Review—Vortex Shedding Lock-on

and Flow Control in Bluff Body

0. M. Griffin Wakes
Naval Research Laboratory,
Washington, DC 20375-5000. The results of recent experiments demonstrate that the phenomenon of vortex shed-
Fellow ASME ding resonance or lock-on is observed also when a bluff body is placed in an incident
mean flow with a periodic component superimposed upon it. This form of vortex
shedding and lock-on exhibits a particularly strong resonance between the flow
M. S. Hall perturbations and the vortices, and provides one of several promising means for
Science Applications International modification and control of the basic formation and stability mechanisms in the
Corporation, near-wake of a bluff body. Examples are given of recent direct numerical simulations
McLean, VA 22102 of the vortex lock-on in the periodic flow. These agree well with the results of
experiments. A discussion also is given of vortex lock-on due to body oscillations
both normal to and in-line with the incident mean flow, rotational oscillations of
thebody, and of the effect ofsound on lock-on. The lock-on phenomenon isdiscussed
in the overall context of active and passive wake control, on the basis of these and
other recent and related results, with particular emphasis placed on active control
of the circular cylinder wake.

Introduction
Vortex streets are formed in the wakes of bluff, or un- quency which is near one of the characteristic frequencies of
streamlined, bodies over a wide range of Reynolds numbers, the structures (Bishop and Hassan, 1964; Koopmann, 1967).
from approximately 50 to 106 and even higher. The physics of Extensive recent reviews of vortex shedding from bluff bodies
vortex street formation and the near-wake flow have been the and vortex-induced oscillations have been given by Sarpkaya
focal point for many past experimental studies, e.g., Roshko (1979) and Bearman (1984). This coincidence or resonance of
(1954, 1955), Gerrard (1966), Bearman (1965, 1967), Griffin the vortex and vibration frequencies is commonly termed lock-
and Ramberg (1974) and, most recently, Unal and Rockwell on. The term phase-locking also has been used in the literature
(1988a, b), Ongoren and Rockwell (1988a, b), and Williamson (Rockwell, 1990). Lock-on or resonance occurs when the body
and Roshko (1988). One reason for this interest has been the is oscillated in-line with the incident flow (Griffin and Ram-
importance of knowing how the mean and fluctuating fluid berg, 1976; Ongoren and Rockwell, 1988b), and the lock-on
forces are generated on the body due to vortex shedding. An- resonance also is induced when a cylinder is forced to oscillate
other reason is the perceived connection of the near-wake flow normal to the flow over the appropriate range of imposed
to the eventual evolution of the overall middle and far-wake frequencies and amplitudes. Two recent studies (Tokomaru
vortex patterns (Cimbala et al., 1988; Browne et al., 1989). and Dimotakis, 1991; Filler et al., 1991) have shown that ro-
One of the most cogent descriptions of the physics of vortex tational oscillations of a circular cylinder can cause lock-on.
streets and bluff body wakes was given by Morkovin (1964) The recent computations of Karniadakis and Triantafyllou
as "a kaleidoscope of challenging fluid phenomena." This show that a lock-on state can be reached when a small spatially
description is in many ways still true today. Modern high- and temporally varying periodic disturbance is introduced into
speed computers and direct and large-eddy numerical simu- the near-wake of the cylinder. The disturbance is analogous
lation techniques now allow and, in the future, will further to a vibrating wire with the appropriate frequency and am-
allow the vortex formation and wake modification and control plitude.
processes to be studied computationally at high resolution Vortex resonance, or lock-on, has been observed also when
(Karniadakis and Triantafyllou, 1989, 1990; Grinstein et al., the incident mean flow has a sufficiently large periodic com-
1990, 1991). ponent superimposed upon it (Barbi et al., 1986; Armstrong
If a bluff cylinder is flexible and lightly damped, or rigid et al., 1986, 1987). In this case the cylinder remains stationary,
and flexibly mounted, then resonant oscillations can be excited but the vortex lock-on resulting from the inflow perturbation
by the incident flow. As a consequence of this flow-induced modifies the character of the near-wake flow. There is a com-
resonance, the body and wake oscillations have the same fre- plete equivalence between this case and in-line oscillations of
the cylinder when the acoustic wavelength is long compared
Contributed by the Fluids Engineering Division for publication in the JOURNAL to the cylinder's diameter. The introduction of an appropriate
OF FLUIDS ENGINEERING. Manuscript received by the Fluids Engineering Division sound field also can cause lock-on to occur (Blevins, 1985).
March 19, 1991. All of these external disturbances represent potential means

526 / Vol. 113, DECEMBER 1991 Transactions of the ASME

Copyright © 1991 by ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms
 for active control of the bluff body near-wake flow (Rockwell,
1987, 1990). Active control of the vortex shedding in the wake LEGEND:
In-line oscillations;
of a stationary circular cylinder by means of acoustic feedback O , +T fa = 190. Griffln and Romberg (1976);
A, Re = 80; A, Re = 4000, Tanida et at. (1973);
was demonstrated in the recent experiments of Ffowes Williams ®, Re = 100, Tatsuno (1972).
and Zhao (1989). Passive control of the shedding process can Flow perturbations;
• , Re = 3000; H, Re = 40000, Barbi et al. (1986).
be accomplished by geometric alterations such as a wake split-
ter plate (Bearman, 1965; Roshko, 1954,1955; Mansingh, 1986). 2AU 2a
Recent discussions of the stability and control of separated cod ' d

flows in general are given by Oertel (1990) and Rockwell (1990). A O

Vortex lock-on and resonance phenomena have numerous +
practical engineering applications. These applications abound 1+ #
in offshore exploration and drilling, Naval and marine hy- NON LOCK-ON ^ Q

drodynamics, and underwater acoustics. Other areas of en- $

gineering practice impacted by these phenomena are civil and * +++
'-$ -*$
wind engineering, nuclear and conventional power generation, 0.0 -J I— I 1-
and electric power transmission. Modification and control of 0.1
the flow can be employed to reduce the intensity of the wake
in order to reduce the drag, for example. These same processes Fig. 1 Limits of the lock-on regime as a function of amplitude and
also can be used to intensify the wake flow in order to enhance frequency for in-line oscillations and flow perturbations. For the region
heat transfer, mixing and combustion. enclosed by the dashed lines, see Fig. 2.
The emphasis of this review paper is on vortex shedding
resonance and lock-on in the near-wakes of bluff bodies. Vor-
tex shedding in a flow with a periodic component superimposed
on the basic mean flow is introduced here as the first case for \ \ i
0.02
study. This is an interesting bluff body flow which has not
been studied previously in detail. The more widely studied cases \ \ /
lm
of vortex shedding resonance and lock-on due to body oscil- +
lations both normal to, in-line with the incident mean flow,
and rotational are also discussed in some detail. The intro- 2AU
wd
' 1
+ + NON LOCK-ON
(CIRCULAR CYLINDER)

duction of sound also is discussed for the relatively few con-

tributions which are available. The discussion here is directed
principally toward the circular cylinder, but limited discussion
0.01 y
LOCK-ON
;'
LEGEND:
of other body configurations is introduced at places where it @, circular cylinder;
seems appropriate to do so. £, flat plate;
+ , D-section cylinder;
Re = 15000 - 35000.
Near-Wake Flow Scaling
0.0 I
Roshko (1954, 1955) and Bearman (1967) originally showed 1 1.9 2.0 2.1 2.2
that a characteristic group of nondimensional parameters for f
scaling of the wakes of bluff bodies could be derived by ap-
plying relatively simple physical arguments. The most recent Fig. 2 Limits of the lock-on regime as a function of amplitude and
formulation (Griffin, 1978, 1981, 1989) is a universal wake frequency for in-line flow perturbations; data from Armstrong et al. (1986)
Strouhal number St* for vortex shedding based upon measured
parameters of the bluff body near-wake flow.
If one considers two shear layers a distance d' apart, with A wake Reynolds number Re* is defined in a corresponding
the velocity just outside the layers equal to Ub, the mean ve- way as
locity at separation, then a wake Strouhal number can be * U»d' =ReiH
pRe*= i, J—d
'
defined as . . , „ (5)
v \d
where Re=Ud/v is the usual free-stream Reynolds number.
These scaling relations are employed later in the paper to cor-
The characteristic frequency fso associated with the flow is
relate the near-wake flow properties with one another.
assumed to be proportional to the ratio Ub/d'. Here the clas-
sical Strouhal number of the vortex wake is
Flow Perturbations and In-Line Oscillations
fsod
St = (2) The recent experiments of Armstrong et al. (1986,1987) and
U
of Barbi et al. (1986) were conducted to examine the problem
where d is the cylinder diameter and U is the incident flow of vortex lock-on for a cylinder in a stream consisting of a
velocity. When Bernoulli's equation is applied to the flow just steady flow with a periodic component superimposed upon it.
outside the boundary layer at separation, the base pressure In earlier experiments, Hatfield and Morkovin (1973) at-
coefficient is tempted to study the same problem, but the results were in-
n 2(pb-pm) Ub • conclusive because the flow perturbation amplitude and
(3) frequency were too low to cause lock-on. The results obtained
by Barbi et al. and Armstrong et al. show some very basic
If the base pressure parameter or velocity ratio K= Ub/U is similarities with the earlier experiments of Griffin and Ramberg
introduced, then (1976), which were conducted to examine vortex shedding lock-
Kl=\-C, pb (4) on for a cylinder oscillating in-line in a steady incident flow.
The vortex lock-on regime measurements by Barbi et al. are
and compared with those of Griffin and Ramberg in Fig. 1. The
St id' vertical axis represents two different measures of the pertur-
(4fl)
bation amplitude. For the experiments of Griffin and Ram-

