Homework assignment 1

1. For a stationary, smooth circular cylinder (with D = 10 cm and L = 200 cm) in a uniform flow (with U∞ = 30
m/s),

(a) Calculate the CD for the wake velocity profile u ( y ) = U ∞ sin πy and u ( y ) = U ∞ y , respectively. a = 2.2D =
2a a
half wake width.

(b) Compare the calculated CD with the published data. Discuss the discrepancy if there is any.

2. Why is there a CL′ acting on a circular cylinder? Discuss its importance and its migitgation techniques. List at
least five techniques and explain how they work (from fluid mechanics viewpoint).

3. Explain why CL < 9 for very low spin ratio (= ωD/2U∞ < 0.5). D is the diameter of the cirulr cylinder.

4. For a smooth, stationary semi-infinate flat-plate flow with w = 30 cm (plate width), δ = 3 mm (boundary-
layer thickness), U∞ = 20 m/s, x = 30 cm (distance downstream from the leading edge of the plate), and
= 2( ) − ( ) 2 , determine the CD and compare it with the published data.
u( y) y y
U∞ δ δ


5. Water flows steadily past a porous flat plate. Constant wall suction (with Vwall = −0.3 ˆj mm/s) is applied
along the porous section. The boundary-layer velocity profile at L = 2 m (downstream from the leading edge
of the plate) is u ( y ) = 3( y ) − 2( y )1.5 . U∞ = 3 m/s, w is plate width = 1.5 m, and δ = 1.5 mm.
U∞ δ δ

(a) Find the total drag force using both Reynolds Transport Theorem and Karman momentum integral
equation. Assuming that the “densely” perforated wall covers up to L = 2 m.

(b) Compare the Ff determined with the value produced on the flat plate with no wall suction (i.e., with
= 2( ) − ( ) 2 instead). Discuss the difference(s).
u( y) y y
U∞ δ δ

1
6. Explain why θ s ≈ ±140 deg for transition boundary layer on a stationary and smooth circular cylinder in a
uniform crossflow. θs is the boundary-layer separation point.

7. Explain, based on below published St-Re curve, why the vortex-shedding wake disappears and is re-
established for ReD > 500,000. St = fsD/U∞ = S, fs = vortex-shedding frequency = n, D = cylinder diameter, and
U∞ = freestream velocity = V. ReD = ρ∞U∞D/µ = R.

8. For a tall pine tree with a diameter D and length L in a hurricane with a wind speed U, the force acts on the
tree can be larger or smaller than ½ρU2DL? Describe the flow pattern around the tree for both situations.
The pine tree can be treated as a vertical circular cylinder.

9. Explain why (from fluid mechanics point of view)

(a) For the U2 spy plane, whose purpose is to loiter for long period of time, why is the aspect ratio of its
slender wings tremendously large (AR = b2/S = 14.7, and b = wingspan = 104 ft)? Why there is a
takeoff and landing problem? Explain how this liability problem was resolved?

(b) Why does the Boeing 747 (designed to fly at M0.85) have backward swept wings (with a sweeping
angle of 37.5˚)?

(c) Why does the Concorde, which is designed to fly at M = 2.0, use a delta wing? What is the purpose of
the drooped nose? Why is the fuel burn so high compared to a Boeing 747?

10. Explain from a fluid mechanics point of view the purpose of the seams and stitches on a baseball. How
do pitchers throw curveballs and knuckleballs?

11. Explain why Canadian migrating geese fly in a Λ formation (or chevron-shape formation). Explain the pros
and cons of aircraft formation flight (i.e., vortical wake surfing).

2
12. Explain how the lift-induced drag is developed and produced both conceptually, graphically, and physically.

13. Discuss the impact of wingtip vortices on aerodyanmics, flight hazards, and airpot control, including
their control of comventional fixed-wing aircraft.

Tipi wingtip modification BERP (British Experiemntal Rotor Project)

14. Explain the impact of wingtip vortices on rotorcraft.

15. Calculate the pressure immediately above and beneath the rotor disk of a V-22 Osprey aircarft in hover.
Assuming that the flow is 1-D, inviscid and incompressible and that the meomentum theory is applicable.

