulation in its feeding sheet decreases to its minimum, making the
Rayleigh [13], the Strouhal number is a function of the Reynolds
sheet most susceptible to rapid diffusion. Simultaneously, the sheet
number for a given body 2 and / should be written as / = So(Re) VI
deforms, diffuses, and is drawn across the wake by the action of the
D.
base pressure and the vortex growing on the other side of the wake.
Then the shedding cycle repeats itself.1 The aforementioned mech- At relatively low subcritical Reynolds numbers, the energy con-
anism, exhibited by numerical experiments [10] is quite similar, but taining frequencies are confined to a narrow band, and the Strouhal
not identical, to that suggested by Gerrard [11]. number is about 0.20 for smooth cylinders [8]. For Reynolds numbers
larger than about 20,000, only an average Strouhal number may be
The separation point at the side of the cut sheet just passes through
defined. In the range 0.5 X 106 < Re < 2.5 X 106 no periodicity is ob-
its minimum angle. The sheet at the opposite side of the wake bears
served in either the unsteady lift or the unsteady pressures. In the
nearly the maximum circulation, relative to any other time, and its
range 2.5 X 106 < Re < 6.5 X 106 a small peak is observed at S 0 = 0.2
separation-point angle just passes through a maximum. The motion
in the unsteady lift and in unsteady pressures only at the front part
of the stagnation point is such that it is 180 deg out-of-phase with the
of the cylinder. No peak is found behind the separation point. For Re
separation points, i.e., the separation point is below the x -axis (flow
> 6.5 X 106, a definite peak is found in both the unsteady lift and the
from left to right) when a vortex is shed from the bottom of the cyl-
unsteady pressures at So = 0.3. The said periodicity occurs both at
inder.
the front and rear part of the cylinder [16]. The foregoing categori-
The shedding of the vortices gives rise to an oscillating side thrust, zation is somewhat arbitrary. Jones, et al., [17], classified the unsteady
upon a cylinder of suitable afterbody, in a direction away from the last lift force in three regimes: wide-band random for 1.1 X 106 < Re < 3.5
detached vortex. The relationship between the vortex-shedding fre- X 106; narrow-band random for 3.5 X 106 < Re < 6 X 106; and quasi-
quency, cylinder diameter, and the velocity of the ambient flow has periodic for 6 X 106 < Re < 18.7 X 106.
been discovered by Strouhal [12] almost exactly 100 years ago in
The spectral content of the exciting forces is particularly important
connection with his work on a special method of creation of sound.
His data showed that the product fD/V is nearly constant for a wide

2
So = 0.21(1 - 20/Re) for 40 < Re < 1000 (Goldstein [14]; S0 = 0.212 (1 -
1
For a circular cylinder there is a 60-deg phase difference between the lift 12.7/Re) in the range 400 < Re < 100,000 according to a correlation obtained
maximum and the sheet cutting time, i.e., a vortex is shed when the lift force by Roshko [15]. In the range 400 < Re < 1200 the results of Roshko's correlation
decreases to 50 percent of the absolute value of its maximum. are slightly below the best-fit line to the existing data [8].

242 / VOL. 46, JUNE 1979 Transactions of the ASME


for bodies which may undergo in-line and/or transverse oscillations Table 1 Universal Strouhal numbers
since the vortex shedding frequency locks on to the frequency of the So Value Body shape Source
transverse oscillations of the cylinder when there is a transverse force
and when the vortex-shedding frequency is in the neighborhood of foh'/VK 0.163 Two-dimensional Roshko [19]
foUV 0.181 Two-dimensional Bearman [20]
the natural frequency of the cylinder. When the flow about a freely folJVK 0.190 Axisymmetric Calvert [21]
oscillating cylinder is in the critical regime, the sharp rise in the folw/VK 0.163 Two-dimensional Simmons [22]
Strouhal number does not occur and remains at a value nearly equal
to that found at subcritical Reynolds numbers [18].
Various attempts have been made to devise a universal Strouhal
Table 2 Correlation lengths
number which would remain constant for differently shaped two-
Reynolds No. range Correlation length Source
dimensional and axisymmetric bodies (see Table 1).
In Table 1, h' represents the wake width as determined from 40 < Re < 150 15D-20D [34]

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


Roshko's notched hodograph theory [19]; lu, the longitudinal spacing 150 < Re < 10 5 2D - 3D [34]
of the vortices as determined by Kronauer's minimum drag criteria 10" < Re < 4.5 i< 10" 3D - 6D [32]
as reported by Bearman [20]; lw, the wake width at a distance of the Re > 10 s 0.05D [34]
wake establishment region; and K = (1 — C p t) 1 / 2 where Cpt, is the base Re = 2 X 105 1.56D [29]
pressure coefficient.
The foregoing definitions suffer from the obvious drawback that
they require the solution of the wake formation problem first, or the d'Alembert flow and obtained So = 0.2028. His analysis does not
measurement of one or more flow characteristics. consider the wake formation and thus it is not expected that flows
Theoretical or semiempirical predictions of the wake characteristics around objects that induce large wakes could be treated with a simple
have been attempted by various researchers. Birkhoff [23] demon- perturbation of the unseparated flow.
strated that the longitudinal spacing /„ is trivially invariant since the The constancy of the Strouhal number over a broad range of
longitudinal velocity of vortices is uniformly bounded. He has shown Reynolds numbers does not imply that the base pressure remains
further that in an inviscid fluid the lateral spacing also remains con- constant and that a simple two-dimensional vortex emanates from
stant at a value equal to the initial spacing of the shear layers. These a separation line. In reality, there is not only a phase shift between
led Birkhoff to the conclusion that hv/lv is determined by its initial various sections of the vortex, separated by a correlation length, 3 but
value and not by von Karman's stability criteria oihvllv = 0.281. The also variations in both the intensity and the frequency of vortex seg-
use of a wake oscillator model ("the wake swings from side to side, ments.
somewhat like the tail of a swimming fish" [23]) and some experi- The variation of the base pressure with Reynolds number, in the
mentally determined values led Birkhoff to So = 0.2. One must keep range where the Strouhal number remains practically constant, may
in mind that vorticity in real vortices is not concentrated in points, be related to the variation of the mean vorticity flux or to the variation
the vortices are noncircular and distort and rotate as they move of the correlation length with the Reynolds number, turbulence,
downstream [24], vorticity diffuses and is swept across the wake [25], length-to-diameter ratio, and surface roughness [29-33]. Table 2, as
and dissipated by turbulence [8]. Vortices are subjected to strain fields compiled by King [3], gives an approximate idea about the typical
imposed by nearby vortices. The resulting patterns are ever-changing values of the correlation length.
vortex shapes encompassing elliptic and pearlike geometries. The
The net effect of the span wise variations of the vortex tube is that
complexity of the interaction between strained distorting vortices is
the transverse force (lift) coefficient obtained from a pressure inte-
further exacerbated by the addition of turbulence to the wake, as this
gration is not necessarily identical with that obtained from the direct
is likely to produce a more diffusive vorticity distribution and thus
measurements of the lift force. Partial spanwise correlation leads to
an additional shear field [24]. Thus the near constancy of wake
variations in both the frequency and the amplitude of the lift force,
characteristics in the range of Reynolds numbers where a vortex street
the variation of the latter being more pronounced than that of the
might be observed is primarily due to the slow variation of lu, h„, and
former.
So with respect to the strength and deformation of vortices. Never-
The lack of correlation exits not only spanwise but also chordwise
theless, theoretical idealizations of the wake give some clues about
[35-38] and that the chordwise correlation is related to the spanwise
the asymptotic behavior of the wake. In particular, one obtains from
correlation. Comparison between various results suggests that with
folu = V - us, where V — us is the velocity relative to the body, and
increasing Reynolds number over the range 104 to 106, the chordwise
from use of von Karman's stability criteria that
correlation for square and circular cross-section cylinders is main-
-_ 1 T/VD tained or improved. This leads to an increase in fluctuating lift coef-
(1) ficient. The reasons for these variations are not quite clear. The end
° ~ IJD 2V2(IJDT-
effects [29, 39-41], wall boundary layers, free-stream turbulence
where r represents the strength of a vortex. Assuming that lu/D re- [42-44], nonuniformity of the flow are mentioned often as possible
mains nearly constant (lu/D rn 5), one concludes that the smaller the reasons. The complexity of the three-dimensional nature of the flow
strength of the vortices, the larger is the Strouhal number (smaller about a cylinder is clearly demonstrated with measurements by
drag and narrower wake) and vice versa. The vortices which are Tournier and Py [45].
subjected to smaller dissipation linger a littler while longer in the near It would not be correct to assume that the mobility of the separation
wake relative to those which are dissipated more, i.e., strong vortices points is primarily responsible for the imperfect coherence. Even
get stronger. However, the dependence of So on F/VD is rather weak. bodies such as 90-deg wedges, square cylinders, with fixed separation
This explains in part the reason for the success of the discreet vortex lines, do not exhibit perfect correlation. However, the variation of the
models in predicting the Strouhal number fairly accurately [26, 27] base-pressure coefficient for bodies with mobile separation lines is
in spite of the fact that the calculated vortex strengths are about 35 greater than that for bodies with fixed separation lines [46]. In fact,
percent larger than those estimated experimentally. However, weak, for flow over symmetrical wedges the base pressure appears to be
the dependence of So on F/VD becomes a primary factor in the lock- insensitive to the nose angle [46].
ing-on of the vortex shedding to the natural frequency of the body. As noted earlier, the strength of the vortices plays an important role,
The vortices which are stronger continue to be fed by their shear layers particularly in the near wake. Laboratory and numerical experiments
a longer time period, thus further reducing the Strouhal number. In have shown that [7,19, 47-50] the net circulation of a rolled-up vortex
this process the mobility of the separation points is important but not
necessary.
3
Sacksteder [28] pursued a theoretical approach to determine the The equivalent length over which the velocity fluctuations in the wake may
Strouhal number at large Reynolds numbers by perturbing the be described as perfectly correlated.

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 243


of the street is 40 to 60 percent smaller than that generated in the
boundary layer during a shedding cycle. Prandtl determined that the
initial vorticity decreases to about half where the first vortex centers 100

appear. Vorticity is ultimately dissipated by viscosity to which it owes Jj = 0 - 0 0 1 0 3


its generation. Nevertheless, one may think of loss of circulation
a = 0.00166
through cancellation of oppositely signed vorticity. Primarily, there o
1.5 50
are three mechanisms whereby oppositely signed vorticity are brought S = 0.198
close together: vorticity generated on the forebody is carried by the //
shear layers near that generated on the afterbody; vorticity of the
deformed and cut sheet is carried across the near wake by the en- /
trainment of the irrotational fluid; and, finally, vorticity is swept
across the entire wake [25]. The percentages quoted in the literature

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


[8] for the total loss of vorticity often imply that the vortices, once
having acquired a certain circulation, retain that circulation l.o
throughout the rest of their motion. The fact that circulation decreases
continuously with time or distance is demonstrated clearly by the
2.0
experiments of Schmidt and Tilmann [50] and Bloor and Gerrard [48].
The amount of vorticity generated in the boundary layers and the
amount dissipated are of prime importance not only for the flow about V |U" IX
stationary bluff bodies but also for those undergoing resonant oscil- 1.5
0.6
lations. In fact, the entire bluff body problem may be reduced to the
determination of the vorticity distribution throughout the flow field.
0.5
This is not yet possible for Re larger than about 100. The determi-
nation of the vortex strengths is difficult and sensitive to the theo-
//!/ 1.0 0.4

retical and experimental means employed.