Journal of Fluids Engineering DECEMBER 1991, Vol. 113 / 527

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 berg, the amplitude parameter is defined by the ratio of the
peak-to-peak amplitude of cylinder displacement 2a and the
cylinder diameter d. And for the experiments of Barbi et aI.,
the normalized "peak-to-peak" incident velocity perturbation
is given by 2!1Vlwd. The horizontal axis is the ratio of the
vibration frequency / and the Strouhal frequency /50 of a sta-
tionary cylinder. Also shown are the cylinder vibration results
of Tanida et al. (1973) and of Tatsuno (1972), reproduced
from the paper by Griffin and Ramberg. The dashed lines
enclose the region occupied by the results of Armstrong et al.
(1986, 1987) which are shown on an expanded scale in Fig. 2.
Vortex lock-on and cross-flow oscillations usually occur near
the Strouhal shedding frequency /50' For in-line oscillations
and flow perturbations, the lock-on is caused by frequencies
which occur near twice the Strouhal frequency, /=2/50 , since
the fluctuating drag force is in the flow direction. However,
in many cases the actual lock-on frequency is near the Strouhal
frequency, or half the oscillation or perturbation frequency.
There is generally good agreement between the bounds of
the lock-on regime for the two different types of external
disturbance or flow control, though there is some scatter at
the highest amplitudes. This is most likely due to Reynolds
number effects, as noted by Barbi et al. The latter experiments
were conducted at Re between 3,000 and 40,000, whereas the
results of Tanida et aI., Tatsuno, and of Griffin and Ramberg
were conducted at Re between 80 and 4,000. The overall dif-
ferences are relatively small in any case.
In Fig. 2 the vertical and horizontal axes have been scaled Fig. 3 Time sequence of locked-on vortex shedding produced by cyl·
in the same way as in the previous figure. The original results inder oscillations In·line with the flow at an oscillation frequency of
of Armstrong et al. had been plotted in terms of the rms velocity t = 2 t,.; from Ongoren and Rockwell (1988b)
u' and the reduced velocity VI/sod. Three body shapes were
investigated, i.e., a circular cylinder, a D-section cylinder, and on due to in-line oscillations of a circular cylinder was con-
a vertical flat plate. It is clear that the circular cylinder, with ducted by Ongoren and Rockwell (1988b). These experiments
free separation points, has a lock-on range of about twice the also included oscillations of the cylinder at inclination angles
breadth of the two bodies with fixed separation points. This to the flow between a = 0 degrees (in-line) and IX = 90 degrees
basic difference in the lock-on behavior for these types of bluff (cross-flow), but the emphasis of the discussion here is on the
bodies was previously discussed by Bearman and Davies (1975) in-line oscillations. The cross-flow oscillations (Ongoren and
and by Bearman (1984) for the case of body oscillations only. Rockwell, 1988a) are discussed in the next section. A wide
As shown by the former, the afterbody shape plays an im- range of vortex patterns was visualized by introducing pulsed
portant role in the character of the lock-on or resonance, e.g., hydrogen bubbles into the incident flow about a circular cyl-
in terms of the response of the base pressure and near wake inder mounted vertically in a free-surface water channel.
flow to the forcing. Both symmetric and asymmetric vortex patterns were ob-
The base pressure coefficient Cpb is influenced by the flow served over a wide range of oscillation conditions. For the in-
perturbations in much the same manner as in the case of cyl- line oscillations, vortex lock-on was observed at / = 2, 3, and
inder oscillations. For the stationary cylinder the base pressure 4 /50' with an asymmetric street formed at twice the basic
coefficient is near Cpb = - 1.44; this value, though somewhat Strouhal frequency and a symmetric street formed at three
low for a circular cylinder, is in reasonable agreement with the times the Strouhal frequency. The asymmetric pattern was
results of West and Apelt (1982) for a comparable wind tunnel complex in that one row consisted of a line of single vortices,
blockage ratio of nine percent. When the flow perturbation whereas the other row consisted of a line of oppositely rotating
was largest, the base pressure was decreased to Cpb = - 1.85 vortex pairs. The vortex lock-on at three times the Strouhal
at the point of maximum resonance, a reduced velocity of VI frequency resulted in the formation of a symmetric street of
/sod=2.5 (half the Strouhal value). The measured vortex for- vortices. In these cases the basic patterns persist downstream
mation region length lfwas reduced by this level of perturbation over a large number of oscillation cycles. When the oscillation
to 0.9d from 1.2d, the value measured for the unperturbed frequency is four times the Strouhal frequency, a symmetric
flow (Armstrong et al. 1987). pattern is formed but rapidly loses its coherence in the early
Lesser decreases in Cpb were measured for smaller levels of wake.
the flow perturbation, with an overall dependence upon re- A time sequence over a full in-line oscillation cycle is shown
duced velocity VI/sod. The mean drag coefficient CD increased in Fig. 3 for the condition/=2fso' Figure 3(a) was taken with
from 1.28 to 1.52 for the perturbed flow as compared to the the cylinder in its forwardmost position and shows a vortex
unperturbed flow. The base pressures of the flat plate and D- shedding from one side of the cylinder as in Figs. 3(c) and (d).
section bodies also were decreased by the introduction of the As the cylinder moves through its maximum downstream po-
incident flow perturbations. But the decrease was only half of sition and changes direction, a second vortex is formed and
that measured for the circular cylinder at the same perturbation shed from the same side of the cylinder as in Figs. 3(e) and
amplitude, which further shows the effect of free versus fixed (f). Then a single vortex is formed from the other side of the
separation points on the vortex resonance. These experiments cylinder as the motion cycle continues as shown in Fig. 3(g).
were conducted at Reynolds numbers between 15,000 and Ongoren and Rockwell observed that the pattern persisted over
35,000, and the base pressure coefficients of all three stationary 50 or more cycles of the oscillation, but often, if the flow was
bodies in the unperturbed flow were effectively constant over stopped and restarted, a mirror image of the pattern was
this range. formed. This is but one example of the complexity of the flow
A recent experimental study of vortex resonance and lock- patterns which accompany the oscillations. In this case the

528/ Vol. 113, DECEMBER 1991 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 v(x,y,t)
-0.13287 0.00233

Fig. 4 Instantaneous streamlines, at Re = 200, of the forced wake re-

sponse at f=2.2fao for vortex shedding from a cylinder in a perturbed
flow. The wavelength \ of the vortex street is 4.66 cylinder diameters.
The spectral element computational grid is superimposed on the stream-
line pattern.

Fig. 6 Phase plane plot of the fluctuating velocity components, at

Re = 200, for the location and forcing conditions given in Figs. 4 and 5.
The regularity of the plot is demonstrative of the locked-on state of the
vortex shedding.