16. Show the ideal induced power during vertical ascend of a helicopter with an ascend velocity VC = 1.5Vi.

17. Russian built military helicopters oftentimes are equipped with a pair of stubby wings attached to the
fuselage. Explain the role theses stubby wings and their impact on the induced power in hover.

3
18. List the advantages and disadvantaes of tiltwing aircraft as compared to tiltrotor aircraft. Discuss the
remedies.

19. Expalin why the advance ratio (Vf/Vtip) of a helicopter has to be kept below 0.3 for Vtip eqauls to, say,
M0.18. List three applicable remedies.

20. List the six fuel saver techniques (from reading assigment). Explain how they work.

21. Consider a plane vortex sheet, initially occupying the plane y = 0. Let the velocity jump across the sheet be
perfectly uniform, say from u = -Ui where y < 0, to u = +Ui where y > 0. Suppose the velocity at all times is
the uo induced by the vorticity in the diffusing sheet, and assume that both u and ζ will be forever
independent of x and z. Finally, assume that both ρ and µ may be treated as constants. Find u(y,t) and
ζ(y,t). Also, plot these functions versus y for various values of t. Find also the characteristic thickness of a
free shear layer.

Note: A similarity solution is one in which the number of independent variables is reduced by at least one, usually
by a coordinate transformation. The coordinates themselves are collapsed into dimensionless groups that
scale the velocities. Similarity solutions are limited to certain geometries and certain boundary conditions.

4
5
22. Determine δ , δ * , θ , H ,τ w , and C f via the Karman momentum integral boundary-layer equation for
steady, incompressible flow over a semi-infinite flat plate with zero pressure gradient for the following
guessed velocity distributions:

a) u = a + by b) u = a + by + cy 2
u∞ u∞
c) u = a + by + cy 2 + dy 3 d) u = a + by + cy 2 + dy 3 + ey 4
u∞ u∞
 πy 
e) u = sin  
u∞  2δ 

Compare the approximated results with the Blasius solutions.

23. Reading assignments

6
1
2
J . Fluid Mech. (1991), vol. 224, pp. 77-90 77
Printed in Great Britain

Rotary oscillation control of a cylinder wake

By P. T. TOKUMARU A N D P. E. DIMOTAKIS
Graduate Aeronautical Laboratories, California Institute of Technology 301 -46, Pasadena,
CA 91125, USA

(Received 24 July 1989 and in revised form 19 July 1990)

Exploratory experiments have been performed on circular cylinders executing forced

rotary oscillations in a steady uniform flow. Flow visualization and wake profile
measurements at moderate Reynolds numbers have shown that a considerable
amount of control can be exerted over the structure of the wake by such means. In
particular, a large increase, or decrease, in the resulting displacement thickness,
estimated cylinder drag, and associated mixing with the free stream can be achieved,
depending on the frequency and amplitude of oscillation.

1. Introduction
It has long been appreciated that the dynamics of the flow over a bluff body are
dominated by large-scale vortical structures shed in its wake. Weihs ( 1972) proposed
a methodology for predicting the structure of such a wake and found that it could
become wider, narrower, or remain the same, depending on the initial strength and
spacing of these vortices. His work was primarily aimed a t describing the wake
behind an oscillating airfoil (e.g. Bratt 1953 ; Koochesfahani 1987). Nevertheless,
Weihs’s results do suggest the potential for effective control of wakes in general.
The effects of periodic forcing on a wake were investigated by Roberts (1985), and
Roberts & Roshko (1985). They demonstrated that a significant amount of control
over the wake vortex structure and the associated mixing was possible, by pulsing
one stream relative t o the other. Okajima, Takata & Asanuma (1975) examined the
forces acting on a rotationally oscillating cylinder for Reynolds numbers, based on
cylinder diameter, in the range