It is evident from the foregoing and from a more detailed perusal
of the references cited that the description of the near wake of a bluff 0.2
body is in a primitive state. Much of what is known about the conse-
quences of separation has come from laboratory experiments. It has
not yet been possible to develop a numerical model with which ex-
periments may be conducted to explain the observed or inferred re- 1.0 V/unD 1.2
lationships between various parameters and to guide and complement
the laboratory experiments. The principal difficulties are as fol- Fig. 2 Response characteristics of a freely oscillating cylinder, from Feng
lows: [54], (see also [2])

1 Separation Points. They represent a mobile boundary be-


tween two regions of vastly different scales. This in turn leads to stretching of vortex filaments and in the redistribution of vorticity
complex physical nonuniformities in relatively narrow regions which in all directions. The numerical models are not in a position to account
cannot be handled within the framework of the boundary-layer theory for such complex effects. One may hope to assess the effects of
[51]. Finite-difference and Marker and Cell (MAC) techniques require three-dimensionality by means of two-dimensional numerical ex-
in such regions very small computational times. The discretization periments.
of the continuous process of vorticity generation by line vortices in It is against this background that the nonlinear fluid-structure
the vicinity of a mobile or fixed singular point (discrete vortex model) interaction projects an order of magnitude more complex problems.
strongly affects the existing nonuniformities and promotes earlier The picture emerging from the numerous experiments and mathe-
separation. Attempts to preserve the prevailing flow conditions, say matical models is that there are phenomena to be explained, param-
by limiting the influence of the nascent vortices, while satisfying a eters to be quantified, and conjectures to be justified. It is a research
relatively simple separation criteria lead to hydrodynamical incon- area which will continue to challenge the imagination of both the
sistencies and nondisposable parameters. experimentalist and the theoretician for generations to come.
2 Reynolds Number. Finite-difference schemes for bluff-body
flows are limited to relatively small Reynolds numbers whether the Lock-In or Synchronization
scale of flow is assumed to be governed by a constant viscosity or by Description and Consequences of the Phenomenon. Numerous
a constant eddy viscosity. The large recirculation region of the flow experiments have shown that when the vortex-shedding frequency
is often comprised of turbulent vortices even when the boundary layer brackets the natural frequency of an elastic or elastically mounted
is laminar. The transition to turbulence moves upstream in the shear rigid cylinder with a suitable afterbody (capable of giving rise to a
layers as the Reynolds number is increased from about 10 3 to 5 X 104. transverse force), the cylinder takes control of the shedding in ap-
At Re = 5 X 104, it reaches the shoulder of the cylinder [52]. It does parent violation of the Strouhal relationship. Then the frequencies
not move appreciably further upstream before the critical Reynolds of vortex shedding and the body oscillation collapse into a single
number is reached. Thus the distribution, turbulent diffusion, and frequency close to the natural frequency of the body (see Fig. 2). This
decay of vorticity and the interaction between the wake and the phenomenon is known by various names: lock-in, locking-on, syn-
boundary layers cannot be subjected to numerical simulation without chronization, hydroelastic or fluid-elastic oscillations, wake capture,
recourse to some heuristic turbulence models and inspired fore- self-controlled or self-excited oscillations, etc. 4
sight.
The representation of the wake by clouds of point vortices or
4
discretized spiralling sheets (see, e.g., [10,26,49,53]) is not immune A vortex-excited oscillation is actually a forced one having a self-excited
to scaling problems. In fact, not a particular Reynolds number but character also to some degree due to lift force amplification through nonlinear
interactions. By definition, self-excited oscillation is one where the alternating
only a particular flow regime may be specified, depending on the force that sustains the motion is created or controlled by the motion itself; when
separation criteria used. the motion stops, the alternating force disappears. For the phenomenon dis-
Three-Dimensionality. As noted earlier, even a uniform flow cussed herein the alternating lift force does not disappear when there is no os-
about a stationary cylinder exhibits chordwise and spanwise varia- cillation. In fact, when there is no alternating force (a flat plate normal to the
flow) there is no lock-in. Apparently, the often-used definition of "self-excited
tions. These three-dimensional effects may play a major role in the oscillation" is a misnomer.

244 / VOL. 46, JUNE 1979 Transactions of the ASME


The facts which have emerged from two decades of work on vor- of the correlation length is responsible only for about 10 percent of
tex-induced oscillations may be summarized as follows: the lift increase, i.e., it is not the major factor for the lock-in phe-
nomenon.
1 When a body is close to its linear resonance conditions it can
13 A hysteresis behavior may exist in the amplitude variation and
undergo sustained oscillations at a frequency close to its natural fre-
frequency capture depending on the approach to the resonance
quency [55]. That a cylinder should be excited at its natural frequency
range—whether from a low velocity or from a high velocity [54]. No
when the frequency of the exciting force is equal to its natural fre-
universal behavior is noted and the reasons are not clear. The jump
quency is not surprising. But that the phenomenon encompass a rahge
condition (double amplitude response) may originate in the fluid
of ±25 to 30 percent of the natural vortex-shedding frequency and
system, and therefore in the lift force, and not in the cylinder elastic
that the vibration and vortex-shedding frequencies lock together and
system [2]. The "combination-oscillation" model [68] (based on the
control the shedding process are surprising.
assumption that the lift oscillator has two components, one at the
2 The interaction between the oscillation of the body and the cylinder frequency, representing lock-in, and the other approximately

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


action of the fluid is nonlinear (e.g., [56, 57]). at the frequency of vortex shedding for the stationary cylinder) gives
3 The vortex shedding does not necessarily result in an alternating rise to jumps between two stable oscillation amplitudes but only at
transverse force. There must be a suitable afterbody, and hence an the ends of the lock-in range and not in the middle of the said range,
alternating lift force, and the shedding frequency should coincide with as found by Feng [54] (see Fig. 2). The jump may be the consequence
or be near the natural frequency of the body, or one of its harmonics, of a variable structural damping (dependent on the interaction of the
for sustained oscillations to occur. Thus the magnitude and occurrence ambient velocity and structure support, vibration velocity and/or
of sustained oscillations strongly depend on the lift coefficient of the amplitude, etc.). [69, 70] or of a nonlinear spring behavior [71]. It is
stationary body. also entirely possible that the jump condition originates in both the
4 The parameters Vr = V/fnD and K = m^/pD2 [58, 59] or R = fluid and the cylinder elastic system, including blockage effects [72]
m^/pD2CjJo [60] or Ar = (27rf)(TSo)2/pr are of major importance in and the vibrations of the test apparatus [54].
determining the amplitude of oscillations and the range of synchro-
14 The absolute value of the base pressure and the in-line or drag
nization for a given body [59-61].
force increase with the amplitude of oscillation [3,56,61, 72, 73]. For
5 The velocity range over which synchronization occurs increases a circular cylinder, Cpb increases as much as 60 percent for A/D = 0.29
with oscillation amplitude [62]. [72] . 6 For a D -shaped cylinder, Davies [24] found an increase of 70
6 The excitation range of cross-flow oscillations in air extends over percent relative to the stationary D -cylinder value.
4.75 < Vr < 8 and maximum amplitudes occur in the range 5.5 < Vr 15 The phase difference between the displacement and force
< 6.5 [3]. In water, the excitation range of the transverse oscillations signals increases almost linearly at first and then rapidly [54, 56].
can be increased to 4.5 < Vr < 10 with maximum amplitude falling
16 The mobility of the separation points is important but not
within the range of 6.5 < Vr < 8.
necessary for synchronization. The separation point on a circular
7 For a circular cylinder with large LID, synchronization begins cylinder moves over an arc whose magnitude depends on the fre-
when /o — fn and ends at about fo/fn ea 1.4. The maximum amplitude quency and amplitude of the oscillation. The magnitude of the sep-
occurs near the middle of this range. At the end of the lock-in range, aration-point travel reaches a peak when the cylinder frequency is
vortex-shedding frequency jumps to that governed by the Strouhal about 0.9 of the wake frequency [74, 75].
relationship, but the cylinder continues to oscillate at fc ca /„. This
17 The actual instantaneous value of the wake angle is greater
is true at both ends of the lock-in range [54] and shows that the re-
than that between the ambient flow velocity and the relative fluid
sponse is not a simple forced vibration at the exciting natural Strouhal
velocity [74,75].
frequency.
18 The overall wake width is practically unaffected in spite of the
8 For a cylinder with fixed separation points (e.g., a D-shaped lateral movement of the body that generates it. Koopman [62] inferred
cylinder) sychronization extends over a range of 0.96 to 1.1 of the from flow-visualization studies an apparent decrease in lateral vortex
resonant velocity (fo/fn from about 0.85 to 1.1) [24]. According to Feng spacing in the resonant wake of an oscillating circular cylinder. Davies
[54], the said range extends over a range of 5•< Vr < 7. Gowda [39] [24] found an about 20 percent increase for a D -shaped cylinder
found that the lock-in begins already when / 0 is about 80 percent of synchronized in a flow at much higher Reynolds numbers (7 X 103 <
fn and ends when /o — 2/„. The maximum amplitude occurs near the Re < 4 X 104, AID = 0.2). Griffin [67] found that the lateral spacing
end of the synchronization. of the vortices decreases with increasing amplitude and is unaffected
9 In synchronization, the slantwise vortex shedding is replaced by changes in frequency of oscillation.
by parallel vortex shedding [44,62].
19 Longitudinal spacing /„ does not change with the amplitude
10 The correlation length increases rapidly with amplitude [57,
of motion in the synchronization region but it varies inversely with
62-65]. The increase of the correlation length in smooth flow is much
the frequency ratio [77]. For a cylinder undergoing forced transverse
larger than in turbulent flow [66]. In smooth flow the correlation
oscillations with the frequncy /,., /„ decreases from its value a t / c = /„
length is estimated by numerical extrapolation to increase from about
as fdfo increases and vice versa. In other words, wake contracts for
3.5D to 40C for Re = 1.9 X 104 in the range 6 0.05 < A/D < 0.1. In
fdfo > 1 and expands for fdfo < 1.
turbulent flow, it is again estimated to vary from about 2.5D to 10D
20 The formation length // exhibits a similar variation with fdfo-
in the same A/D range. The rate of increase is steeper than linear but
However, unlike the longitudinal spacing, the formation length de-
does not show any abrupt change which would indicate a sudden de-
creases systematically with increasing amplitude of vibration [76].
velopment of the lock-in once a threshold amplitude is achieved.
For a given fdfo, the vortices in the near wake go through a spacing
11 For bodies, with mobile or fixed separation points, undergoing adjustment from an amplitude-dependent spacing to an amplitude-
sustained oscillations, the vortex strength is increased [24, 67]. This independent longitudinal spacing.
could be either or both due to increased rate of vorticity flux or due
21 Forced vibrations may differ from the naturally occurring
to less destruction of vorticity in the near wake. The growing vortex
vortex-induced vibrations since the latter depend, to some extent, on
on a vibrating cylinder seems to roll up more quickly [24].
the history of the fluid motion. History effects may be particularly
12 The transverse force needed to excite a cylinder to large am- important in cases where hysteresis occurs. Driven models overshadow
plitude oscillations is far greater than that exerted by vortex shedding,
i.e., the transverse force is amplfified (e.g., [54, 56, 61]). The increase

6
Fixing separation at 9, = ±70 deg by a tripping wire on a circular cylinder
5
The generality of these values is not certain since in Novak and Tanaka's yielded a nearly constant Cpb at - 1.39 for AID = 0.1 for all values of fdfo from
[66] experiments the cylinder length (between the end plates) was only 12 di- 0.4 to 1.3 [72]. The reason for this anomalous behavior is not clear. Confirmatory
ameters. experiments are needed.

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 245


the intricate interaction between the evolution of the motion and the prevents unlimited growth. The fifth term provides the feedback from
variations of the pressure distribution. the body motion to the fluid motion and hence to the lift force.
22 Frequency demultiplication or frequency division (existence Equation (2) is not the only form which could provide the desired
of the natural vortex-shedding frequency in the state of synchroni- variation in C/,, but it is the simplest.
zation in the wake) may occur when the forced cylinder frequency fc Of three parameters (a, y, and B), two must be chosen to provide
is an integral multiple of the shedding frequency 7 [56, 57, 72, 77]. the best fit to the data. This is not too unusual and all other models
23 In-line oscillations occur within two adjacent regions [79-86]. require some retrofitting to experiment. The ratio a/y is related to
The first is in the range of 1.25 < Vr < 2.5, maximum amplitudes oc- the amplitude of the steady-state oscillation of the lift force by
curring at Vr =* 2.1. The second region extends from Vr ^ 2.7 to Vr CAO = V W 3 7 .
=* 3.8 with maximum amplitudes at Vr =* 3.2. The first instability Experiments have shown that in large-amplitude, steady-state,
region is accompanied by symmetric vortex shedding (as if the flow vortex-induced oscillation, the displacement and the exciting force