1
' I I I ! i I i | i
0.70

\
0.60
-f\ -
Frequency, fd/U 0.50
Fig. 5 A typical power spectrum of the flow velocity, at Re = 200, for \ \\ t -
the location (x = 2d, y = 2 d ) and the forcing conditions given in Fig. 4.
2a/d
\ \ \ •\ \ 1
0.40 - (
Only the locked-on shedding frequency, /=1.1 fso, the perturbation fre-
quency, f=2.2fso, and their harmonics are dominant in the spectrum.
0.30 \ \ ''
oscillation and vortex frequencies are phase-locked, but under - \\\ / -
B. \ °/
other nonresonant conditions there was competition between
the symmetric and asymmetric modes. Under these conditions,
0.20 \\ //
\ \\ LOCK-ON AS.
the lock-on persists in one mode over a specified number of V - -di?/-
i// -
cycles and then switches to the other mode. The mode com- 0.10
THRESHOLD
petition also is influenced by the upstream feedback of dis- DISPLACEMENT
turbances from the near-wake of the cylinder. Complex patterns 0 i I i 1 i 1
of three vortices such as these also were photographed by 0.7 0.8 0.9 1.0 1.1 1.2 1.3
Griffin and Ramberg (1976) at similar frequencies during their
wind tunnel experiments. Fig. 7 Limits of the lock-on regime as a function of amplitude and
frequency for cross-flow oscillations; from Koopmann (1967)
Numerical simulation provides yet another method of ex-
amining the effects of inflow perturbations and cylinder os-
cillations on the wake. This consists of superimposing an so that lock-on does occur and the results are in keeping with
oscillatory component on the inflow boundary condition for those shown in Fig. 1.
a domain such as that shown by the spectral element grid in After an initial period of time corresponding to the quasi-
Fig. 4. The example given here was computed at NRL using steady stage in the forced perturbed flow calculation, the re-
a computer code similar to one employed extensively by Kar- sulting streamwise velocity history at a point in the near wake
niadakis and Triantafyllou (1989, 1990). The grid consists of is periodic. The corresponding power spectrum in Fig. 5 con-
56 spectral elements, each of order iV=6. Results of the com- tains primary peaks a t / = l . l and 2.2 fso as expected, and
putation are shown in Figs. 4, 5, and 6, in which the lowest secondary peaks at superharmonics of these values. No ad-
dominant frequency of the resulting vortex wake is near half ditional peaks appear in the spectrum. The phase plane plot
the perturbation frequency when a boundary condition of the corresponding to this case is shown in Fig. 6, and with the
form power spectrum gives evidence of lock-on in the fully developed
M=1.0 + (0.8)sin(4.4ir/OT t) flow. Streamlines corresponding to this case are also shown
in Fig. 4. The vortex spacing here is approximately X = 4.66rf,
y = 0.
representing a decrease of seven percent over the unforced
is enforced at the inflow. This also represents an in-line os- value of X = 5d. The normalized frequency \/d(f/nfso) = 5.13
cillation, and is thus expected to result in a shedding frequency for this case. These values compare well with the results of
near / = 1.1 fso if lock-on occurs at one-half the perturbation experiments which are discussed later in the paRer. A more
frequency. The amplitude of the oscillatory component can be extensive discussion of the spectral element computations of
expressed as the perturbed flow cylinder lock-on is given by Hall and Griffin
a = 0.8 = 7.037r/M, (1991).

Journal of Fluids Engineering DECEMBER 1991, Vol. 113 / 529

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 1.5

1.4

1.3

l / 1.2
:g

~
1.1
~
1.0

0.9

0.8 +

Fig.8(a) 0.7
1.0 2.8
iF/d'
Fig. 9 The measured dependence between the initial circulation K of
the vortices and the ratio of the formation region length If and the wake
width d' at formation. The Reynolds number is Re == 144; lrom Griffin
and Ramberg (1974).

flow perturbations shown in Figs. 1 and 2. However, the im-

posed oscillations are near the Strouhal frequency fso rather
than twice its value.
Many other effects of the cross-flow oscillations are also
similar. For example, the longitudinal spacing of the vortex
street adjusts in a similar manner to the example shown earlier;
oscillation frequencies less than fso expand the vortex street
while frequencies greater than Iso contract the pattern. In-
creasing the amplitude of oscillation reduces the lateral spacing
of vortices to the point of zero spacing, after which there is a
drastic change in the appearance of the pattern as the flow
Flg.8(b)
adjusts to preclude the transition to a thrust-type vortex street.
Three previously unpublished examples from experiments at
NRL which demonstrate this effect are shown in Fig. 8. The
street behind a stationary cylinder appears in Fig. 8(a) and
shows the well-known geometry which has been visualized by
numerous investigators. When the amplitude of oscillation is
increased as shown in Fig. 8(b), the lateral spacing is much
reduced. For still higher amplitudes of oscillation, beyond the
limit of zero lateral spacing, a complex asymmetric pattern
such as that shown in Fig. 8(c) emerges. These photographs
were taken in a wind tunnel using an aerosol as the indicator.
This emergence of the asymmetric pattern also has been ob-
served by Ongoren and Rockwell (1988a) in water, using hy-
drogen bubbles as the flow indicator.
The formation region of the vortices as defined by the model
of Gerrard (1966) also varies inversely with frequency in the
resonance or lock-on regime (Griffin and Ramberg, 1974; On-
Fig.8(e)
goren and Rockwell, 1988a), and is reduced in length by in-
creasing amplitude of oscillation at any given constant
Flg.8 Flow visualization of the vortex shedding from a circular cylinder
oscillating In cross·flow at a Reynolds number of Re == 190. (a) stationary frequency. These changes in the near wake vortex formation
cylinder, unforced wake; (b) 2yld == 0.5, flf,. == 0.9; (c) 2yld == 1.1, flf,. == 0.9. cause corresponding changes in the strength or circulation of
the vortices. Reductions in the vortex formation length result
in increasing the vortex strength by as much as 75 percent at
Cross-Flow Oscillations a Reynolds number of 144. An example is shown in Fig. 9
The excellent reviews by Sarpkaya (1979) and by Bearman where the nondimensional initial circulation of the vortices is
(1984) dealt with cross-flow oscillations of flexibly-mounted plotted against the ratio of the formation length to the wake
bluff bodies, bodies which were free to oscillate, and those width at formation. The basic importance of the length scales
which were forced. The purpose of the present paper is to to the near wake flow physics is discussed later in the paper.
complement these works, to compare with some pertinent past This increase in the vortex strength is accompanied by a cor-
results, and to highlight more recent developments in the con- responding increase in the rate of vorticity generation with
text of flow control and modification. The basic character of amplitude of oscillation.
cross-flow l.ock-on due to forced oscillations can be represented The base pressure coefficient Cpb also is reduced significantly
by the measurements of Koopmann (1967) which are shown by the oscillations in the lock-on or resonance regime (Stansby,
in Fig. 7. The appearance of the lock-on range is very similar 1976). As an example, the minimum base pressure on a circular
overall to the corresponding cases of in-line oscillations and cylinder at resonance was decreased by 33 percent as the am-

530 I Vol. 113, DECEMBER 1991 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 o

.,

II

'- r.~ I"

1

Fig. 10 Effect of lhe ratio of oscillation frequency' to the natural

shedding or Strouhal frequency's. on the vorlex formation region of a
circular cylinder; from Ongoren and Rockwell (1988a). All of the photo.
graphs were taken with the cylinder at its maximum negative position.