4 0 < R e = - Umd
V
< 6 x lo3, (1)

where Um is the free-stream velocity far ahead of the cylinder, d is the cylinder
diameter, and u is the kinematic viscosity. His measurements were for a normalized
peak rotation rate,

of 0.2 to 1.0, and a forcing Strouhal number,

sr-uf9
=-
d
m (3)

ranging from 0.05 to 0.3, where b, is the peak rotational rate of the cylinder, ve, is the
peak circumferential velocity, and f is the forcing frequency. He noted a
78 P. T. Tokumaru and P. E. Dimotakis
‘synchronization ’ similar to that observed, for example, by Bishop & Hassan (1964),
Koopman (1967), and more recently, Ongoren & Rockwell (1988a, b), and by
Williamson & Roshko (1988) for a cylinder in transverse and in-line oscillation. Both
Okajima et al. and Bishop & Hassan reported a hysteresis in their measurements for
increasing and decreasing S,. Investigations a t comparable forcing frequencies,
amplitudes and Re were also performed by Wu, Mo & Vakili (1989).Taneda (1978)
demonstrated that in the range 30 < Re < 300, the ‘dead water’ region behind a
cylinder can be removed for a, sufficiently large(a,> 7 to 27, depending on such
factors as the cylinder span and tank width). For somewhat larger a,, the
streamlines converged downstream. Similar behaviour was also documented by
Williams & Amato (1988) using a line of unsteady pulsing jets embedded in the
trailing edge of the cylinder, at a comparable Reynolds number of 370.
I n the work to be discussed here, we examined the efficacy of forced rotary
oscillations of a (fixed axis) cylinder a t a moderate Reynolds number of 1.5x 104, for
the purpose of controlling the unsteady separated flow in its wake. The effect of the
cylinder oscillations on the structure of the wake was investigated using flow
visualization, as well as surveys of the resulting wakes using laser-Doppler velocity
measurements.

2. Experimental facility
The experiments documented here were performed in the 18 in. wide, by 19 in.
deep, low speed water channel, as well as the 20 x 20 in. free surface water channel
a t GALCIT. A 4 in. diameter Plexiglas cylinder was supported 10 in. above the
bottom of the channel, using 0.5in. thick Plexiglas fairings placed flush to the
sidewalls of the channel. Drive belts and pulleys were entirely enclosed in the fairings
and did not interfere with the flow. Rotation about the axis of the cylinder was
achieved using a high-performance servo-controlled motor, capable of tracking an
arbitrary command signal within the bandwidth and slewing rates that were
investigated. This command signal was generated by a function generator, or a
computer digital-to-analog converter output channel. Flow visualization was
accomplished by introducing dye upstream of the cylinder, or through small holes on
the cylinder surface a t midspan. Streamwise mean and r.m.s. velocity profiles were
measured using a Bragg cell frequency-shifted laser-Doppler velocimeter.

3. Results and discussion

The cylinder was rotated sinusoidally in time at a forcing frequency f and a
normalized peak rotational rate of Q, (see ( 2 ) ) ,i.e.

-ea
- - Q(t) = a, sin (2njt). (4)
2u,
SZ, was chosen such that the peak circumferential velocity of the cylinder would be
comparable with the velocity just outside the boundary layer of the cylinder
(approximately two times the free-stream velocity). We anticipated that amplitudes
of this magnitude or greater would be necessary to send the vorticity stored in the
boundary layer into the wake in a regulated manner. For these experiments,
we examined the control parameters 0,and S, in the range 0 < 52, < 16 and
<
0.17 < S, 3.3. The free-stream velocity was approximately 15 cm/s, yielding a
 Rotary oscillation control of a cylinder wake 79
- ucv)

...........................
FIGURE
1. Rotating cylinder flow geometry.

Reynolds number based on cylinder diameter of Re = 1.5x 104. I n addition, wake

mean and r.m.s. velocity profiles were measured at a single streamwise station,
located 4.5 diameters downstream of the cylinder axis. See figure 1.