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


started impulsively from rest at each cycle) and the second by alter- have nearly sinusoidal forms and oscillate at the same frequency o>Ct
nate vortex shedding. close to wn, with a phase angle 0. Thus
Wake-Oscillator Models. Several mathematical models have
been proposed [87-91]; (see also [2, 55, 71, 92] for additional models) xr = xrsmQcT and CL = CL sin (£2cr + 0) (7a, b)
in an attempt to simulate and/or explain some of the experimental Substituting equations (7) in equation (2) and equating to zero the
results just summarized. These models do not include the analysis resulting coefficients of sin QCT and cos flcT, Hartlen and Currie [87]
of the flow field and the fluid-mechanical arguments invoked in their obtained
evolution are not altogether convincing. Thus their worth should be
measured no so much by their capacity to obtain functional relations (1 - Q c 2 )x r = OOSVCL cos 0 (8)
among significant parameters that lead to the basic understanding 2
2ffi c x r = a 0 fio C L sin 0 (9)
of the phenomenon but rather by their ability to produce results which
are qualitatively similar to those obtained experimentally. Likewise, substituting equations (7) in equation (2), dropping terms
The most noteworthy among the oscillator models is the one pro- arising from CL3, and using equations (8) and (9), one has
posed by Hartlen and Currie [87] where a van der Pol-type soft non-
2aoBf,fl c 2 l- 1 / 2
linear oscillator for the lift force is coupled to the body motion by a
linear-dependence on cylinder velocity. This model is based partly
on a suggestion by Birkhoff and Zarantonello [23], mentioned earlier which shows that fic < An or fc < f0.
in connection with the estimation of the Strouhal number, and by Hartlen and Currie applied their analysis to a comparison with
Bishop and Hassan [56] in connection with their now classic experi- Jones, et a l , [17], data (Re = 3.6 X 10B to 1.9 X 107, xr = 0.08) with
ments with oscillating cylinders in uniform flow. The model has its good qualitative results. However, some doubt has been raised [8] as
roots in mechanics and electricity rather than in the equations of fluid to whether the oscillations in Jones, et al., experiments had reached
motion. synchronization and whether an amplitude of xr = 0.08 was suffi-
The pair of equations 8 which result from this concept are ciently greater than the threshold amplitude [44, 62] necessary to
2
bring about synchronization. Furthermore, Jones, et al., data did not
xr + 2f s x r + xr = a0a0 CL (2) show an increase in drag 10 in the transcritical range of Reynolds
CL ~ ««oCt + (y/U0)CLs + Q02CL = Bxr (3) numbers.
Hartlen and Currie also used Ferguson's [63] data for the circular
in which 9 cylinder and obtained the results shown in Fig. 3. These results are
in qualitative agreement with those shown in Fig. 2. There are, how-
0>„ = 2irf„ = Vkjm , fin - fo/fn, fie = fc/fn, ®u = fulfn
ever, some important quantitative differences:
T = coJ, So = fgD/V, xr = x/D, xr = AID, xrm = (A/D)max 1 The phase angle between the exciting force and response vary
2
continuously, as does o>c, CL, and xr. There is no oscillation hystere-
xr = dxr/dr, CL = 2FL/(pLDV*), a0 = pLD 2 / 0 2 / ( 2 m S 0 V Q 0 ) sis.
Ar = f«/o 0 = (2irf,)(irS 0 ) /p r , Pr = P/ps = 27r 3 a 0 S 0 2
2 2 In Feng's data (Fig. 2) the cylinder is seen to continue to oscillate
at wc =* w„ on both sides of the region outside the lock-in range. This
«o = 2ir/o, i»v = 27r/u, UIC = 2ir/ c (4) feature is not presented in Fig. 3.
3 The force and response maximums occur in Fig. 2 at about the
The parameters a and y are the van der Pol coefficients and B is
same Vr- value. In Fig. 3, CL remains constant over a broad range of
the interaction parameter. Finally, f„ represents the material damping
synchronization, reaching a maximum at a Vr -value considerably
factor for the elastic system. There is no other damping imposed on
smaller (relative to the width of the lock-in range) than that for xr.
the body.
The explanation of these discrepancies and the relationship between
In equation (2), first and fourth terms generate a simple harmonic
the particular changes in the wake and/or the elastic system which
oscillator of normalized frequency S2n> the second term (the so-called
trigger the hysteresis remained unresolved.
negative lift damping) provides the growth of CL, and the third term
Griffin, et al., [61], in an attempt to improve the Hartlen-Currie
model, introduced a slightly modified version of equation (3), con-
taining several empirical parameters, in a form more amenable to
7
For example, x = 2fc~1/2 sin at is the exact solution of* - 2aa>(l - bx2)x + comparison with experiment. This version of the oscillator model
oi2x = — 4ab~1/2 sin (3a>£) and is a special case of the phenomenon called "fre- suffers from the same inadequacies as those of Hartlen and Currie.
quency demultiplication" by van der Pol [78]. Additionally, the model has been criticized because it is based on
8
The particular form of equation (3) is more correctly identified as the
Rayleigh equation given by damping measured in a still fluid. Nevertheless, the model's predic-
tions are within the data scatter band.
y-rt+ (%W + y =0 (5) Additional attempts to improve the oscillator model have been
where n is a parameter. The differentiation of equation (5) with respect to t and made by Landl [97] and Szechenyi [98]. Landl introduced a nonlin-
the substitution of u = y yields earity of fifth order in the damping term in the lift-oscillator equation.
0, — ixii + iiu2u + u- 0 (6)
which is known as the van der Pol equation. It has been subjected to consider-
able investigation particularly through the use of the Kryloff-Bogoliuboff
10
(K-B method); (see, e.g., [93]). A drag increase of about 16 percent would have been expected on the basis
9
Additional parameters appearing in (4) will be referred to later. of experiments at subcritical Reynolds numbers [57,73,94-96].

246 / VOL. 46, JUNE 1979 Transactions of the ASME


1.01

(b)

0.99

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


0.80 0.85 , ^ 0.90

Fig. 3 Response characteristics as predicted by the Hartlen-Currie model; from Hartlen and Currie [87].

The results of his model predicted, as it was designed to, most of the amplitude.11 However, the results of equation (11) are quite sensitive
observed features. He was able to explain "hard excitation" in Feng's to the Strouhal number, particularly for Ar smaller than about 0.4.
experiments (see, e.g., [99] for a detailed discussion of soft and hard Iwan [90] extended his model to flexible structures. Subsequently,
excitation). Szechenyi's [98] model is closer to the original proposal Blevins and Burton [91], following a somewhat different approach,
of Birkhoff [23]. He assumed a fictitious, symmetric, oscillatory airfoil used random vibration theory and represented the vortex-induced
attached to the cylinder and examined its oscillation under the action forces by a narrow-band, stationary random process. This model ac-
of a periodic lift force through well-known aerodynamic consider- commodates for the variation in correlation length and accounts for
ations. The details of the analysis are rather cumbersome and it is the dependence of the vortex shedding upon the amplitude of oscil-
difficult to separate the assumptions from the predictions. However, lation and the L/D ratio through judicious selection of three con-
he did obtain good correlation with the experimental values of stants.
Ci(rms) and the phase angle. Further developments of the oscillator models dealt with the ex-
Iwan and Blevins [89] arrived at the Hartlen-Currie model through planation of the occurrence of two stable amplitudes of oscillation for
considerations based on the vortex street. However, their fluid- the circular cylinder in the range of flow velocities producing syn-
mechanical arguments are not altogether convincing and shed no more chronization. Currie, et al., [71], attributed the observed phenomenon
light on the subject than those of Hartlen and Currie (for an appli- to the slight nonlinearity of the mechanical spring, as suggested by
cation of the Iwan-Blevins model and more favorable opinions see the behavior of hard and soft springs [99]. They have modified
[100]). equation (2) so that the restoring force term xr was replaced by xr (1
The maximum amplitude at perfect synchronization is given by - txr2) which represents, for e « 1, a slightly nonlinear mechanical
Iwan and Blevins as [1, 89]. spring. They have shown that such small nonlinearities in the re-
storing force could produce a double-amplitude response in the res-
0.44 7 4.52 S0 1/2
onant range of the two frequencies involved. Since nonlinearities of
(A/DU 2 0.3 + - 2 (ID
Ar + 11.94 So Ar + 11.94 So the order of 2 percent produced double-amplitudes, they have con-
in which y is the dimensionless mode shape factor and is given by cluded that this might be the explanation of the phenomenon. Sub-
[1] sequently, Oey, et al., [68], proposed an alternative theory based on

j\2(.y)dy
1/2
11
7 = tmax(y/L) (12) Griffin, et al., [61], least-square fit to the existing data is given by
J^V(y)dy (A/D)mM = 1.29T/[1 + 0.43Ar*)-3-36 (13)
Sarpkaya [73] obtained
where ^(y) is the mode shape at each spanwise point y. y = 1.0 for a
spring-supported rigid cylinder, 1.305 for the first mode of a cantile- {A/D)mSL% = 0.327/[0.06 + A,.*]-"2 (14)
ver, and 1.155 for a sinusoidal mode. The results of equations (12)-(14) will be compared later in connection with
Equation (11), a least-squares fit [61] to the existing data, and the discussion of experimental results. Ar* in equation (13) includes both the
Sarpkaya's [73] formulation yield similar results for the resonant structural and still-fluid effects.

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 247


equations (2) and (3), assuming constant £„ and spring constant. This flow-field experiments as to whether a highly nonlinear-equation
so-called "combination-oscillation" model is based on the assumption system is being conditioned to suit qualitatively the observations or
that the lift oscillator has two components, one at the cylinder fre- whether one is gradually approaching toward the modeling of the lift
quency, representing lock-in, and the other approximately at the force behavior in flow about a cylinder undergoing synchronized os-
frequency of vortex shedding for the stationary cylinder. Redefining cillations.
T as T = o>t and xr and CL as Finally, a comment must be made on a question often raised in
connection with the oscillator models, namely, "why an equation of
xr = xr sin r (15) the van der Pol kind should at all describe the fluid-body interaction?"
C L = C F sin (QFT/UC + <t>F) + CH sin (T + <I>H) (16)
First, as evidenced from the foregoing, it does not quite describe the
said interaction. Second, the qualitative agreement between the model
and using equations (2) and (3) as before, they were able to show that predictions and measurements (specifically, the limit cycle behavior

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


there exist two stable solutions to the oscillator equation. The dou- and frequency entrainment) is a preconditioned reflex of the van der
ble-amplitude response is shown to be associated with the physical Pol oscillator. In fact, as noted by van der Pol [78], there are many
system jumping from one of these solutions to the other as the fre- other instances of relaxation oscillations such as a pneumatic hammer,
quency detuning is continuously varied. Oey, et al.'s, [68], explanation the scratching noise of a knife on a plate, the waving of a flag in the
also incorporated a hysteresis effect for sufficiently small mechanical wind, the humming noise of a water tap, the motion of sleeping
damping. The results were taken as further evidence of the validity flowers, the beating of the heart, etc., which may be described quali-
of equation (3). tatively by a van der Pol-type equation. Evidently, the gross features
The combination-oscillation model gives rise to jumps between two of the vortex-induced oscillations fall in the same general category.
stable oscillation amplitudes only at the ends of the lock-in range and Added Mass, Damping, and Dynamic Response Measurements.
not in the middle of it as found by Feng [54]. Thus, Oey, et al., model There are some fundamental questions in the current approaches to
left the hysteresis problem unresolved while providing a possible empirical correlations of the dynamic response and to modeling of the
explanation for the amplitude and frequency modulation at the ends vortex-induced oscillations which have not been adequately dealt with
of the lock-in range. by any researcher. These questions stem from the use of such concepts
Having explored the higher-order nonlinearity in CL [70, 97], 12 as still-water added mass and fluid damping for the case of flow with
nonlinearity of the spring [71], and the combination-oscillation with vortex shedding. As they are presently used, added mass and fluid
constant fs and k [68], researchers turned their attention to the pos- damping are quantities which are defined only for the zero flow con-
sible nonlinearity of the material damping constant fs in equation (2). dition and they are, in fact, measured under these conditions (e.g.,
Wood and Parkinson [69] and Wood [70] noted that the effective by plucking excitation is still fluid). There is no theoretical basis for
structural damping in Feng's [54] data was not constant, but varied assuming that these quantities should play the same roles in the case
with flow velocity and cylinder amplitude. They have approximated of flow with vortex shedding.
the effective structural damping by an empirical relation of the Some wake oscillator models require that both the still-fluid added
form mass and damping be added to the "structural" part of the equation.
Certain other models require only that the added mass be added to
fs = ftd + FSL<?Xr) (17) the structural element. Both of these approaches are fundamentally
which provides a dependence of system damping on the force acting in error and both require more information for the structural system
on the cylinder and the cylinder displacement. 13 The modified than should be required. Hence, rather than arbitrarily introducing
equation governing the cylinder response then became these quantities in the structural part of the equation describing i he
model, the "fluid" part of the model should be constructed from -lie
xr + 2 rod + FQ,Q2xr)xr + xr = a 0 fio 2 C L (18) outset so as to include all fluid dynamic effects.
The analysis of equations (3) and (18) by standard methods for weakly There is no reason in the laws of fluid dynamics that the added mass
nonlinear oscillations [93,101] have demonstrated at least in principle, should be the same for a body oscillating, at its natural frequency, in
the system's ability to generate multiple amplitudes in xr, CL, 4>, and a fluid otherwise at rest with that oscillating in a fluid in motion (for
fic for a considerable band of Oo-values within the lock-in region and additional discussion see [73,96,102,103]). Experiments have shown
realistic system behavior outside the lock-in range (combination- that [102, 103] the added mass coefficient of a circular cylinder os-
oscillation form of the solution). Although the inclusion of a variable cillating in still liquid fluid is slightly under unity (the inviscid-flow
damping term has resulted in some improvement of the oscillator value being exactly 1.0) for A/D less than about unity. This does not
model within the lock-in range, there remained significant difficulties prove that it remains constant for all types of fluid motion. In fa<'l,
with the amplitudes of xr and CL and with trends in the phase angle added mass depends on the type of motion of the body or of the fluid
for flc > 1. about the body; proximity of other bodies, free surface, etc., and time.
It is not always possible nor advisable to determine the instantaneous
On the basis of the foregoing one would conclude that the oscillator
value of the added mass. In short, it would be wrong in general to
model with a variable &> a constant ox variable k, and the combina-
equate the two flow situations where in one case the cylinder oscillates
tion-oscillation modification, and the suitable values of CLo and a and
with amplitudes of A/D smaller than unity in still fluid and the other
B [or 7 and B, see equation (3)] could predict qualitatively all of the
case where the cylinder oscillates at similar amplitudes in a transverse
observed features of cylinder oscillation in Feng's data. The important
(or in-line) direction in steady flow. The former does not involve
distinction between the variable—r model of Wood and Parkinson
vortex shedding (xr < 0.7) whereas the latter is accompanied hy
[69,70] and the variable—CL damping model of Landl [97] and Wood
complex separation and vortex-shedding phenomenon even though
[70] and combination-oscillation model of Oey, et al., [68], is that while
the amplitude of oscillation is of comparable magnitude. However,
the former assumes that the jump condition originates in the cylinder
as will be seen shortly, the perfect lock-in state appears to be an ex-
elastic system, and therefore not with CL (as originally argued by
ception.
Parkinson [2]), while the two latter models assume that it originates
in the fluid system, and therefore with CL- Furthermore, all models The in-phase and out-of-phase components of the transverse forvr
assume the validity of the oscillator model. In any case, it is difficult have been measured by Sarpkaya [73, 96] and Mercier [104] H with
to decide without further guidance from laboratory and numerical cylinders subjected to forced oscillations. Freely oscillating cylinder
measurements were made by a number of researchers and the forv.e
coefficients have been calculated or estimated through direct or in-
12
Wood [70] has shown that the application of a combination-oscillation form
of solution to Hartlen and Currie's original model, and the extension to a sev-
14
enth-order nonlinearity in CL in equation (3) are unproductive. , Mercier's measurements are not quite extensive and precise. The torn'
13
Wood [70] chose F = 4 on the basis of Feng's [54] data. components can only be estimated from his data.