Fig. 11 Flow visualizalion of synchronized vortex shedding due to roo

plitude of oscillation (measured in diameters from equilibrium) lational oscillalion of a circular cylinder. Legend for data polnts:!l, =8;
increased from O.ld to 0.3d. The Reynolds number of the Re = 15,000; (a) St,= 0.3, (b) 5t,= 0.5, (e) 5t = 0.7, (d) St,= 0.9; from Tok·
experiment was Re = 8600. As mentioned earlier herein, Arm- umaru and Oimolakis (1991).
strong et al. (1987) measured comparable decreases in the base
pressure. However, the level of inflow perturbations was much substantial decrease in the length of the vortex formation re-
less than the oscillation levels required to achieve the same gion with frequency of oscillation also is evident from the
level of base pressure modification. As noted earlier, the latter photographs.
reduction in base pressure was accompanied by a reduction in Other effects of the cross-flow oscillations observed by On-
the length of the vortex formation region from 1.2d to 0.9d. goren and Rockwell included a large swinging motion of the
The most comprehensive recent study of cross-flow oscil- circular cylinder wake about the body at frequencies less than
lations is that of Ongoren and Rockwell (1988a). The Reynolds the Strouhal frequency. This swinging motion is largest near
number range of the experiments was Re = 580 to 1300. They the Strouhal frequency and then it abruptly disappears. There
found that two fundamental types of lock-on take place; at a is evidence of some correlation between the phase shift in the
frequency of one-half of the Stouhal frequency, a subharmonic shedding and the abrupt disappearance of the swinging motion.
form of lock-on takes place whereby the shed vortex is always
The swinging motion does not appear in the case of bodies
from one side of the body, whereas at frequencies near the with fixed separation points. There were numerous complex
Strouhal frequency the classical form of lock-on described wake patterns observed at other frequencies. For instance,
above takes place as vortices are shed alternately from the small-scale vortices were shed at superharmonic frequencies
body to form an altered Karman vortex street pattern. of n = 2, 3, 4, and higher nonharmonic frequencies, and the
It has been known for some time (Bearman and Currie, downstream wake eventually recovered to a lock-on and altered
1979; Zdravkovich, 1982) that a drastic change in the phase asymmetric pattern similar in form to the classical Karman
of the vortex shedding, relative to the body oscillations, occurs vortex pattern. However, the altered pattern departed sub-
in the vicinity of the natural shedding frequency. However, stantially from the classical Karman street, with the frequency
Ongoren and Rockwell and, earlier, Bearman and Davies (1975) taking the values la =11n. An interesting aspect of the paper
showed that the afterbody shape plays an important role in by Ongoren and Rockwell is the number of historical refer-
the phase shifting in that bodies with a short or nonexistent ences, e.g., Meier-Windhorst (1939).
afterbody, i.e., a circular or triangular cylinder, experience a
large phase shift, while a body such as a square or rectangular
cylinder with a relatively large afterbody experiences little or Rotational Oscillations
no phase shift. This phase shift results in the switch of the Vortex lock-on and control of the near-wake flow also can
initially shed vortex from the upper to the lower side of the be realized with small rotational oscillations of a circular cyl-
cylinder or vice versa. The presence of the afterbody appears inder. There are very few studies of this aspect of the problem,
to induce reattachment of the initially shed vortex and to reduce the most recent being those of Tokumaru and Dimotakis (1991)
the likelihood of the phase shifting. This is yet another indi- and of Filler et al. (1991). An important distinction between
cation of the importance of the vortex formation region and the two studies is that in the experiments of Tokumaru and
near-wake flow to the shedding process. Dimotakis the maximum rotational velocity of the cylinder
The changes in the vortex formation region with the fre- .was on the order of the velocity outside the boundary layer at
quency of the oscillations are shown by the photographs in separation (approximately 1AU), whereas in the experiments
Fig. 10 from Ongoren and Rockwell (1988a). The flow was of Filler et al. the maximum rotational velocity of the cylinder
visualized in water in the manner described in the previous was only 0.03U. These studies seem to show, however, that
section of the paper, and again the results are remarkably the range of lock-on frequencies, though still probably am-
similar to the earlier wind tunnel photographs of Griffin and plitude-dependent, is much reduced from what has been ob-
Ramberg (1974). All of the photographs were taken with the served for imposed oscillations and flow perturbations as
cylinder at its lowest position in the oscillation cycle, and the discussed earlier. There are essentially two types of forcing
shift in phase of the shedding relative to the cylinder can be which can be introduced by rotational oscillations. The first
seen by comparing the wakes at/=0.9/so and/= 1.05/so • The is the classical form of lock-on or resonance which takes place

Journal of Fluids Engineering DECEMBER 1991, Vol. 113/531

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 0 I 1 1 1 1 1 1
0 0.2 0.4 0.6 0.8 1.0 1.2
Forcing Frequency F = f d / U
Fig. 12 A typical frequency response curve showing the normalized
velocity response amplitude as a function of the forcing frequency for
Re = 920, the vibration parameter SI, = 0.14, and d = 1.27 cm; from Filler
etal. (1991) 360.0 370.0 380.0 390.0 400.0
Sound Frequency f(Hz)

Fig. 13 Entrainment of natural or unforced vortex shedding at 392 Hz

when the oscillations are near the Karman vortex shedding by sound at 380 Hz. The vertical scale is the same for all three spectra;
from Blevins (1985).
frequency. When the Reynolds number is greater than about
Re = 500, oscillations of higher frequency also can excite the
Bloor-Gerrard (Bloor, 1964) frequency of instability in the The Effects of Sound
shear layers separating from the cylinder. This excitation of The application of an appropriate sound field to the flow
the B-G instabilities is likely to be caused by inflow pertur- about a rigid cylinder can induce vortex lock-on and resonance
bations and other types of imposed body oscillations as well. in the wake. There are even fewer reported studies of the effect
A series of photographs of vortex lock-on at increasing val- of sound than of rotational oscillations, the principal example
ues of the oscillation Strouhal number St/ are shown in Fig. of the former being that of Blevins (1985). The only other
11. The oscillation parameter Q1 = yi/(7, where v is the peak directly related work is that of Okamoto et al. (1981). As noted
circumferential velocity, was kept fixed during the sequence by Blevins, the effect on the vortex shedding of an acoustic
of conditions shown in this figure, from Tokumaru and Di- wave propagating along the axis of a circular cylinder was
motakis (1991). The experiments were performed in a CalTech examined by the latter. Only a minor influence was observed
water channel at a Reynolds number of Re= 15,000, and the for sound excitation levels above 20 Pa.
flow was visualized by introducing a mixture of colored food The experiments performed by Blevins were at Reynolds
dye into the water upstream of the cylinder. The dye mixture numbers in the range Re = 20,000 to 40,000 in a wind tunnel
was made neutrally buoyant by diluting it with ethyl alcohol that allowed a transverse sound field to be applied such that
(P.E. Dimotakis, private communication). In all of the cases the cylinder was located at the node of the acoustic pressure
shown the vortex shedding is locked-on in the classical manner field. This is the point of maximum induced velocity due to
with the imposed rotational oscillations. When the oscillation the sound. It was observed by Blevins that the vortex lock-on
Strouhal number was increased to St/= 1.5, the wake forcing was induced by the velocity rather than the pressure.
went through a transition to the Bloor-Gerrard shear layer
forcing. For a fixed value of oscillation Strouhal number of An example of the results reported by Blevins (1985) is shown
St/= 1, the transition to shear layer forcing takes place at an in Fig. 13. The frequency of the vortex shedding was/ s = 392
oscillation parameter of approximately 0= 16. Hz, and the frequency of the applied sound field was/= 380
Hz. Thus the lock-on occurred at a frequency less than the
A plot of the velocity u' in the wake from Filler et al. (1991) Strouhal frequency. The average spectral output from a flush-
as a function of the oscillation Strouhal number St/ is shown mounted hot-film probe mounted on the cylinder is plotted
in Fig. 12. Here the oscillation parameter 0] =wid/2U, where for the three test runs. The spectrum labeled 1 shows the typical
oi! is the frequency of the rotational oscillations. At the lower averaged spectrum for a rigid cylinder in a uniform flow, with
frequencies near the usual Karman shedding frequency a large the broad peak in this case centered at the vortex shedding
resonant peak is seen when the oscillations are in that range. frequency of 392 Hz. When a 100 Pa sound field is applied
However, at the higher imposed frequencies there is a sec- two peaks are present—a sharp peak at 380 Hz induced by the
ondary broad peak in the range of the shear layer instability applied sound, and a broader reduced and shifted peak due
frequencies. In the Karman frequency range of vortex shedding to vortex shedding. An increase in the applied sound field to
the wake behaves like a nonlinear oscillator near resonance. 250 Pa produces a typical lock-on spectrum with a single sharp
This behavior is well known (Bishop and Hassan, 1964; Bear- peak at the frequency of the sound. The lock-on frequency
man, 1984) and has been explored by numerous investigators here is less than the Strouhal frequency, and it was observed
for the cases of cross-flow and in-line oscillations. The forced by Blevins that the induced resonance always was stronger at
Bloor-Gerrard shear layer instabilities are simply convected the reduced frequencies as compared to frequencies higher than
downstream in the near wake. An important finding by Tok- the Strouhal frequency. The sound field and the vortex shed-
umaru and Dimotakis (1991) is that active control of the near- ding were phase locked over a range of phase angles which
wake vortex formation and flow physics by rotational oscil- varies nearly linearly with the applied sound frequency.
lations of the cylinder can reduce the drag on the cylinder by
as much as a factor of six! This decrease in the drag coefficient An additional observation by Blevins (1985) was that tur-
CD is accompanied by a comparable decrease in the wake bulence in the free stream suppressed the influence of sound
displacement thickness 5* (a factor of five) as defined by an on the vortex shedding. The results suggest that the induced
integral of the cross-stream wake velocity distribution over the sound field velocity must exceed the turbulence velocities in
height of the channel. order for the sound to influence the vortex shedding. Also,
the introduction of sound substantially increased the coherence
There are very few studies of rotational oscillations on wake of the vortex shedding along the span of the cylinder as is
flow control and vortex resonance or lock-on. Examples from usually found when a circular cylinder is oscillated.
the two most recent have been given here. For earlier examples
the reader should refer to the work of Okajima et al. (1975)
and of Taneda (1978). This is a new and potentially exciting The Near-Wake Flow Field
approach to the active control of vortex formation and bluff There is a physical dependence between the wake width d'
body wake flows. at the end of the vortex formation region of a bluff body and