3.1. Displacement thickness and drag coeficient estimation

I n order to provide a quantitative measure of the effects of forcing on the flow over
a cylinder and the resulting wake, the cylinder drag and wake displacement thickness
were estimated from wake streamwise mean and r.m.s. velocity profiles. For flow
that is two-dimensional, we can define the displacement thickness in terms of the
integral

I n this expression, 7 = y / h is the normalized vertical (cross-stream) position, u ( y ) is

the mean streamwise velocity, y and h are the vertical position and water channel
depth respectively, and Uo is the velocity in the (free-stream) region outside the
cylinder wake and water channel boundary layers. See figure 1. By analogy, we may
define the ‘displaced area’ for a three-dimensional flow, i.e.

where A is the test-section cross-sectional area, and 5 = z / b is the normalized

spanwise coordinate, with z the spanwise coordinate, and b the water channel span.
Continuity then yields for U,, the free-stream velocity far ahead of the cylinder,

(7)

Using (6) and (7) we then find

This expression is useful in that it provides information about the flow over an entire
cross-section of the water channel, while requiring only the measurement of U, and
u
r
n*
In order to remove the contribution of the initial boundary layer (in the absence
of the cylinder) to leading behaviour, it is useful to compare Uo (with the cylinder in
80 P. T. Tokumaru and P. E . Dimotakis
16

14

12

10

2. Variation of A^*/A with S, and 52,.

FIGURE

place) with the undisturbed flow velocity in the absence of the cylinder a,
a t the
same streamwise location. This yields an expression for the approximate displaced
area of the wake:
d* 4
- = - 1---. (9)
A u,
Figure 2 depicts the variation of d * / A with S, and a,. The dashed curve in this figure
denotes the j * / A level for the unforced case. The parameter space above this curve
represents an increase in A*/A over the unforced case, and below represents a
decrease. Note that the displacement area can be made substantially larger or
smaller by varying the frequency and amplitude of oscillation. The minimum for
j * / A over the range shown occurs in the neighbourhoo_dof S, x 1, and 52, z 3, and
is roughly half of the unforced case. Measurements for A*/A were taken in the range
0.17 < S, < 3.3in steps of about 0.17, and 0 < SZl < 16 in steps of 1. Each of the data
points presented in figures 2 and 3 were averaged over 2 minutes (approximately 40
natural unforced shedding cycles). Wake midspan velocity profiles were measured for
S, in the range of 0.15 to 1.4, holding SZ, fixed a t 2. Figure 3 compares the wake
profiles for several S,, and SZ, = 2.
For finite test sections and flow that is two-dimensional in the mean, one can show
that the sectional drag coefficient can be estimated by the expression (e.g. Dimotakis
1978) :

In this expression, u' and v' are the streamwise and cross-stream r.m.s., velocities and
6* is the displacement thickness defined in (5).The displacement thickness and drag
 Rotary oscillation control of a cylinder wake 81

I I
-0.11 0 2.5
-2.5
Cv-YJld

3. Cylinder wake mean velocity profiles, measured at midspan (SZ, = 2) : 0,

FIGURE unforced ;
0, S, x 0.2; A, S, x 1.

coefficient estimates summarized in figure 4 were calculated from mean midspan

velocity profiles, using ( 5 )and (lo),neglecting the fluctuating terms. As can be seen,
thcre is a broad minimum in C, and 6 * / h around S, = 1. The estimated C, for the
unforced case is a factor of six greater than for this forced case. Similarly, there is a
factor of five reduction in 6*/h. Note that the value of 6*/h for a particular S, and
52, is generally less than the corresponding value of a * / A , indicating that there is an
increase in 6*/h outside the midspan region. Note also that since C, increases with
6*/h it must also increase outside this midspan region. This spanwise variation in
6*/h is due to the interaction of the sidewalls with the shed vortices. See Roberts
(1985), Roberts & Roshko (1985),Koochesfahani (1987),and Kurosaka et al. (1988).
Another local minimum exists near S, x 0.2. The flow visualization in figure 5
suggests that this corresponds to a mode in which the cylinder releases two vortices
of opposite sign during each half-cycle. When one of these two vortices shed from the
bottom appears on top, a dip is observed in the corresponding velocity profile. See
the circles on figure 3.
We should mention that Okajima et al. (1975) observed a hysteresis with increasing
and decreasing S, and f2,= 0.2. We also observed some hysteresis for small values of
52, but none was discernible in experiments with Q1 3 2.
Finally, we suspect that had it not been for the finite height of the test section, the
wake could have been made even larger by forcing at S, < 0.2.