248 / VOL. 46, JUNE 1979 Transactions of the ASME


1.0 "


0.5 -
? A/D = 0 . 5 0

?'*
<J" O •*- _2
7
•y v.
'"$,• A/D = 0 . 5 0

•H,

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


-0.5

• r

— 1.0 -
L*
Fig. 4(b) Variation of the drag coefficient as a function of V, for AID = 0.5
[96]

spring constant or other structural properties and mode shape (see,


e.g., [1]).
The out-of-phase component of the force changes sign from nega-
tive to positive as the vibration amplitude is increased from A/D =
0.75 to 1.03 [73]. This is a convincing evidence of the self-limiting
Fig. 4(a) Variation of the inertia coefficient as a function of vr for nature of the vortex-excited oscillations. Mercier [104] also noted a
AID = 0.5 [96] similar change in the sign of Cdh as A/D was increased to 1.3.
The problems associated with the determination of the damping
ratio of a structure in vacuum, f„; in still fluid, fs/ (fs/ = £» + f/); and
in flowing fluid, at resonance, f// (may or may not include f s ), are far
direct measurements of the transverse force [58,106-108]. Sarpkaya more complex and inherent in choosing the appropriate damping ratio •
[73, 96] decomposed the lift-force coefficient as for all vibrating structures. Experiments in vacuum are difficult to
perform and slight changes in support conditions may lead to large
CL = Cmh sin QCT - Cdh cos QCT (19) differences. Experiments in still fluid are relatively easier but the
results are not easy to interpret since f/ is a nonlinear function of Re<„
where Cmh and Cdh represent the in-phase and out-of-phase, Fou- wD2/v and A/D, For a circular cylinder undergoing harmonic oscil-
rier-averaged, transverse force coefficients.16 Figs. 4(a, 6) show the lations in still fluid, separation occurs at about A/D = 0.2 and the
variation of Cmh and Cdh as a function of V/fcD for A/D = 0.5. Data vortex shedding at about A/D = 0.7 [102]. An approximate theoretical
for other values of A/D are given in [73, 96]. In the lock-in region analysis of the problem is possible only for very small amplitudes (A/D
(V/fcD = 5± in Fig. 4) Cdh is negative for A/D smaller than about < 0.1) [110, 111]. As will be demonstrated shortly, fs may not be as-
unity and thus energy is transferred to the cylinder. Cmh decreases sumed to be equal to fs/ (measured in still air), particularly for small
rapidly with increasing V/fcD, becomes nearly equal to unity at V/fcD values of Re„.
= 5, and then becomes negative. Evidently, the use of Cmh = 1 (for a
cylinder) in off-resonance conditions is not justified. Batchelor [110] has shown that for small values of A/D (less than
about 0.05) f/ (fluid damping alone) for a circular cylinder is given
King [109] conducted a series of experiments in still water (by
by
plucking a cantilevered cylinder) and also in flowing water at the
fundamental mode of the flow-excited vibrations. From a comparison
(22)
of the measured frequencies he concluded that the still-water and
flow-excited frequencies are virtually equal, i.e., the added mass is
As A/D increases, the effect of separation and hence the pressure drag
unaffected by streaming flow. Apparently, this conclusion is valid only
increases. Experiments have shown that [102, 103, 112] Cop —
at the resonance condition. In general the matter is more complex.
1.25A/D for AID < 1.6 and Re„, < 10". Thus noting that
The change in Cmh may be regraded not only as a change in mass m
but also as a change in the restoring force because of the unique phase 8 pA
relationship among these three quantities. For sake of simplicity, the Cop (23)
resonance natural frequency in liquid fluid, fnw, may be related to that
in still air, /„, by which is based on equivalent energy dissipation per cycle, and com-
bining with the viscous-dissipation contribution given by equation
fnlfnw = V(/n + Am)/m (20) (22), one has

where Am is the added mass. For a circular cylinder, this reduces p 4-\/2
f -
= +0.34(A/Z)) 2 (24)
to P:
It should be emphasized that f/ does not include the structural
fnlfnw = V I + Cmhplps = V I + 2C m / ,7r 3 OoS 0 2 (21)
damping.
Simple string equations exist for the calculation of /„ in terms of Skop, et al., [113], carried out experiments at Re,,, = 1720 and 6080
and proposed the following empirical equation (see also [114]):
15
Attempts [105] to interpret Cmj, cos 4>, Cmh sin $,Cn, cos 0, and Cdh sin 4.5
+ 0.454 (A/D - 0.4)ff (25)
<l> as fluid added mass, fluid damping, fluid inertia, andfluidexcitation are at o=-
best superficial and do not bring any additional understanding to the phe-
nomenon. i which H = 0 for AID < 0.4 and H = 1 for A/D > 0.4. The linear

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 249


2.0
2.5

2.0

1.5
IK
1.0 o
A

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


0.5
»A X

0.0
0 0.2 0.4 CL 0.6 0-8

Fig. 5 Various lift-coefficient data for oscillating cylinders, from King [3]

Fig. 6 Cylinder response as a function of the response parameter A r ; full


circles represent various data points as compiled by Griffin, Skop, and
dependence of f/ on AID for AID > 0.4 is questionable. The use of Ramberg [61]
equation (24) is recommended.
Equation (24) shows that, for suitable values of plps and Re^, {y may
be considerably larger than fs even for very small values of AID, (e.g.,
for f„ = 6.8 X 10" 4 , Re., = 780, plps = 0.02, and AID =* 0, f/ = 41 X Evidently, only for relatively large values of Ar and ao (a conflicting
10 -4).ie demand unless fs is very large) that xrm is uniquely determined by
Several facts emerge from the foregoing: Ar-11
1 Still-air damping may be considerably different from the actual The relative significance of the roles played by f„ f/, and f// may
damping. be further elucidated by writing
2 Still-air damping, determined with a very small amplitude, may
not be assumed to consistently characterize a flow parameter (e.g., Arsf = Ars + &rf = Ua0 + f//a 0 (28)
Ar or any other response parameter) for a vibration at a much larger
and combining with equation (24) to yield
amplitude.
3 Material damping may be determined by first vibrating the Ar'f = Ar> + 2 T T 3 S 0 2 [ 4 V 2 Re<,- 1/2 + 0.34(A/D) 2 ] (29)
structure in still air, at a relatively high Re<„ and at very small am-
plitudes (AID < 0.1), so as to determine fs/, and then subtracting f/ The last term in equation (29), representing Arf, is plotted in Fig. 6
through the use of equation (24). This procedure is simpler and more together with the experimental data (based on fa, obtained in still
reliable than experiments in vacuum. However, one must ascertain air at often unspecified amplitudes) and equations (11), (13), and
that f5 remains independent of the amplitude of vibration at ampli- (146). Evidently, for Rea = 6000, AID less than about 0.2, and Ar
tudes corresponding to the vortex-excited oscillation. larger than about 0.5, the contribution of A / to Ar is small and be-
comes insignificant as Ar increases. Thus the region of small ampli-
4 Finally, fs should be used in all equations and parameters
tudes or large A r 's is material-damping dominated. This is of course
characterizing the phenomenon.
the region where fa is mostly due to fS| i.e., the use of fa in lieu of f„
Among others, Vickery and Watkins [58] and Griffin, et al., [61],
does not change the correlation in Fig. 6. The region where Ar is
have attempted through the use of a relatively crude analysis, based
smaller than about 0.2 is dominated by fluid-damping. In this region,
on the balance of energy transferred to the structure and that lost
in particular, fa may not be assumed to be equal to £„. Fortunately,
through damping, that AID is uniquely determined by the so-called
as the experiments show, AID does not significantly vary with Ar for
"response parameter" or "damping ratio" Ar = f/W Historically, f
Ar less than about 0.2 or 0.3. Thus the use of fa in lieu of fs does not
was taken as fsf, measured in still air (occasionally in still water), and
materially affect the correlation even if fa is several times larger than
included fs, as previously noted. Equation (2) shows that the response
of the system is independently governed by fs and ao and that the said f..
equation cannot be expressed, in general, in terms of a single damping A third region, between the two regions cited in the foregoing, may
ratio Ar. be identified where the use of fa in lieu of fs leads to larger scatter in
The foregoing may be clarified further through a relatively more the data (see Fig. 6) since both f, and f/ acquire equal importance. The
rigorous analysis. Using equations (la) and (76), multiplying equation data reported in the literature are not always comprehensive enough
(2) with xr, assuming <?/, at resonance to be independent of the reso- to delineate the effect of fs let alone to assess the amplitude or fre-
nance amplitude (see Fig. 5), and integrating, one has quency dependence of fs.
Fig. 6 also shows that fa needed to yield, say, Ar = 0.05 in a flow-
iirlQc

X 2f s x r 2 dT
Jo
aoilcPCiXrdj (26)

which reduces to 17
Note that a 0 and Qe are not independent. Also note that as Ar -» 0, xrm ~
fio2Ci atflo2Ci,/(l - fie2)- Experiments have shown that the limiting value ofxrm ' 8
(A/D) max = x„ (27) between 1.0 and 1.5. Assuming xrm = 1.25, Qo = 1-05, and Ci - 0.56, one has
2
0 2\2- 1/2 a 0 /(l — Qc2) - 2.02. Assuming this value to remain invariant, equation (27)
2fi„ A r + reduces to (A//J) maI = 0.317/(0.062 + Ar2)~1/2. This is a slightly modified ver-
2Sicoo
sion of equation (14) and will be identified as equation (14b). For Ar > 1, Xrm
=a 0.31Ar-1. This result as well as equations (11) and (13) tend to overpredict
slightly the amplitude for Ar > 3 or xrm < 0.1. Reduced spanwise correlation
at such amplitudes may yield relatively smaller response, explaining in part
' This example corresponds to the values given in [106]. the behavior of equations (11)—(14).