532 / Vol. 113, DECEMBER 1991 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 Table 1 Legend lor data plotted in Fig. 14
Symbol Bluff body type Method Investigators
» Half cylinder Wind tunnel experiment Armstrong et al.
(D-section) (1986, 1987)
1- Flat plate Wind tunnel experiment Armstrong et al.
(1986, 1987)
© Circular cylinder Wind tunnel experiment Armstrong et al.
9 Circular cylinder Wind tunnel experiment (1986, 1987)
Barbi et al.
BDI Wide splitter plate Computation (1986)
(D-section) Grinstein et al.
1 Hydrofoil Wind tunnel experiment (1991)
Blake et al.
+ D-section cylinder Wind tunnel experiment (1977)
V- V D-section cylinder with Wind tunnel experiment Simmons (1975)
and w/o splitter plate Bearman (1965)

pressure parameter or separation velocity ratio K represents

the entire regime over which vortex shedding takes place over
bluff bodies, whether the shedding is natural or unforced, or
the shedding is controlled or modified by some means such as
oscillations, in-flow perturbations, sound, and wake splitter
plates. The shaded area represents a host of results for oscil-
lating circular cylinders, D-section cylinders, and flat plates
under a wide variety of conditions (Griffin, 1989). There is
generally good agreement between the new perturbed and steady
flow results and the earlier data, except for the two circular
cylinder results from Armstrong et al. which are displaced to
the right of the overall trend of the data set. This departure
from the overall trend of the results is due to the much reduced
base pressure measured on the circular cylinder in those ex-
periments, as previously mentioned. In the experiments of
Armstrong et al. the circular cylinder and flat plate base pres-
sures were virtually the same under otherwise unvarying con-
ditions of blockage, incident flow, axial uniformity, etc. One
might expect the base pressure coefficient for a circular cylinder
at the Reynolds numbers studied to be closer to the values of
Cpb = - 1 to 1.1 measured by Barbi et al. and others at the
same Reynolds numbers.
The wake widths for flow over a D-section cylinder com-
Fig. 14 Wake width d'Id as a function of the base pressure parameter puted by Grinstein et al. (1990, 1991) are also plotted in Fig.
or velocity ratio K. The legend is given in Table 1. The shaded area in 14. The results shown represent a two-dimensional computa-
the figure corresponds to a variety of body geometries and flow con- tion using the flux-corrected transport (FCT) algorithm, but
ditions given in Griffin (1989).
comparable results were obtained with companion three-di-
mensional computations described by Grinstein et al. (1990).
the base pressure coefficient - Cpb or the related velocity ratio These computations were made for compressible flows with
K. As discussed earlier, these are important characteristic phys- freestream Mach numbers in the range 0.3 to 0.6 at standard
ical parameters of the bluff body wake. In general, the bluff- temperature and pressure conditions. However, for these Mach
ness of a given body shape is represented by a wider wake, numbers compressibility effects are relatively small and rea-
and corresponding lower base pressure or higher velocity ratio. sonable comparisons can be made with incompressible flow
For a cylinder vibrating normal to the incident flow, the var- experiments. The wake widths plotted in Fig. 14 were obtained
iation of both the measured base pressure and wake width with directly from computed contour plots of rms velocity u' at
the frequency ratio f/fso over the lock-on regime show the the end of the vortex formation region. The results shown
same resonant behavior (Griffin, 1989). Both -Cpb and d' represent flow over the body both with and without a splitter
increase to a maximum value and then gradually decrease as plate attached and they agree remarkably well with the ex-
the upper limit of the lock-on range of frequencies is reached. perimental results of Blade et al. (1977), Simmons (1975) and
Only the base pressure variation was measured by Armstrong Bearman (1965). The D-section bluff body results overall rep-
et al., but the wake width can be estimated as a function of resent the lowest regime of K which has been observed thus
U/fsdd, using the wake similitude relationships summarized far.
earlier in the paper and measured values of St and Cpb (or K). A comprehensive experimental study of the effects of wake
For the range of Reynolds numbers corresponding to the ex- splitter plates on vortex shedding from a circular cylinder was
periments of Armstrong et al., the wake Strouhal number St* made recently by Unal and Rockwell (1988b). The experiments
is essentially constant at a value of 0.16, so that were conducted in the Reynolds number range Re =140 to
5000, and the primary objective was to study the effects of the
d'/d=(St*/St)K passive wake interference on the formation region of the vor-
from Eq. (4a). tices. An unusual aspect of these experiments was the ratio of
The wake widths for several cylinders and flow conditions the plate thickness h to the cylinder diameter d, which was
are plotted as a function of K in Fig. 14. The legend for the approximately 0.5. Also, the plate length was approximately
data in the figure is given in Table 1. This range of the base lp = 24d cylinder diameters, in contrast to the splitter plate

Journal of Fluids Engineering DECEMBER 1991, Vol. 113 / 533

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 Table 2 Longitudinal vortex spacing or wavelength in the near-
wake of a circular cylinder vibrating in-line with an incident uni-
form flow
Vibration Frequency Vortex Relative Vortex
equency, ratio, spacing, change, convection speed
/(Hz) X/d AX/X 1/2/X/U
Reynolds number == 190
69.2 1.88 5.2 + 0.07 0.94
73.6 2.00 4.9 0 0.93
75.6 2.06 4.7 -0.05 • 0.91
78.9 2.14 4.7 -0.04 0.96
80.4 2.18 4.4 -0.09 0.92
Average = 0.93