3.2. Flow visualization

A clear picture of the control that can be exercised over the separated flow in the
wake of the cylinder emerges from our flow visualization data. These demonstrate
the overall narrowing or widening of the wake, depending on the values of S, and Q1,
the control parameters, as well as the formation of distinct vortical structures in the
wake.
In figure 6 (plate l ) , dye is introduced ahead of the cylinder by a comb of dye
 I
82 P. T . Tokumaru and P . E . Dimotakis

0.12
O'I4 1 O O
0

0.08

0.06 1 0
0 0
0 0
'.04'F
0.02
I I I I I I I 1.6
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
SJ

c, 0.8 1 O O O

o.6
0.4
t1 0

I I I I I I I
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 6

SJ
4. Cylinder wake displacement thickness and drag coefficient variation with 8,
FIGURE
(a,= 2 ) : ( a ) 6*/h, ( b ) c,.

FIQURE
5. Forced shedding of two same-sign vortices during a half forcing cycle,
s,= 0.2, a, = 2, R~ = 1 . 5 104.
~
 Journal of Fluid Mechanics, El. 224 Plate 1

(a)

FIGURE6. Close up and full flow visualization views with and without active control.
-
(a) Unforced (non-rotating) cylinder in uniform flow;(b)S, 1, a, Z 3.

TOKUMARU
& DIMUTAKIS (Facing p. 82)
 Kotury oscillntion control of CI cylindrr iuuke 83

7 . Forred shedding a t fixed A’, = 1 (Re = 1.5 x 104). ( a ) R,

FIGITRE = 8, ( h ) R, = 16, (c) 52, = 32,
( d ) R, = 64.

injectors. I n the unforced case, figure 6 ( a ) ,the dye can be seen to be dispersed and
mixed across the full height of the test section, within a few diameters downstream
of the cylinder. I n contrast, the dye marker in the forced case corresponding t o the
minimum wake width found in 53.1, figure 6 ( b ) ,occupies approximately the same
fraction of the test section height on leaving as it did upon entering.
The data in figures 7 and 8 (recorded with dye issuing from the surface of the
cylinder) illustrate how the wake may be made wider or narrower by varying the
control paramctcrs S, and 52,. Weihs (1972) examined this phenomenon by
considering the mutual influence of the vortices in the wake and those being shed.
Although his analysis is highly idealized, considering only point vortices in potential
flow, it does provide an intuitive physical argument for the various trajectories taken
by the vortices. Additional flow visualization data (also recorded with dye issuing
from the surfacc of the cylinder), at fixed SZ, = 8 and increasing S,, arc depicted in
figures 9 and 10.
84 P . T . Tokumaru and P . E . Dimotakis

(4

FIGURE 8. Mode 11: wake structure is synchronized with the forcing frequency
(Q, = 8, Re = 1 . 5 104).
~ ( a ) S, = 0.3, ( b ) 8, = 0.5, (c) S, = 0.7, ( d ) 8, = 0.9.