250 / VOL. 46, JUNE 1979 Transactions of the ASME


induced resonant situation is considerably smaller than fs/ for iden- Table 3 Characteristics of the elastic system
tical values oiA/D and a<\. In other words, the effective fluid damping Xrm
Ijf at resonance may be many times smaller than fs/ or f/ and much (numerical
closer to fs (all assumed to be determined at resonance AID). This Set No. f« a0 A/ model)
happy coincidence is brought about by the nonlinear behavior of the
fluid-structure interaction which minimizes f// near perfect syn- 1 0.004 0.0080 0.5 0.75
2 0.040 0.0800 0.5 0.78
chronization. On either side of resonance, f// increases (more rapidly 3 0.004 0.0040 1.0 0.50
for Vr larger than about 6) (see, e.g., [106]) as would be expected on 4 0.040 0.0400 1.0 0.53
physical grounds. This is partly the reason for the failure of correla- 5 0.008 0.0050 1.6 0.25
tions and the predictions of the oscillator models at off-resonance 6 0.020 0.0125 1.6 0.27
7 0.040 0.0250 1.6 0.29
conditions. 8 0.080 0.0500 1.6 0.31
Flow-Field Models and Some Recent Work. A number of 9 0.040 0.0125 3.2 0.12

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


flow-field models have been proposed [2]. They are based primarily 10 0.080 0.0250 3.2 0.13
on the use of one or two concentrated point vortices. The position and
the time-dependent strength of these vortices are chosen to give good
agreement with some of the measurements. These models are highly 0
empirical and lack the mathematical and phenomenological sophis- N N
CD + iCL = i i r „ ( z „ - z 0 ) - (l/(z„ - zo) - iz0 £ r „ - vrz'o
tication necessary to relate the motion of the body to the changes in l l
the boundary layer, separation points, in-line and transverse forces,
(34)
and the evolution of the wake. The numerical solution of the Nav-
ier-Stokes equations for a transversely oscillating cylinder can provide The discrete vortex model developed for the stationary cylinder
basic information only for laminar wakes where Re < 300. Even then is used together with equations (33) and (34) to determine the fluid
the difficulty of describing the time-dependent conditions with a fi- forces acting on the cylinder. Equation (32) is integrated through the
nite-element grid and the computer time needed for a few cycles of Use of a fourth-order Runge-Kutta method. Otherwise, no additional
oscillations pose extremely difficult problems and confine the modifications are introduced into the procedures described in [10].
Reynolds number range to about 100. The boundary-condition In fact, the only specification made was the time at which the cylinder
problem may be overcome by introducing a transformation of the was allowed to respond to the fluid forces. Such a decision was nec-
Navier-Stokes equations to a noninertial frame of reference [115]. essary to avoid the coupling of two transient states resulting from the
Potential flow-field models can provide realistic information on impulsive start of the flow and the transient response of the cylinder.
practically every phase of the motion and allow one to conduct nu- First, the cylinder was allowed to shed a sufficient number of vortices
merical experiments or sensitivity analysis on any part of the flow field (by regarding the cylinder stationary for t* < 20) and then the cyl-
or on the variation of any parameter characterizing the elastic system inder was allowed to move. Calculations were repeated for each value
or the flow. In the following the main features of such a model will be of An for a given set of fs and an, characterizing the mechanical system.
presented. Each calculation for a given flo, fs, and an required about 3 hr of
Recently, Sarpkaya and Shoaff [10] developed a comprehensive computer time, on a CDC-6600 to reach a steady state at about t* =
discrete-vortex model based on potential flow and boundary-layer 400.
interaction, rediscretization of the shear layers, and the circulation Mathematically speaking, there are no particular restrictions on
dissipation to determine the characteristics of an impulsively started the selection of the numerical values of fs and ao other than the fact
flow about a circular cylinder. The evolution of the flow from the start that they should be physically realizable. From the point of view of
to very large times, lift and drag forces, Strouhal number, oscillations actual calculations, however, one must place restrictions on, fs and
of the stagnation and separation points, and the vortex-street char- ao to limit the computation time. The smaller the value of fs, the
acteristics have been calculated and found to be in good agreement larger is the amplitude of oscillation. Also, the larger the mass (smaller
with those obtained experimentally. The numerical model was then ao), the larger is the transient time during which T] asymptotically
applied [116] to the flow about a transversely oscillating circular approaches its final value. Consequently, a very small value of fs
cylinder. coupled with a relatively small value of ao requires many cycles of
The equation of motion for an elastically mounted and linearly oscillation before the cylinder reaches its final amplitude. This makes
damped cylinder of unit length is written in dimensional form as the computation very time-consuming. With these considerations in
mind, the fs and ao-values shown in Table 3 were used in the calcu-
mx + 2mwn^3x + mo>n2x = pV2cCL (30) lations.
The numerical results obtained with fs = 0.004 and ao = 0.0040 are
in which c is the radius of the cylinder. In terms of the following
shown in Fig. 7. Other fs and ao combinations yielded similar results
normalized variables:
[116]. Apparently, the results follow essentially the same trends as
i) = x/c, t* = Vt/c, r/ = dri/dt*, fio = u o / % , xr = ri/2 (31) those shown in Fig. 2. However, no hysteresis effect was found for any
fs, ao, and Qo combination within the lock-in range. Furthermore,
one has unlike in Fig. 2, UL reached its maximum at a V/wnD-value, (Vr =
2irV/anD = fio/So), smaller than that corresponding to xrm. Also note
r, + (27rrsS0/floH + US0/a0)2v = 27r2S02aoCL (32)
that Cz, (max) in Fig. 7 is about one half that reported in [2] 18 and
somewhat smaller than that given in Fig. 5. The latter difference may
The solution of equation (32) requires the instantaneous value of CL.
be partly due to the effects of the three-dimensionality of the flow in
It is evaluated through the use of the complex velocity potential
the laboratory experiments and partly due to the characteristics of
modified to take into account the motion of the cylinder. Thus, for
the potential flow model.
V = c = 1, one has
The predictions of the numerical model for all A r s values considered
are presented in Table 3. Two facts are of importance. First, the nu-
w(z) = -[(z- zo) + (1 - z0)/(z - z0)l + (i/2ir) £ Tn\Ln(z - zn)
l

- Ln[(z - zo) - l/(7^z~~o)]} (33) 18


CL is not plotted in Feng's [54] thesis. Few CL values were evaluated
in which zn denotes the instantaneous position of the center of the through the integration of the measured pressure distribution. The CL curve
in Fig. 2 has been reproduced from Parkinson's plot given in [2]. It appears that
cylinder; in, its velocity; Tn, the strength of the nth discrete vortex; Feng's data have been strongly influenced by the tunnel blockage effects. As
and z„, its position. The use of equation (33) together with the gen- will be noted later, Feng's CL data, as reported in [2] do not satisfy equation
eralized Blasius theorem [117,118] yields (9).

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 251


180

150
2.
- 120 CD
5-s = 0.004

a = 0.004 90
0
cr
S =0.205 60

1.4 30
cn
CD
0 o 0 •

1.2

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


0|
0.75
1.0
;.-tf.T.-"**-«'--",,:*r-"i'
Si .•«•••' 0.50
1.0

0.25

V 0.8
0.00
lo"1
0.6
-.25

-.50
X
r ,»•
0.4
-.75

0.2 Fig. 8 (a) Rate of change of circulation versus time; (ft) displacement versus
time

_1_ _L- a_ _i_ Jo.o


0.8 1.0 V/unD 1.2 1.4

Fig. 7 Response characteristics as predicted by the discrete-vortex


model

merical values of xrm are slightly larger than those predicted by


equations (11), (13), and (146). The use of fs, rather than fs/ (in air),
in the definition of Ar should have resulted in relatively smaller
# rm -values. Evidently, the small difference between Ar*f and A / in
the material-damping dominated region is more than offset by the
larger values of Cj, in the numerical model. Second, it appears that
xrm is not very sensitive to the individual variations of fs and ao for
the A r -values encountered in the calculations.
Representative computer plots of the instantaneous values of r ,
Xr, CDJ, CL, and the separation and stagnation angles, all at perfect
synchronization, are shown in Figs. 8-10 for f, = 0.08 and ao = 0.05.
Other flow characteristics such as the vorticity and velocity distri-
bution along the vortex sheets, the strength of the shed vortices, wake
angle, net circulation, etc., have been calculated. They will not be
presented here for sake of brevity.
The strength of the vortices in the near wake was found to be about
25 percent larger than that for a stationary cylinder [10]. The reasons
for this will become clear in the next section.
The amplitude of the separation point excursion, A6S, satisfactorily
compared, where possible, with that measured by Mei and Currie [74].
They have found A6S ai 9 deg for xrm = 0.13. The numerical model
yielded A0S = 10 deg for the same amplitude. Mei and Currie's ex- -1.0,
'0. 50. 100. 150. 200. 250. 300.
periments were confined to xrrn < 0.18.
The mean in-line force and its amplitude of oscillation were found Fig. 9 (a) Drag coefficient versus time; (b) lift coefficient versus time
to increase with increasing amplitude as in [3, 56, 61, 72, 73]. The
maximums of the numerically obtained mean values of the in-line drag
coefficient, CDI, are compared in Fig. 11 with those obtained experi-
mentally. A calculation based on the stationary-cylinder drag coef- The wake width far downstream was found to remain practically
ficient, CDO, and the apparent projected area for the oscillating cyl- constant at hv/D = 1.3. Likewise, the longitudinal spacing of the ,
inder, i.e., CDI = Cjjod + Zxrm), yields values that are almost equal vortices did not change with the amplitude of oscillation and remained s
to the maximum CDI- values shown in Fig. 11, (see also [65] for addi- at IJD = 4.96, in conformity with the experimental results [77]. Fi-:;:
tional correlations). Evidently, the oscillations increase the absolute nally, the numerical model has shown that the cylinder continues to :
value of the base pressure as confirmed by experiments [24, 72]. oscillate at a frequency close to its natural frequency at both ends of rf

252 / VOL. 46, JUNE 1979 Transactions of the ASMEf


150.

lea.

50.
(a)

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


1
g I I i i \ l i i i i i t i -L ' i 1 i i ' i
0. 50. 100. 150

100. 150. 200.


Fig. 10 (a) Oscillations of the separation points; (b) oscillations of the
stagnation point
Fig. 12 Vortex-shedding sequence for an oscillating cylinder

2.0

1.5

d 1.0
CD

0.5

0.0

11 Variation of the in-line drag coefficient with 1/Vr [96]

-.6
the synchronization region (see Fig. 7). The small amplitude oscilla-
tions exhibited nearly periodic modulations. Fig. 13 (a) Variation of the rate of change of circulation with time during
Mechanism of Synchronization. The sequence of events revealed the early stages of cylinder oscillation; (b) evolution of the cylinder re-
by the numerical experiments may be summarized as follows. Assume sponse
that /o ^ /„. As soon as the cylinder begins to move upward under the
action of the transverse force, the lower attached-vortex sheet lags
behind because of its inertia. Thus the wake axis passing through the
instantaneous center of the cylinder rotates clockwise (see Fig. 12, t*
= 189). In the meantime, the velocity relative to the cylinder rotates stantaneous value of the net circulation about the cylinder. The vortex
clockwise an angle dictated by the velocity of the uniform flow and spiral which is attached to the upper separation point does not move
the instantaneous velocity of the cylinder. The rotation of the velocity out of the cylinder's way because of its inertia and consequently comes
vector coupled with the rotation of the wake axis moves the upper closer to the approaching cylinder. These facts, namely, the increase
separation point further downstream and the lower separation point of circulation and the decrease of the relative distance between the
further upstream relative to those on a stationary cylinder, as sug- cylinder and the upper vortex spiral, confirm the observations made
gested by Marris [6]. r at the upper separation point increases rapidly by Davies [24] that "the growing vortex on vibrating cylinders seems
and decreases at the lower separation point. In other words, maximum to roll up more quickly and with a strength about 35 percent larger
and minimum values of r do not occur simultaneously, as on a sta- than that for a stationary cylinder."
tionary cylinder, and the f versus t* curves become skewed (see Fig. The phenomenon is, however, relatively more complex than the
13). The net effect of the foregoing is to increase the strength of the simplified introductory remarks would suggest. There are two other
vortex toward which the cylinder is moving and to increase the in- factors which enhance the strengths of the attached vortices. The first

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 253


-VORTEX SHEDDING 1.0

E< 0
0.5

0.0

-.5

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


-1.0
VORTEX SHEDDING
Fig. 15(a) Variation of lift coefficient, displacement, velocity, and accel-
BEGINS eration as a function of time for E < 0
Fig. 14 The relation between the lift force and cylinder displacement at
synchronization