of the streamwise and cross stream components of the wake

velocity fluctuations which can be employed comparably to
assess the temporal state of the near-wake.
Two examples from Karniadakis and Triantafyllou are shown
in Fig. 15. The upper instantaneous streamline pattern cor-
responds to the unforced wake at Re =100 while the lower
pattern corresponds to a wake forced by a spatially localized
acceleration which varied with time in the near-wake vortex
formation region with normalized amplitude and frequency,
respectively, of y4=0.10 and f/fso = 0.75, and which decayed
(a) exponentially in the far field. The center of the disturbance
was located at x=2, y = 0, measured in multiples of cylinder
diameter. This is a unique form of control disturbance which
had not been investigated in any previous work. For the un-
forced wake \ = 5d and for the forced wake X = Id, an increase
of forty percent. A similar example from the NRL perturbed
flow computations is given in Fig. 6.
Comparable measurements were made at Re = 190 by Griffin
and Ramberg (1976). The results are summarized in Table 2.
The cylinder oscillations were in-line with the flow over a range
of frequencies near twice the Strouhal frequency (as in Fig. 1)
(b)
and, for the cases shown, a single vortex was shed during each
Fig. 15 (a) Instantaneous streamlines at Re = 100 for natural shedding.
oscillation cycle. Thus this basic forced wake pattern shared
(b) Instantaneous streamlines at Re = 100 and near-wake forcing at the many of the same overall features of the wake forced with the
excitation frequency, fo = 0.75rs (a lock-in state); from Karniadakis and cross-flow oscillations. The measured changes in the forced
Triantafyllou (1989). wake vortex spacing correspond directly with those from the
direct numerical simulations; for f<2fs0 the wavelength is
increased while for/> 2fso the wavelength is decreased as shown
geometries discussed above. The leading edge of the plate had in Table 2. Extrapolating the measured results in the table to
a tapered and sharply pointed configuration. the case shown in Fig. 15 using a least-squares straight line
These experiments yielded several important conclusions and given by Griffin and Ramberg (1976), the vortex spacing is
reinforced those previously reached by other investigators. The A = 6.2d. This compares reasonably well with the computed
first is a reiteration of the importance of the dynamics of the results of X = ld. The measured vortex spacing for the sta-
formation region to the overall near-wake flow and the pos- tionary cylinder at Re = 190 (X = 4.9d) is virtually identical to
sibility of control and modification of the vortex wake. Second, the computed value at Re= 100 (\ = 5d).
the wake formation is dominated by an absolute instability The vortex street wavelengths computed by Karniadakis and
and there is a close relation between the vortex formation Triantafyllou and more recently at NRL are compared further
process and the dynamics of the near-wake Karman vortex with measured street wavelengths for both forced and unforced
street. Other conclusions are the importance of Reynolds num- conditions in the range of Reynolds numbers from 100 to 2000
ber to the wake formation, at least over the range examined, in Fig. 16. The computations fit well with the overall trend of
but that no locking-on or vortex resonance occurs in the pres- the measured data, which show only a very slight dependence
ence of the passive wake interference. Nonetheless, the results on Reynolds number in this range. The vertical scale in the
of these and the other experiments and computations discussed figure essentially is a normalized form of the convection speed
here have demonstrated the potential importance of both pas- of the vortices, or the downstream speed of the vortex cores.
sive and active control to the flow physics of bluff body wakes. The constant phase or convection speed is representative of a
The downstream vortex spacing or wavelength is a valuable non-dispersive physical system.
and important diagnostic parameter for the state of the spatial • Several measurements of the vortex phase or convection
structure and development of the near-wake. Measurements speed, i.e., the speed at which the vortex cores travel down-
of the spacing for a variety of in-line and cross-flow oscilla- stream, are given in Table 2. Though there is some scatter, the
tions, and also for the unforced wakes of stationary cylinders data generally are grouped around the average value of
were reported originally by Griffin and Ramberg (1976). These t/^, = 0.93t/. This gives some evidence that both forced and
can be compared to the direct numerical simulations of Kar- unforced or natural periodic vortex wakes have the same basic
niadakis and Triantafyllou (1989) and our recent NRL simu- non-dispersive properties.
lations (Hall and Griffin, 1992). The vortex spacing or The experiments of Tokumaru and Dimotakis (1991) also
longitudinal wavelength can be employed as a measure of the included measurements of the vortex street wavelength X over
spatial state of the flow as compared to phase plane diagrams a wide range of the oscillation Strouhal number Stf=fd/U.

534 / Vol. 113, DECEMBER 1991 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 when an initial infinitesimally small disturbance grows expo-
nentially in place at the location where it is introduced. In the
6.0 - case of a convective instability the initially small disturbance
is transported or converted away from the point of its intro-
> duction, leaving behind an undisturbed region of flow. Chomaz
5.0 - & et al. (1988) have introduced the term global instability. They
Oy y O note that the existence of a local absolute instability is only a
X/d(f/nf > o necessary, but not sufficient, condition for the existence of a
4.0 - - global instability in a shear flow, and that the localized region
of absolute instability must physically grow to a large size to
become globally unstable.
3.0 - LEGEND:
Experiments: Recent stability calculations based upon computed and
9. O. I!•>!,.- i A , / / „ / , „ = 0.9;
x . / / » / „ = 1.07;
measured mean velocities in the wakes of stationary circular
2.0 Computations;
• , / / n / „ » 0.75 1-00 (Karniadakis and Triantafyllou 1989);
cylinders suggest that the vortex formation region is absolutely
+ , / / „ / „ - O.Sl 1.00,1.10 (Hall and Griffin 1991). unstable while the fully-formed vortex street is convectively
unstable. The vortex formation region is thought to be a com-
1 1 1 i 1
plex global region of upstream and downstream propagating
10* I0J
vorticity waves, self-excitation of the flow, and modal com-
Re
petition and interaction (Rockwell, 1990). This region of self-
Fig. 16 Longitudinal vortex spacing \ld( flnfs„) as a function of Reynolds excited oscillations and flow resonance plays an important role
number Re. All of the measurements were made in the wakes of sta-
tionary and oscillating cylinders. Data points at Re = 190 and 200 cor- in frequency selection, and thus in programming and control
respond to in-line oscillations and flow perturbations with n = 2; all other of the flow separation and the overall unsteady flow. More
results correspond to cross-flow oscillations (experiments) and near- extensive stability calculations of this nature may also lead to
wake perturbations (computations) with n= 1. better understanding of bluff body wakes, and their active
control and modification by any of the means discussed here.
LOCK-ON BOUNDARY
A better understanding of passive wake control and modifi-
cation, e.g., by splitter plates and base bleed, may be achieved
in the same manner. The most recent and comprehensive dis-
cussions of absolute/convective and local/global instabilities
for spatially developing shear flows, including bluff body
wakes, are given by Huerre and Monkewitz (1990), Oertel
(1990), and Rockwell (1990).
Karniadakis and Triantafyllou (1989) conducted a linear
RECEPTIVITY BOUNDARY
stability analysis of the time-averaged flow in the near-wake
which was derived from their direct numerical simulation of
the cylinder wake at Re =100. The flow was assumed to be
locally parallel and slowly varying in the downstream direction.
The averaged flow was found to be absolutely unstable for
approximately 2.5 diameters downstream from the cylinder.
This corresponds generally to the length scale of the vortex
Fig. 17 State selection (amplitude versus frequency) diagram for lam- formation region at subcritical Reynolds numbers below 104
inar wakes. The plot should only be interpreted in a qualitative sense; (Bloor and Gerrard, 1966), a wide range of Reynolds numbers.
from Karniadakis and Triantafyllou (1989). At greater downstream distances the flow is convectively un-
stable. Thus the continuous formation of the vortex street is
thought to be sustained by the near-wake absolute instability.
For the observed range of lock-on where St/=0.3 to 1.0, the Good global agreement was found between the stability anal-
geometry of the vortex street adjusted in terms of the wave- ysis and the computation in that the Strouhal number was
length A and the frequency / to form yet again an essentially St = 0.179 in both cases. This is slightly higher than most ex-
non-dispersive street pattern, though there is some slight scatter periments (Roshko, 1954; Williamson, 1988), which may be
in the measurements. The lateral spacing of the near-wake due partly to finite grid size and the extent of the computational
vortex street also decreased with increasing St/ over this lock- domain, and to the three-dimensional effects which invariably
on range. The Reynolds number for these experiments was exist in a cylinder wake. Williamson (1988) has shown the
Re =15,000, and some complementary smooth cylinder ex- importance of the span wise variation of the cylinder wake flow.
periments were conducted at a lower Reynolds number of
Re = 3,300 and with surface roughness added to the cylinder
at the higher Reynolds number. The results were similar in Summary and Concluding Remarks
both cases. These observations for the case of active control
with rotational oscillations confirm and extend the conclusions Previous observations of vortex resonance or lock-on for
drawn earlier for both forced and unforced wakes. bluff body near-wakes which have been accumulated by nu-
merous researchers over the years now have been extended to
several additional types of imposed disturbances of both fun-
Near-Wake Flow Stability damental and practical importance—a bluff body in a per-
The introduction of the absolute/convective theory of fluid turbed incident flow consisting of a mean flow with a periodic
dynamic stability has led to a promising new approach and a component superimposed upon it, rotational oscillations of
new theoretical framework for understanding the physics of the body, and sound. And the first of these cases has been
vortex formation and near-wake flow development (Koch, shown to be fundamentally identical, under appropriate con-
1985; Triantafyllou et al., 1986,1987; Monkewitz and Nguyen, ditions, to the lock-on or vortex resonance of a cylinder os-
1987; Rockwell, 1987; Chomaz et al., 1988; Unal and Rock- cillating in line with an incident uniform flow.The vortex lock-
well, 1988a, b; Ongoren and Rockwell, 1988a, b; Karniadakis on results from the periodic flow exhibit a particularly strong
and Triantafyllou, 1989; Oertel, 1990; Huerre and Monkewitz, form of resonance, with a relative perturbation amplitude of
1990; Rockwell, 1990). A flow is said to be absolutely unstable 2AU/LOD = 0.014 resulting in a reduction in base pressure from

Journal of Fluids Engineering DECEMBER 1991, Vol. 113 / 535

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 Cpb= - 1.44 to - 1 . 8 5 , or 22 percent, for a circular cylinder fluid dynamics and bluff body flows supported by the Naval
(Armstrong et al., 1986, 1987). For a circular cylinder oscil- Research Laboratory and the Office of Naval Research. We
lating in cross-flow, as noted earlier, a peak-to-peak vibration are grateful to Dr. George Karniadakis of Princeton University
amplitude of 2a/d=0.20 to 0.30 is required to provide a com- for several helpful discussions on the spectral element com-
parable reduction in Cpt, (Stansby, 1976). Conditionally-av- putations and absolute/convective stability concepts. We are
eraged wake velocity measurements by Armstrong et al. (1987) also grateful to our colleagues who provided figures and pho-
have indicated that the strength of the vortices was increased tographs from their research, and to the referees and editors
by 29 percent and the spacing was decreased by 25 percent for who made numerous constructive and helpful comments to us.
the largest perturbation levels of their experiments.
These are quite remarkable modifications of the near-wake
References
flow for such a relatively small perturbation amplitude. Thus,
Armstrong, B. J., Barnes, F. H., and Grant, 1., 1986, "The Effect of a
seemingly small perturbations of the basic wake flow can pro- Perturbation on the Flow Over a Cylinder," Phys. Fluids, Vol. 29, pp. 2095-
duce large changes in vortex strength, base pressure and drag 2102.
on a bluff circular cylinder or other cross-section. Modification Armstrong, B. J., Barnes, F. H., and Grant, I., 1987, "A Comparison of
and control of the basic formation or instability mechanisms the Structure of the Wake Behind a Circular Cylinder in a Steady Flow With
of the wake thus can provide a means for making substantial That in a Perturbed Flow," Phys. Fluids, Vol. 30, pp. 19-26.
Barbi, C , Favier, D. P., Maresca, C. A., and Telionis, D. P., 1986, "Vortex
changes in the near-wake vortex pattern, and possibly even the Shedding and Lock-On of a Circular Cylinder in Oscillatory Flow," J. Fluid
middle- and far-wake patterns as well (Cimbala et al., 1988, Mec/i., Vol. 170, pp. 527-544.
Browne et al., 1989). Bearman, P. W., 1965, "Investigation of the Flow Behind a Two Dimensional
Model With a Blunt Trailing Edge and Fitted With Splitter Plates," J. Fluid
Karniadakis and Triantafyllou (1989) have characterized the Mech., Vol. 21, pp. 241-255.
state of both the forced and unforced vortex wakes by means Bearman, P. W., 1967, "On Vortex Street Wakes," J. Fluid Mech., Vol. 28,
of a state diagram as sketched in Fig. 17. At a small but finite pp. 625-641.
amplitude the transitions corresponding to the upper and lower Bearman, P. W., 1984, "Vortex Shedding From Oscillating Bluff Bodies,"
limits of the lock-on are given by two bounding frequencies; Ann. Rev. Fluid Mech., Vol. 6, pp. 195-222.
Bearman, P. W., andCurrie, I. G., 1979, "Pressure Oscillation Measurements
within these limits only periodic lock-on states exist. Two quasi- on an Oscillating Circular Cylinder," J. Fluid Mech., Vol. 91, pp. 661-677.
periodic regions are thought to develop at frequencies well Bearman, P. W., and Davies, M. E., 1975, "The Flow About Oscillating
above and below the lock-on regime, together with chaotic Bluff Structures," Proc. 4th Int. Conf. Wind Eff. Build. Struct., ed. K. J.
states in narrow regions immediately adjacent to the lock-on Eaton, pp. 285-295, Cambridge Univ. Press.
Bishop, R. E. D., and Hassan, A. Y., 1964, "The Lift and Drag Forces on
boundaries. These are conceptually similar in overall appear- a Circular Cylinder," Proc. R. Soc, London, Series A, Vol. 277, pp. 51-75.
ance to the vortex resonance or lock-on measurements of Blevins, R. D., 1985, "The Effect of Sound on Vortex Shedding From Cyl-
Koopmann (1967) for cross-flow oscillations of a cylinder, inders," J. Fluid Mech., Vol. 161, pp. 217-237.
reproduced here in Fig. 7, and for the in-line flow perturbations Blake, W. K., Maga, L. J., andFinkelstein, G., 1977, "Hydroelastic Variables
Influencing Propellor and Hydrofoil Singing," Noise and Fluids Engineering,
and cylinder oscillations shown in Figs. 1 and 2. ASME: New York, pp. 191-199.
As the threshold amplitude a, is approached, these finite Bloor, M. S., 1964, "Transition to Turbulence in the Wake of a Circular
regions shrink to a single frequency fe. For cross flow oscil- Cylinder," /. Fluid Mech., Vol. 19, pp. 290-304.
lations of the cylinder fe =fso while for in-line oscillations of Bloor, M. S., andGerrard, J . H . , 1966, "Measurements on Turbulent Vortices
in a Cylinder Wake," Proc. Roy. Soc, Lond., Series A, Vol. 294, pp. 319-342.
the body and periodic perturbations of the mean flow / = 2fs0 Browne, L. W. B., Antonia, R. A., and Shah, D. A., 1989, "On the Origin
as shown in Fig. 2. The small amplitude perturbations intro- of the Organized Motion in the Turbulent Far-Wake of a Cylinder,'' Exp. Fluids,
duced by Armstrong et al. (1986, 1987) correspond qualita- Vol. 7, pp. 475-480.
tively to those discussed by Karniadakis and Triantafyllou. Chomaz, J. M., Huerre, P., and Redekopp, L. T., 1988, "Bifurcations to
Local and Global Modes in Spatially Developing Flows," Phys. Rev. Let., Vol.
The relatively large amplitude cylinder oscillations and flow 60, pp. 25-28.
perturbations investigated by Koopmann (1967), Griffin and Cimbala, J. M., Nagib, H. M., and Roshko, A., 1988, "Large Structure in
Ramberg (1974, 1976) and Barbi et al. (1986) and shown in the Far Wakes of Two-Dimensional Bluff Bodies," /. Fluid Mech., Vol. 190,
Figs. 1 and 7 introduce nonlinearities and complex changes in pp. 265-298.
Ffowcs Williams, J. E., and Zhao, B. C , 1989, "The Active Control of
the near-wake flow field as shown by the flow visualization Vortex Shedding," J. Fluids Struct., Vol. 3, pp. 115-122.
studies of Griffin and Ramberg (1974, 1976), Williamson and Filler, J. R., Marston, P. L.,andMih, W. C , 1991, "Response of the Shear
Rashko (1988), and Ongoren and Rockwell (1988a, b). At the Layers Separating From a Circular Cylinder to Small Amplitude Rotational
largest amplitudes of oscillation, highly complex vortex flow Oscillations," J. Fluid Mech., Vol. 231, pp. 481-499.
Gerrard, J. H., 1966, "The Mechanics of the Formation Region of Vortices
patterns were observed; and a more extensive kaleidoscope of Behind Bluff Bodies," J. Fluid Mech., Vol. 25, pp. 401-413.
complex vortex patterns over an even wider range of frequen- Griffin, O. M., 1978, "A Universal Strouhal Number for the 'Locking-On'
cies and amplitudes was observed experimentally by William- of Vortex Shedding to the Vibrations of Bluff Cylinders," J. Fluid Mech., Vol.
son and Roshko (1988). 85, pp. 591-606.
Griffin, O. M., 1981, "Universal Similarity in the Wakes of Stationary and
Further research based upon these new analytical and com- Vibrating Bluff Structures," ASME JOURNAL OF FLUIDS ENGINEERING, Vol. 103,
putational approaches described herein is likely to lead to new pp. 52-58.
and more complete fundamental understanding of the near- Griffin, O. M., 1989, "Flow Similitude and Vortex Lock-On in Bluff Body
Near-Wakes," Phys. Fluids A, Vol. 1, pp. 697-703.
wake vortex dynamics and vortex lock-on, which until now Griffin, O. M., and Ramberg, S. E., 1974, "The Vortex Street Wakes of
have been studied mostly by using the more traditional mod- Vibrating Cylinders," J. Fluid Mech., Vol. 66, pp. 553-576.
deling approaches combined with experiments. The results dis- Griffin, O. M., and Ramberg, S. E., 1976, "Vortex Shedding From a Cylinder
cussed in this paper suggest that modification and control of Vibrating in Line With an Incident Uniform Flow," J. Fluid Mech., Vol. 75,
the basic instability or formation mechanisms of the wake by pp. 257-271.
Grinstein, F. F., Boris, J. P., Griffin, O. M., Hussain, F., and Oran, E. S.,
imposed oscillations, i.e., cross-flow, in-line and rotational, 1990, "Coherent Structure Dynamics in Spatially Evolving Near Wakes," 28th
incident flow perturbations, and an imposed sound field pro- Aerospace Sciences Meeting (Reno), AIAA Paper 90-0507.
vide a means for making substantial alterations to the near- Grinstein, F. F., Boris, J. P., and Griffin, O. M., 1991, "A Numerical Study
wake vortex pattern, and possibly to the middle- and far-wake of Passive Pressure-Drag Control in a Plane Vortex Street Wake," AIAA J.,
Vol. 29, in press.
flow patterns as well which exist relatively far downstream Hall, M. S., and Griffin, O. M., 1992, "Vortex Shedding and Lock-On in a
from the wake-generating body. Perturbed Flow," in preparation.
Hatfield, H. M., and Morkovin, M. V., 1973, "Effect of an Oscillating Free
Stream on the Unsteady Pressure on a Circular Cylinder," ASME JOURNAL or
FLUIDS ENGINEERING, Vol. 95, pp. 249-254.
Acknowledgments Huerre, P., and Monkewitz, P,, 1990, "Local and Global Instabilities in
This study was conducted as part of a research program in Spatially Developing Flows," Ann. Rev. Fluid Mech., Vol. 22, pp. 473-537.