1.0
is the decrease of the dissipation of circulation with amplitude. The
second is directly related to the difference between the period of os- 0.5
cillation of the cylinder, Tc = l// c =* l//n, and the time required for
the shedding of a vortex from a stationary cylinder, i.e., the time
difference (T c — To). Each growing vortex about an oscillating cyl- 0.0
inder continues to be fed circulation not only at a higher rate but also
for a longer time period. This further reinforces the strength of each
growing vortex. The link between the enhanced vortex strength and -.5
the locking-on of the vortex shedding frequency to the frequency of
cylinder oscillation is through the relationship between vortex
strength and the Strouhal number. The results of the stationary -1.0
cylinder analysis have shown [10] that the larger the strength of the 284. 28S 292. 294.
vortices in the near wake, the smaller is the Strouhal number and vice Fig. 15(b) Variation of lift coefficient, displacement, velocity, and accel-
versa [see also equation (1)]. The decrease of S, or the increase of T„, eration as a function of time for £ = 0
brings T„ closer to T c . This process is reinforced every succeeding
cycle until T„ a Tc =* Tn. It is this reinforcing action only that gives
this type oscillations the "self-excited" character. Otherwise, the
cylinder is literally plucked by each shed vortex at the most favorable 2. r
moment and forced to oscillate at its natural frequency. Consequently,
a more appropriate question is not why the vortex shedding locks onto
the natural frequency in the so-called synchronization region but
rather why does the vortex shedding not lock onto the cylinder fre-
quency for all flow velocities. For To > Tn, the vortex strength cannot
be reduced, relative to its stationary cylinder value, so as to increase
sufficiently the Strouhal number or /„ (at least for a circular cylinder)
in order to make T„ a* T„. Likewise, for T 0 « Tn, i.e., f0 » fn (end
of lock-in range), the strength of the vortices cannot be increased,
enough so as to bring /„ down to a value nearly equal to /„. This latter
situation is rather precarious because of the presence of unusually
strong vortices near the cylinder. An accidental change in the flow
velocity, cylinder frequency, or vortex strength can bring about a fairly
abrupt change in both the amplitude and the lift force. Arguments Fig. 15(c) Variation of lift coefficient, displacement, velocity, and accel-
similar to those presented above may be used to explain the existence eration as a function of time for E > 0
of a second resonance region when /„ = l/3/o [119].
The foregoing explanation dealt only with the kinematics of the nized oscillations, the increased strength of the vortices, growing
fluid motion. In the final analysis, it is the balance between the work ahead of the cylinder motion, increases T„. This, in turn, decreases
done by the fluid on the cylinder and the work done by the internal the phase angle between the force and the displacement which then
friction that determine the stability of the oscillations. The parameter increases the phase angle between the force and the velocity. Con-
which establishes the relationship between the friction mechanism sequently, the energy transferred from the fluid to the cylinder de-
and the given state of oscillation is the phase angle <f> by which the creases below that dissipated by the material damping. This sequence
excitation leads the response. It is given by [see equations (8) and of events decreases the amplitude to the level where the energy ex-
(9)] tracted from the fluid over a cycle is just balanced by that dissipated
2 by internal friction.
t a n 0 = 2f s fl c /(l - Oc ) (35)
For an unstable state where the energy balance does not hold true,
In the case of synchronization, the maximum velocity of the cylinder one has, after multiplying equation (32) with >j and integrating,
leads the maximum lift and the lift force reduces to zero shortly after
the cylinder reaches its peak amplitude (see Fig. 14). In other words, £ = 27T2S02ao CT CLi)dt* -(27rfSo/fio) f i,2dt* (36)
the lift force opposes the cylinder motion for a short time period as Jo Jo
T) begins to decrease from its maximum. If for any accidental reason The variation of ij, T/, T}, and Ci with time is shown in Fig. 15 for
?) were to be increased, while the cylinder was undergoing synchro- E > 0 (amplification), E = 0 (steady-state oscillation), and for

254 / VOL. 46, JUNE 1979 Transactions of the ASWlE


E < 0 (attenuation). Clearly, the phase angle between CL and r) is leads to an alternate vortex shedding for sufficiently large values of
smaller for E > 0 than that for E = 0. The case E = 0 yields V — Um and thus to the second type of instability. The origin of the
exciting force is the drag fluctuations resulting from the alternate
x r = ij/2 = fi02CL sin 4>/(2flcArs) (37) vortex shedding. Additional measurements (particularly forces and
which is identical to that given by equation (27), [see also equation pressures) are needed to properly quantify the in-line oscillations and
to devise methods to calculate the wave-current induced forces on
(9)].
piles.
The results presented in Fig. 7 satisfy equation (37), as they should.
Feng's [54] lift data [2] do not satisfy equation (37). As noted earlier,
his pressure measurements appear to be strongly influenced by the Closing Remarks
large tunnel blockage ratio in his experiments.
The modest objective of this paper has been to review the vortex-
The potential flow model just described produces most of the ex-
induced oscillations in a few specific fundamental cases. Those who

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


perimentally observed features of the lock-in phenomenon. The
desire a more comprehensive or utilitarian treatment of the entire
hysteresis or the double-amplitude response in the middle of the
subject will find the references cited in the text to be most helpful.
lock-in range could not be produced by holding fs and ao constant.
There are a number of relatively complex problems which deserve
One numerical experiment where fs was varied according to equation
considerable attention. Among these are the effects of cylinder in-
(17) yielded a double-amplitude response. Even though additional
clination, body-proximity, turbulence, shear, and the time-depen-
numerical experiments are necessary,19 it is tentatively concluded that
dence of the ambient flow. King [126] and Ramberg [127] have dem-
the double-amplitude response is primarily a consequence of variable
onstrated that the vortex-induced oscillations of circular cylinders
structural damping. Tunnel blockage may play a secondary role
are largely independent of the angle of inclination of the cylinder and
[72].
that the "independence principle" may be used, i.e., the component
In-Line Oscillations. This type oscillations in uniform flow have
of the free-stream velocity normal to the cylinder may be incorporated
attracted some attention partly because of the damaging vortex-
into any and all relations found for the normal incidence flow to de-
induced oscillations of trashracks in currents and of piles in tidal
scribe the flow over a yawed cylinder. Additional experiments at
waters and partly because of the need to determine the wave and
higher Reynolds numbers with direct force measurements are need-
current induced forces on offshore structures [79-86, 104, 120-
ed.
124].
Body proximity, as in cylindrical arrays in crossflow, is of great
Lock-in occurs when the in-line frequency approaches twice the
practical importance [129]. It appears that the tube-proximity effects
Strouhal frequency. The amplitude of the alternating force (drag
considerably reduce the exciting force and the width of the lock-in
fluctuations) and the response of the cylinder are an order of magni-
range [130, 131]. Furthermore, tube vibrations are often limited to
tude smaller than those in the transverse direction. The range of ob-
the first few rows [131] and to relatively small Reynolds numbers (Re
served frequencies and other features of the phenomenon have been
<1200). There seems to be no lock-in at higher Reynolds numbers.
noted previously. The most interesting feature of the phenomenon
is the occurrence of two instability regions separated by Vr =* 2. For There are no systematic experiments on the effect of turbulence
Vr < 2, usually a symmetric vortex shedding is observed [3, 81]. At on vortex-induced vibrations. It is possible that turbulence will in-
Vr e* 2, the vortex shedding changes to the commonly observed form crease vorticity diffusion, reduce vortex strengths, and lead to smaller
of alternate shedding, indicating a radical change in the cause of ex- structural response. The effect of flow shear [128] and time-depen-
citation and dynamic response. dence of the ambient flow [60] remain largely unexplored.
There is an extreme lack of information on the vortex-shedding
The symmetric shedding of vortices in the first instability region
from, and vortex-induced motions of, three-dimensional bluff bodies
may be explained as follows. Consider a relative displacement oiA/D
(for a comprehensive bibliography of the relevant studies see [132]).
= 0.25. For V/fcD = 1.57, V = Um where Um is the maximum velocity
Wind and gust induced vibrations of buildings and towers fall in this
of the oscillating cylinder. Thus, as the cylinder moves downstream
category [133]. Considerable work is needed to check the validity of
from its midposition, the relative velocity about the cylinder increases
the various modeling and scaling techniques; to determine the in-
from zero to V. During that period only two small vortices develop
fluence of the shape of the velocity profile, turbulence intensity and
symmetrically and remain attached since the relative displacement
scale, and of the surface roughness on structural response; to develop
of the cylinder is very small. It has been shown [125] in connection
analytical models; and to devise methods and devices for the reduction
with impulsively started or uniformly accelerating flows that the
[134] of flow excited vibrations.
relative displacement of the fluid should be sufficiently large for the
vortices to grow and become unstable. A relative displacement of 0.25 It appears that future progress will be based on both the availability
or so is not sufficient for the full development of the vortices let alone of reliable data and on the further development of analytical methods.
that of asymmetry. As the cylinder reverses its direction, the relative Experimenters need to obtain reliable data (preferably with freely
velocity increases and reaches a value of 2 V at the midposition of oscillating bodies vice driven models) and report all the parameters
oscillation. The vortices continue to grow symmetrically and give rise and the conditions of the experiments. The work of the past 20 years
to a drag overshoot [125] which is about 25 percent larger than the has shown that there is no such a thing as a simple vortex-induced
steady-state drag. The motion of the cylinder further upstream results vibration experiment. Model makers need to remember that the
in decelerating flow relative to the cylinder. The vortices move away structural and fluid dynamical aspects of the phenomenon are equally
from the cylinder under the influence of the uniform flow, symmet- important. A model which is not fluid mechanically satisfying will
rically at first and then becoming gradually asymmetrical. During a have little chance of survival. Application of the numerical methods
given cycle, there is not sufficient time for the onset of instability or such as the Marker-and-Cell technique [115] and the discrete vortex
for the asymmetric growth of the vortices. Apparently, the symmetric model [10, 116] can complement the experiments and heuristic
vortex shedding owes its existence to a proper range of V/Um -values models. More work is needed to increase their predictive potential.
and to relatively small amplitudes oiA/D. The origin of the exciting Perhaps the most exciting aspect of the entire phenomenon is that
force is the periodic drag overshoot resulting from the symmetric it is an unfinished structural and fluid mechanical problem which
growth and shedding of vortices. touches all branches of engineering.

In the region of V/fcD- values for which V > Um, the relative ve-
locity about the cylinder is always in the downstream direction. This
Acknowledgments
Part of the work reported here was supported by the National
Science Foundation and part by the Civil Engineering Laboratory of
19
One such numerical experiment requires 8 hr of computer time on CDC- the Naval Construction Battalion Center. The writer would like
6600. gratefully to acknowledge the support of these sponsors.

Journal of Applied Mechanics JUNE 1979, VOL.46 / 255


References a Stationary Cylinder in a Fluid Stream (Aeolian Tones)," University of To-
ronto, UTIA, Rev. No. 13,1958.
1 Blevins, R. D., Flow-Induced Vibration, Van Nostrand Reinhold 34 Gerlach, C. R., and Dodge, F. T., "An Engineering Approach to Tube
Company New York, 1977. Flow-Induced Vibrations," Proceedings of the Conference on Flow-Induced
2 Parkinson, G. V., "Mathematical Models of Flow-Induced Vibrations Vibrations in Reactor System Components, Argonne National Laboratory
of Bluff Bodies," Flow-Induced Structural Vibrations, ed., Naudascher, E., 1970, pp. 205-225.
Springer-Verlag, Berlin, 1974, pp. 81-127. 35 Chaplin, J. R., and Shaw, T. L., "On the Mechanics of Flow-Induced
3 King, R., "A Review of Vortex-Shedding Research and Its Application," Periodic Forces on Structures," Dynamic Waves in Civil Engineering, eds.,
Ocean Engineering, Vol. 4,1977, pp. 141-172. Howells, D. A., Haigh, I. P., and Taylor, C , Wiley-Interscience, London, 1971
4 McCroskey, W. J., "Some Current Research in Unsteady Fluid Dy- pp.73-94
namics," ASME Journal of Fluids Engineering, Vol. 99,1977, pp. 8-39. 36 Vickery, B. J., "Fluctuating Lift and Drag on a Long Cylinder of Square
5 Morkovin, M. V., "Flow Around Circular Cylinders. A Caleidoscope Cross Section in a Smooth and in a Turbulent Stream," Journal of Fluid Me-
of Challenging Fluid Phenomena," ASME Symposium on Fully Separated chanics, Vol. 25,1966, pp. 481-494.
Flows, Philadelphia, Pa., 1964, pp. 102-118. 37 Gerrard, J. H., "A Disturbance Sensitive Reynolds Number Range of

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


6 Marris, A. W., "A Review on Vortex Streets, Periodic Wakes, and In- the Flow Past a Circular Cylinder," Journal of Fluid Mechanics, Vol. 22,1965
duced Vibration Phenomena," ASME Journal of Basic Engineering, Vol. 86, pp. 187-196.
1964, pp. 185-196. 38 Wilkinson, R. H., Chaplin, J. R., and Shaw, T. L., "On the Correlation
7 Mair, W. A., and Maull, D. J., "Bluff Bodies and Vortex Shedding—A of Dynamic Pressures on the Surface of a Prismatic Bluff Body," Flow-Induced
Report on Euromech 17," Journal of Fluid Mechanics, Vol. 45, 1971, pp. Structural Vibrations, ed., Naudascher, E., Springer-Verlag, Berlin, 1974, pp
209-224. 471-487.
8 Berger, E., and Wille, R., "Periodic Flow Phenomana," Annual Reviews 39 Gowda, B. H. L., "Some Measurements on the Phenomenon of Vortex
of Fluid Mechanics, Annual Reviews Inc., Palo Alto, Calif., Vol. 4,1972, pp. Shedding and Induced Vibrations of Circular Cylinders," Deutsche Luft-und
313-340. Raumfahrt Forschungsbericht No. 75-01,1975.
9 Schlichting, H., Boundary-Layer Theory, 6th ed., McGraw-Hill, New 40 Stansby, P. K., "The Effects of End Plates on the Base Pressure
York, 1968. Coefficient of a Circular Cylinder," Aeronautical Journal, Vol. 87,1974, pp.
10 Sarpkaya, T., and Shoaff, R. L., "An Inviscid Model of Two-Dimen- 36-37.
sional Vortex Shedding for Transient and Asymptotically Steady Separated 41 Etzold, F., and Fiedler, H., "The Near-Wake Structures of a Cantile-
Flow Over a Cylinder," AIAA 17th Aerospace Sciences Meeting, New Orleans, vered Cylinder in a Cross Flow," Zeitschrift fur Flugwissenschaften, Vol. 24,
Paper no. 79-0281, Jan. 15-17,1979. 1976, pp. 77-82.
11 Gerrard, J. H., "The Mechanics of the Formation Region of Vortices 42 Surry, D., "The Effect of High Intensity Turbulence on the Aerody-
Behind Bluff Bodies," Journal of Fluid Mechanics, Vol. 25,1966, pp. 4 0 1 - namics of a Rigid Circular Cylinder at Subcritical Reynolds Numbers," Uni-
413. versity of Toronto, UTIAS Report No. 142,1969.
12 Strouhal, V., "Uber eine besondere Art der Tonerregung," Ann. Phys. 43 McGregor, D. M., "An Experimental Investigation of the Oscillating
und Chemie. New Series Vol. 5,1878, pp. 216-251. Pressures on a Circular Cylinder in a Fluid Stream," University of Toronto,
13 Rayleigh, L., "Acoustical Observations II," Philosophical Magazine, UTIA, Technical Note No. 14,1957.
Vol. VII, 1879, pp. 149-162 (also The Theory of Sound, Dover Publications, 44 Berger, E., "Unterdrueckung der Laminaren Wirbelstroemung und
New York, 1945. des Turbulenzeinsatzes der Karmanschen Wirbelstrasse im Nachlauf ernes
14 Goldstein, S., Modern Developments in Fluid Dynamics, Clarendon Schwingenden Zylinders bei kleinen Reynolds-Zahlen," Jahrbuch der WGLR,
Press, Oxford, 1938. 1964, pp. 164-172.
15 Roshko, A., "On the Development of Turbulent Wakes From Vortex 45 Tournier, C , and Py, B., "The Behaviour of Naturally Oscillating
Streets," NACA Technical Note 2913, Mar. 1953. Three-Dimensional Flow Around a Cylinder," Journal of Fluid Mechanics,
16 van Nunen, J. W. G., "Pressures and Forces on a Circular Cylinder in Vol. 85,1978, pp. 161-186.
a Cross Flow at High Reynolds Numbers," Flow-Induced Structural Vibrations, 46 Roshko, A., "On the Aerodynamic Drag of Cylinders at High Reynolds
ed., Naudascher, E., Springer-Verlag, Berlin, 1974, pp. 748-754. Numbers," paper presented at U.S.-Japan Research Seminar on Wind Loads
17 Jones, G. W., Cincotta, J. J., and Walker, R. W., "Aerodynamic Forces on Structures, University of Hawaii, Oct. 1970.
on a Stationary and Oscillating Circular Cylinder at High Reynolds Numbers," 47 Fage, A., and Johansen, F. C , "The Structure of the Vortex Sheet,"
NASA TR-R-300,1969. Philosophical Magazine, Vol. 7,1928, pp. 417-436.
18 National Physical Laboratory, "Strouhal Number of Model Stacks Free 48 Bloor, M. S., and Gerrard, J. H., "Measurement on Turbulent Vortices
to Oscillate," National Physical Laboratory Aero Report No. 1257,1968. in a Cylinder Wake," Proceedings of the Royal Society, London, Series A, Vol.
19 Roshko, A., "On the Wake and Drag of Bluff Bodies," Journal of 294,1966, pp. 319-342.
Aerospace Sciences, Vol. 22,1955, pp. 124-132. 49 Abernathy, F. H., and Kronauer, R. E., "The Formation of Vortex
20 Bearman, P. W., "On Vortex-Street Wakes," Journal of Fluid Me- Streets," Journal of Fluid Mechanics, Vol. 13,1962, pp. 1-20.
chanics, Vol. 28,1967, pp. 625-641. 50 Schmidt, D. W., and Tilmann, P. M., "Uber die Zirkulationsentwicklung
21 Calvert, J. R., "Experiments on the Low-Speed Flow Past Cones," in Nachlaufen von Rundstaben," Acustica, Vol. 27,1972, pp. 14-22.
Journal of Fluid Mechanics, Vol. 27,1967, pp. 273-289. 51 Williams, J. C , "Incompressible Boundary-Layer Separation," Annual
22 Simmons, J. E. L., "Similarities Between Two-Dimensional and Axi- Reviews of Fluid Mechanics, Annual Reviews Inc., Palo Alto, Calif., Vol. 9,1977,
symmetric Vortex Wakes," The Aeronautical Quarterly, Vol. 28, 1977, pp. pp. 113-144.
15-20. 52 Bloor, M. S., "The Transition to Turbulence in the Wake of a Circular
23 Birkhoff, G., "Formation of Vortex Streets," Journal of Applied Cylinder," Journal of Fluid Mechanics, Vol. 19,1964, pp. 290-304.
Physics, Vol. 24,1953, pp. 98-103, (see also: Birkhoff, G., and Zarantonello, E. 53 Fink, P. T., and Soh, W. K„ "Calculation of Vortex Sheets in Unsteady
H., Jets, Wakes, and Cavities, Academic Press, New York, 1957, Chapter Flow and Applications in Ship Hydrodynamics," 10th Symposium of Naval
XIII). Hydrodynamics, Cambridge, Mass., 1974.
24 Davies, M. E . , " A Comparison of the Wake Structure of a Stationary 54 Feng, C. C , "The Measurement of Vortex-Induced Effects in Flow Past
and Oscillating Bluff Body, Using a Conditional Averaging Technique," Journal Stationary and Oscillating Circular and D-Section Cylinders," MASc thesis,
of Fluid Mechanics, Vol. 75,1976, pp. 209-231. University of British Columbia, 1968.
25 Zdravkovich, M. M., "Smoke Observations of the Formation of a 55 Naudascher, E., ed., Flow-Induced Structural Vibrations, Springer-
Karman Vortex Street," Journal of Fluid Mechanics, Vol. 37,1969, pp. 4 9 1 - Verlag, Berlin, 1974.
499. 56 Bishop, R. E. D., and Hassan, A. Y., "The Lift and Drag Forces on a
26 Clements, R. R., "Flow Representation, Including Separated Regions, Circular Cylinder in a Flowing Fluid," Proceedings of Royal Society, London,
Using Discrete Vortices," AGARD Lecture series No. 86,1977. Series A, Vol. 277,1963, pp. 32-50.
27 Sarpkaya, T., "An Inviscid Model of Two-Dimensional Vortex 57 Toebes, G. H., "The Unsteady Flow and Wake Near an Oscillating
Shedding for Transient and Asymptotically Steady Separated Flow Over an Cylinder," ASME Journal of Basic Engineering, Vol. 91,1969, pp. 493-502.
Inclined Plate," Journal of Fluid Mechanics, Vol. 68,1975, pp. 109-128. 58 Vickery, B. J., and Watkins, R. D., "Flow-Induced Vibration of Cy-
28 Sacksteder, R., "On Oscillatory Flows," The Mathematical Intelli- lindrical Structures," Proceedings of the First Australian Conference, held
gencer, Vol. 1,1978, pp. 45-51. at the University of Western Australia, 1962, pp. 213-239.
29 Humphreys, J. S., "On a Circular Cylinder in a Steady Wind at Tran- 59 King, R., "Hydroelastic Model Tests of Marine Piles—A Comparison
sition Reynolds Numbers," Journal of Fluid Mechanics, Vol. 9, 1960, pp. of Model and Full-Scale Results," British Hydromechanics Research Associ-
603-612. ation (BHRA) Report RR-1254,1974.
30 Phillips, O. M., "The Intensity of Aeolian Tones," Journal of Fluid 60 Sarpkaya, T., "Dynamic Response of Piles to Vortex Shedding in Os-
Mechanics, Vol. 1,1956, pp. 607-624. cillating Flows," Proceedings of the Offshore Technology Conference, Paper
31 Prendergast, V., "Measurement of Two-Point Correlations of the No. 3647, Houston, Texas, 1979.
Surface Pressure on a Circular Cylinder," University of Toronto, UTIA Tech- 61 Griffin, O. M., Skop, R. A., and Ramberg, S. E., "The Resonant Vortex
nical Note 23,1958. Excited Vibrations of Structures and Cable Systems," Proceedings of the
32 El Baroudi, M. Y., "Measurement of Two-Point Correlations of Velocity Offshore Technology Conference, Paper No. 2319, Houston, Texas, 1975.
Near a Circular Cylinder Shedding a Karman Vortex Street," University of 62 Koopman, G. H., "The Vortex Wakes of Vibrating Cylinders at Low
Toronto, UTIAS, TN-31,1960. Reynolds Numbers," Journal of Fluid Mechanics, Vol. 28, 1967, pp. 601-
33 Etkin, B., Korbacher, G. K., and Keefe, R T., "Acoustic Radiation From 512.

256 / VOL. 46, JUNE 1979 Transactions of the ASME


63 Ferguson, N., and Parkinson, G. V., "Surface and Wake Flow Phe- 93 Kryloff, N., and Bogoliuboff, N., Introduction to Nonlinear Mechanics,
nomena of Vortex-Excited Oscillation of a Circular Cylinder," ASME Journal Princeton University Press, N. J., 1943.
of Engineering for Industry, Vol. '89,1967, pp.. 831-838. 94 Taneda, S., and Honji, H., "Determination of the Drag on Vibrating
64 Ramberg, S. E., and Griffin, 0 . M., "Vortex Formation in the Wake Circular Cylinders," Report of the Research Institute of Applied Mechanics
of a Vibrating, Flexible Cable," ASME Journal of Fluids Engineering, Vol. 96, Vol. 15, No. 50,1967, pp. 83-92.
1974, pp. 317-322. 95 Sedrak, M., "Widerstandsmessungen am Schwingenden Zylinder bei
65 Ramberg, S. E., and Griffin, 0 . M., "Velocity Correlation and Vortex kleinen Reynolds-Zahlen," Dissertation, Technical University of Berlin,
Spacing in the Wake of a Vibrating Cable," ASME Journal of Fluids Engi- 1970.
neering, Vol. 98,1976, pp. 10-18. 96 Sarpkaya, T., "Transverse Oscillations of a Circular Cylinder in Uni-
66 Novak, M., and Tanaka, H., "Pressure Correlations on a Vibrating form Flow, Part I," Technical Report No. NPS-69SL77071-R, 1977, Naval
Cylinder," Proceedings of the 4th International Conference on Wind Effects Postgraduate School, Monterey, Calif.
on Buildings and Structures, ed., Eaton, K. J., Cambridge University Press, 97 Landl, R., "Theoretical Model for Vortex-Excited Oscillations," Pro-
Cambridge, 1975, pp. 227-232, and 273. ceedings of the International Symposium on Vibration Problems in Industry,
67 Griffin, 0 . M., "Effects of Synchronized Cylinder Vibrations on Vortex Keswick, England, 1973.
Formation and Mean Flow," Flow-Induced Structural Vibrations, ed., 98 Szechenyi, E., "Modele Mathematique du Mouvement Vibratoire