536 / Vol. 113, DECEMBER 1991 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms

 Karniadakis, G. E., and Triantafyllou, G. S., 1989, "Frequency Selection sional Bluff Bodies," Nat. Adv. Comm. for Aero., Washington, DC, Technical
and Asymptotic States in Laminar Wakes," J. Fluid Mech., Vol. 199, pp. 441- Note 3169.
469. Roshko, A., 1955, "On the Wake and Drag of Bluff Bodies," /. Aero. Sci.,
Karniadakis, G. E., and Triantafyllou, G. S., 1990, "Direct Numerical Sim- Vol. 22, pp. 124-132.
ulation of the Three-Dimensional Vortex Street,' '28th Aerospace Sciences Meet- Sarpkaya, T., 1979, "Vortex Induced Oscillations: A Selective Review," ASME
ing (Reno), AIAA Paper 90-0113. Journal of Applied Mechanics, Vol. 46, pp. 241-258.
Koch, W., 1985, "Local Instability Characteristics and Frequency Determi- Simmons, J. E. L., 1975, "Effect of Separation Angle on Vortex Streets,"
nation of Self-Excited Wake Flows," J. Sound Vib.,, Vol. 99, pp. 53, 83. Proc. ASCE, J. Eng. Mech., Vol. 101, pp. 649-661.
Koopmann, G. H., 1967, "The Vortex Wakes of Vibrating Cylinders at Low Stansby, P. K., 1976, "Base Pressure of Oscillating Cylinders," Proc. ASCE,
Reynolds Numbers," J. Fluid Mech., Vol. 28, pp. 501-512. J. Eng. Mech., Vol. 104, pp. 591-600.
Mansingh, V., 1986, "Experimental Investigation of the Effects of Splitter Taneda, S., 1978, "Visual Observations of the Flow Past a Circular Cylinder
Plates on the Flow Behind a Rectangular Cylinder," Ph.D. thesis, Queen's Performing a Rotatory Oscillation," J. Phys. Soc. Japan, Vol. 45, pp. 1038-
University, Kingston, Ontario, Canada. 1043.
Tanida, Y., Okajima, A., and Watanabe, Y., 1973, "Stability of a Circular
Meier-Windhorst, A., 1939, "Flatterschwingungen von Zylindern im gleich- Cylinder Oscillating in Uniform Flow or in a Wake," J. Fluid Mech., Vol. 61,
massungen Flussigkeitsstrom," Mitteilungen des Hydraulischen Instituts der pp. 769-784.
Technichen Hochscule, Munchen, Heft 9, pp. 9-39. Tatsuno, M., 1972, "Vortex Streets Behind a Circular Cylinder Oscillating
Monkewitz, P. A., and Nguyen, L., 1987, "Absolute Instability in the Near- in the Direction of Flow," Bull. Res. Inst. Appl. Mech., Kyushu University,
Wakes of Two-Dimensional Bluff Bodies," J. Fluids Struct., Vol. 1, pp. 165- Vol. 36, pp. 25-37 (in Japanese).
184. Tokumaru, P. T., and Dimotakis, P. E., 1991, "Rotary Oscillation Control
Morkovin, M. V., 1964, "Flow Around Circular Cylinders. A Kaleidoscope of a Cylinder Wake," J. Fluid Mech., Vol. 224, pp. 77-90.
of Challenging Fluid Phenomena,'' ASME Symposium on Fully Separated Flows, Triantafyllou, G., Triantafyllou, M., and Chrissostimidis, C , 1986, "Stability
ASME, New York. Analysis to Predict Vortex Street Characteristics and Forces on Circular Cyl-
Oertel, H., 1990, "Wakes Behind Blunt Bodies," Ann. Rev. Fluid Mech., inders," ASME J. Offshore Mech. and Arctic Engineering., Vol. 109, pp. 148-
Vol. 22, pp. 539-564. 154.
Okajima, A., Takata, H., and Asanuma, T., 1975, "Viscous Flow Around Triantafyllou, G., Kupfer, K., and Bers, A., 1987, "Absolute Instabilities
a Rotationally Oscillating Circular Cylinder," Inst. Space and Aero. Sci. (U. and Self-Sustained Oscillations in the Wakes of Circular Cylinders," Phys. Rev.
Tokyo), Report No. 532. Let., Vol. 59, pp. 1914-1917.
Okamoto, S., Hirosi, T., and Adachi, T., 1981, "The Effect of Sound on Unal, M. F., and Rockwell, D., 1988a, "On Vortex Formation From a Cyl-
the Vortex Shedding From a Circular Cylinder," Bull. Japan Soc. Mech. Engrs., inder. Part 1. The Initial Instability," J. Fluid Mech., Vol. 190, pp. 491-512.
Vol. 24, pp. 45-53. Unal, M. F., and Rockwell, D., 1988b, "On Vortex Formation From a Cyl-
Ongoren, A., and Rockwell, D., 1988a, "Flow Structure From an Oscillating inder. Part 2. Control by Splitter Plate Interference," J. Fluid Mech., Vol. 190,
Cylinder. Part 1: Mechanisms of Phase Shift and Recovery in the Near Wake," pp. 513-529.
/. Fluid Mech., Vol. 191, pp. 197-223. West, G. S., and Apelt, C. J., 1982, "The Effects of Tunnel Blockage and
Aspect Ratio on the Mean Flow Past a Circular Cylinder With Reynolds Numbers
Ongoren, A., and Rockwell, D., 1988b, "Flow Structure From an Oscillating
Between 104 and 10 5 ," J. Fluid Mech., Vol. 114, pp. 361-377.
Cylinder. Part II: Model Competition in the Near Wake," J. Fluid Mech., Vol.
Williamson, C. H.K., and Roshko, A., 1988, "Vortex Formation in the Wake
191, pp. 225-245.
of an Oscillating Cylinder," J. Fluids Struct., Vol. 2, pp. 355-381.
Rockwell, D., 1987, "Unpublished Manuscript: A View of Hypotheses and Williamson, C. H. K., "Defining a Universal and Continuous Strouhal-Reyn-
Issues on Bluff Body Near-Wake Instabilities," Lehigh University. olds Number Relationship for the Laminar Vortex Shedding of a Circular Cyl-
Rockwell, D., 1990,''Active Control of Globally-Unstable Separated Flows,'' inder," Phys. Fluids, Vol. 31, pp. 2742-2744.
ASME International Symposium on Nonsteady Fluid Dynamics (Proceedings), Zdravkovich, M. M., 1982, "Modification of Vortex Shedding in the Syn-
FED-Vol. 92, pp. 379-394. chronization Range," ASME JOURNAL OF FLUIDS ENGINEERING, Vol. 104, pp.
Roshko, A., 1954, "On the Drag and Shedding Frequency of Two-Dimen- 513-517.

Journal of Fluids Engineering DECEMBER 1991, Vol. 113 / 537

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 07/23/2013 Terms of Use: http://asme.org/terms