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


Naudascher E., Springer-Verlag, Berlin, 1974, pp. 454-470. Engendre par un Echappment Tourbillonnaire," La Recherche Aerospatiale,
68 Oey, H. L., Currie, I. G., and Leutheusser, H. J., "On the Double- Vol. 1975, No. 5, pp. 301-312.
Amplitude Response of Circular Cylinders Excited by Vortex Shedding," 99 Timoshenko, S., Young, D. H., and Weaver, W., Jr. Vibration Problems
Proceedings of the 4th International Conference on Wind Effects on Buildings • in Engineering, 4th ed., Wiley, New York, 1974, pp. 177-186.
and Structures, ed., Eaton, K. J., Cambridge University Press, Cambridge, 1975, 100 Hove, D., Shih, W., and Albano, E., "Hydrodynamic Design Loads for
pp. 233-240 and 271. the OTEC Cold Water Pipe," Science Applications, Inc., Technical Report No.
69 Wood, K. N., and Parkinson, G. V., "A Hysteresis Problem in Vortex- SAI-79-559-LA, 1978.
Induced Oscillation," Proceedings of the 6th Canadian Congress of Applied 101 Minorsky, N., Non-Linear Oscillations, Van Nostrand, New York,
Mechanics (CANCAM), Vancouver, 1977, pp. 697-698. 1962.
70 Wood, K. N., "Coupled-Oscillator Models for Vortex-Induced Oscil- 102 Sarpkaya, T., "Vortex Shedding and Resistance in Harmonic Flow
lation of a Circular Cylinder," MASc Thesis, The University of British Co- About Smooth and Rough Circular Cylinders at High Reynolds Numbers,"
lumbia, Vancouver, Aug. 1976. Technical Report No. NPS-59SL76021, Naval Postgraduate School, Monterey,
71 Currie, I. G., Hartlen, R. T., and Martin, W. W., "The Response of Calif., 1976.
Circular Cylinders to Vortex Shedding," Flow-Induced Structural Vibrations, 103 Sarpkaya, T., "In-Line and Transverse Forces on Cylinders in Oscil-
ed., Naudascher, E., Springer-Verlag, Berlin, 1974, pp. 128-142. latory Flow at High Reynolds Numbers," Journal of Ship Research, Vol. 21,
72 Stansby, P. K., "Base Pressure of Oscillating Circular Cylinders," 1977, pp. 200-216.
Journal of Engineering Mechanics Division of ASCE, Vol. 102, No. EM4,1976, 104 Mercier, J. A., "Large Amplitude Oscillations of a Circular Cylinder
pp. 591-600. in a Low-Speed Stream," PhD Thesis, Stevens Institute of Technology, Ho-
73 Sarpkaya, T., "Fluid Forces on Oscillating Cylinders," Journal of boken, N. J., 1973.
Waterway, Port, Coastal and Ocean Division of ASCE, Vol. 104, No. WW4, 105 Griffin, O. M., "Vortex-Excited Unsteady Forces on Resonantly-Vi-
1978, pp. 275-290. brating Bluff Structures," NRL Memorandum Report 3820, Naval Research
74 Mei, V. C , and Currie, I. G., "Flow Separation on a Vibrating Circular Laboratory, Washington, D. C , 1978.
Cylinder," The Physics of Fluids, Vol. 12, No. 11,1969, pp. 2248-2254. 106 Griffin, O. M., and Koopmann, G. H., "The Vortex-Excited Lift and
75 Raudkivi, A. J., and Small, A. F., "Hydroelastic Excitation of Cylin- Reaction Forces on Resonantly Vibrating Cylinders," Journal of Sound and
ders," Journal of Hydraulic Research, Vol. 12, No. 1,1974, pp. 99-131. Vibration, Vol. 54,1977, pp. 435-448.
76 Griffin, O. M:, and Ramberg, S. E., "The Vortex-Street Wakes of Vi- 107 King, R., "An Investigation of the Criteria Controlling Sustained
brating Cylinders," Journal of Fluid Mechanics, Vol. 66,1974, pp. 553-576. Self-Excited Oscillations of Cylinders in Flowing Water," Proceedings of the
77 Taneda, S., "Experimental Investigation of Vortex Streets," Journal Symposium on Turbulence in Liquids, University of Missouri, Roola, 1977.
of the Physical Society of Japan, Vol. 20,1965, pp. 1714-1721. 108 Hartlen, R. T., Baines, W. D., and Currie, I. G., "Vortex-Excited Os-
78 van der Pol, B., "Frequency Demultiplication," Nature, Vol. 120,1927, cillations of a Circular Cylinder," University of Toronto Report UTME-TP-
p. 363. 6809,1968.
79 Sainsbury, R. N., and King, D., "The Flow-Induced Oscillations of 109 King, R., "The 'Added Mass' of Cylinders," The British Hydrome-
Marine Structures," Proceedings of the Institution of Civil Engineers, Vol. chanics Research Association (BHRA) Report TN-1100,1971.
49,1971, pp. 269-302. 110 Batchelor, G. K., An Introduction to Fluid Dynamics, Cambridge
80 King, R., Prosser, M. J., and Johns, D. J., "On Vortex Excitation of University Press, Cambridge, England, 1967, p. 357.
Model Piles in Water," Journal of Sound and Vibration, Vol. 29, 1973, pp. 111 Chen, S. S., Wambsganss, M. W., Jendrzejczyk, J. A., "Added Mass and
169-188. Damping of a Vibrating Rod in Confined Viscous Fluids," ASME J O U R N A L
81 Wootton, L. R., "Oscillations of Piles in Marine Structures," CIRIA O P A P P L I E D M E C H A N I C S , Vol. 43,1976, pp. 325-329.
Underwater Engineering Group Report No. 40,1972. 112 Verley, R. L. P., "An Investigation Into the Damping of Oscillations
82 Tatsuno, M., "Vortex Wakes Behind a Circular Cylinder Oscillating of a Cylinder in Still Water," River and Harbor Laboratory Report No.
in the Flow Direction," Bulletin of the Research Institute of Applied Me- STF60-A78049, Norwegian Institute of Technology, Trondheim-NTH, 1978.
chanics, Kyushu University, No. 36,1972. 113 Skop, R. A., Ramberg, S. E., and Ferer, K. M., "Added Mass and
83 Tanida, Y., Okajima, A., and Watanabe, Y., "Stability of a Circular Damping Forces on Circular Cylinders," Naval Research Laboratory Report
Cylinder Oscillating in Uniform Flow or in a Wake," Journal of Fluid Me- 7970, Washington, D. C , 1976.
chanics, Vol. 61,1973, pp. 769-784. 114 Dong, R. G., "Effective Mass and Damping of Submerged Structures,"
84 Crandall, S. H., Vigander, S., and March, P. A., "Destructive Vibration Lawrence Livermore Laboratory Report UCRL-52342, Livermore, Calif.,
of Trashracks due to Fluid-Structure Interaction," ASME Paper No. 75- 1978.
DET-63. 115 Hurlbut, S. E., Spaulding, M. L„ and White, F. M., "Numerical Solution
85 Sarpkaya, T., and Isaacson, M., Waves and Wave Forces—In Theory of the Time-Dependent Navier-Stokes Equations in the Presence of an Oscil-
and Application, to be published in 1979. lating Cylinder," ASME Fluids Engineering Division Symposium on Nonsteady
86 Griffin, O. M., and Ramberg, S. E., "Vortex Shedding From a Cylinder Flows—II, Numerical Solution of Nonsteady Flows, Book No. H00118,1978,
Vibrating in Line With an Incident Uniform Flow," Journal of Fluid Me- pp. 201-206.
chanics, Vol. 75,1976, pp. 257-271. 116 Sarpkaya, T., and Shoaff, R. L., "A Discrete Vortex Analysis of Flow
87 Hartlen, R. T„ and Currie, I. G., "Lift-Oscillator Model of Vortex- About Stationary and Transversely Oscillating Circular Cylinders," Technical
Induced Vibration," Journal of the Engineering Mechanics Division of ASCE, Report No. NPS-69SL79011, Naval Postgraduate School, Monterey, Calif.,
Vol. 96, EM5,1970, pp. 577-591. 1979.
88 Skop, R. A., and Griffin, O. M„ "A Model for the Vortex-Excited 117 Milne-Thomson, L. M-, Theoretical Hydrodynamics, The Macmillan
Resonant Response of Bluff Cylinders," Journal of Sound and Vibration, Vol. Co., New York, 4th ed., 1962.
27,1973, pp. 225-233. 118 Sarpkaya, T., "Lift, Drag, and Added-Mass Coefficients for a Circular
89 Iwan, W. D., and Blevins, R. D., "A Model for Vortex-Induced Oscil- Cylinder Immersed in a Time-Dependent Flow," ASME JOURNAL OF A P P L I E D
lation of Structures," ASME JOURNAL OF APPLIED MECHANICS, Vol. 41,1974, M E C H A N I C S , Vol. 32,1963, pp. 13-15.
pp. 581-586. 119 Durgin, W. W., March, P. A., and Lefebvre, P. J., "Lower Mode Re-
90 Iwan, W. D., "The Vortex-Induced Oscillation of Elastic Structural sponse of Circular Cylinders in Cross Flow," ASME Fluids Engineering Divi-
Elements," ASME Journal of Engineering for Industry, Vol. 97, pp. 1378- sion Symposium on Nonsteady Flows—If, Book No. H00118,1978, pp. 193-
1382. 200.
91 Blevins, R. D., and Burton, T. E., "Fluid Forces Induced by Vortex 120 Chen, C. F., and Ballengee, D. B., "Vortex Shedding From Circular
Shedding," ASME Journal of Fluids Engineering, Vol. 95,1976, pp. 19-24. Cylinders in an Oscillating Stream," AIAA Journal, Vol. 9, 1971, pp. 340-
92 Eaton, K. J., ed., Proceedings of the Fourth International Conference 362.
on Wind Effects on Buildings and Structures, Cambridge University Press, 121 Hatfield, H. M., and Morkovin, M. V., "Effect of an Oscillating Free
Cambridge, 1975. Stream on the Unsteady Pressure on a Circular Cylinder," ASME Paper No.

Journal of Applied Mechanics JUNE 1979, VOL. 46 / 257


72-WA/FE-12. Journal of Fluid Mechanics, Vol. 74,1976, pp. 641-665.
122 Davenport, A. G., "The Application of Statistical Concepts to the Wind 129 Savkar, S. D., "A Survey of Flow-Induced Vibrations of Cylindrical
Loading on Structures," The Institution of Civil Engineers, Vol. 19,1961, pp. Arrays in Cross Flow," ASME Paper No. 76-WA/FE-21.
449-472. 130 Pettigrew, M. J., Sylvestre, Y., and Campagna, A. 0., "Vibration
123 Tseng, M., "Drag of an Oscillating Plate in a Stream," Schiffstecknik, Analysis of Heat Exchanger and Steam Generator Design," Nuclear Engi-
Vol. 19,1972, pp. 28-34. neering Design, Vol. 48,1978, pp. 97-115.
124 Goddard, V. P., "Numerical Solutions of the Drag Response of a Cir- 131 Grover, L. K., and Weaver, D. S., "Cross-Flow Induced Vibrations in
cular Cylinder to Streamwise Velocity Fluctuations," PhD Thesis, University a Tube-Bank Vortex Shedding," Journal of Sound and Vibration, Vol. 59,1978,
of Notre Dame, 1974. pp.263-276.
125 Sarpkaya, T., "Impulsive Flow About a Circular Cylinder," Technical
Report No. NPS-69SL-78-008, Naval Postgraduate School, Monterey, Calif., 132 Rail, R. D., Hafen, B. E., and Meggitt, D. J., "Flow-Induced Vibrations
1978. of Three-Dimensional Bluff Bodies in a Cross Flow, an Annotated Bibliogra-
126 King, R., "Vortex Excited Oscillations of Yawed Circular Cylinders," phy," Technical Note No. N-1493, Civil Engineering Laboratory, Port Huen-
ASME Journal of Fluids Engineering, Vol. 99,1977, pp. 495-502. eme, Calif., 1977.
127 Ramberg, S. E., "The Influence of Yaw Angle Upon Vortex Wakes of 133 Simui, E., and Scanlan, R. H., Wind Effects on Structures: An Intro-

Downloaded from https://asmedigitalcollection.asme.org/appliedmechanics/article-pdf/46/2/241/4753961/241_1.pdf by University Of Windsor Libraries user on 28 August 2019


Stationary and Vibrating Cylinders," Memorandum Report 3822, Naval Re- duction to Wind Engineering, Wiley, New York, 1978.
search Laboratory, Washington, D. C, 1978. 134 Fabula, A. G., and Bedore, R. L., "Tow Basin Tests of Cable Strum
128 Stansby, P. K., "The Locking-on of Vortex Shedding Due to the Reduction (Second Series)," Naval Undersea Center Report NUC TN 1379,
Cross-Stream Vibration of Circular Cylinders in Uniform and Shear Flows," 1974.

258 / VOL. 46, JUNE 1979 Transactions of the ASME

You might also like