Several qualitatively different vortex shedding modes were observed. They are
presented here as modes I to IV, in order of incrcasing frequency. In the forced mode
I (e.g. figure 5 ) ,the cylinder releases two vortices of the same sign per half-cycle. This
second vortex may be attributed to an additional separation caused by the most
recently generated vortex. In mode I1 (e.g. figure 8), the wake structure is
synchronized with the forced cylinder oscillation and persists beyond the end of the
test section. I n mode 111 (e.g. figure 9), the near-wake structure is also synchronized,
but becomes unstable and evolves into a structure with a lower spatial frequency
some distance downstream of the cylinder. I n mode IV (e.g. figure lo), the effect of
forcing is primarily observed in the shear layers separating from the cylinder. Note
that while the flow visualization photographs of the forced cylinder wake, a t the
largest values of S, (cf. figure 10, bottom), resemble those for the unforced case (cf.
figure l l ) ,we find that d*/A in these forced cases is still noticeably less than in the
unforced case.
 Rotary oscillation control of a cylinder wake 85

9. Mode 111: transition from wake forcing to shear layer forcing (52, = 8,
FIGURE
Re = 1.5 x 104). (a)S, = 1.1, ( b ) S, = 1.5.

10. Mode IV: primarily shear layer forcing (a,= 8, Re = 1.5 x 104).
FIGURE
(a)8,= 2.0, ( 6 ) s, = 3.3.

FIGURE1 1 . Natural (unforced) shedding. Flow visualized with dye issuing as for the data in
figures 5, 7-10.
86 I). 7’.Tokumuru and 1’. E. Dimotakis
I .o

0.8

0.6

SA

0.4

0.2

I I I
0 0.5 1 .o 1.5 2.0
s,
FrUuRE 12. \’ariation of S, with 8, (Q, % 8)

The transition between modes I1 and IV occurs gradually through mode 111.
Figures 8-10 demonstrate how the ‘breakup ’ of the synchronized structure in this
mode moves toward the cylinder with incrcasing S,, until finally only the shear layers
are forced. This breakup in the flow structure in modes I11 and IV may be
attributable t o the stability characteristics of the evolving mean velocity profiles (as
suggested by Cimbala, Nagib & Roshko 1988 for lower Reynolds number flows far
downstream), vortex coalescence. three-dimcmsional effects, or some combination
thereof.
Behaviour similar t o modes I, 11, and 111 was previously observed by Roberts
(1985), and Roberts & Roshko (1985) for the case of forcing the wake behind a
splitter plate. More generally, the bluff-body wake can be compared t o a nonlinear
oscillator (e.g. Provansal, Mathis & Uoyer 1987) which, when forced, can reveal
synchronized or phase-lockd regions in their parameter spaces. as well as regions
which may be periodic, quasi-periodic, chaotic, unstable, or even stable. See, for
example, Guekenheimer & Holmes (1983, Ch. 2). S o t e t h a t both Taneda (1978), and
Williams & Amato (1988) show stable (non-vortex-shedding) regions in their
frequency-amplitude parameter spaces.
Figure 12 shows the variation of the normalized phase speed.

with S,. h is the initial streamwise vortex spacing, measured from photographs like
those in figures 8-10. Note the rapid change in S, around S, = 1. This change is
reminiscent of Tritton’s ( 1959) high-speed and low-speed vortex shedding modes.
Our mode I1 resembles Tritton’s high-speed mode and our mode 111, his low-speed
mode.
Observations of unforced vortex shedding show t h a t vorticity is introduced into
the flow from a more-or-less steady separation point, with the resulting vortical
structures forming in thc wake of the cylinder. This is in contrast t o the forced
 Rotary oscillation control of n cylindm ioakp 87

FIGIJRK 13. Kear-wake region: forced shedding from the upper cylinder surface with AS, z 1 and
R,z 2. The sequence ( a d ) represents one forcing period. The cylinder rotational velocity is zero
in ( a ) and (I). clockwise in (6-f), and countcrcalockwise in ( g - k ) (dye was introduced ahead of
the cylinder).

shrdding process, representative of modes 1-111, which is characterized by an

unsteady separation ‘point ’. The forced shcdding procws for S, NN 1 and SZ, NN 2 is
illustrated in figure 13. The shaded areas in this figure denote fluid marked with dye
introduced just ahead of the ten o’clock position on the cylinder. This forcing
replac.es the natural shcdding mechanism with one that gathers vorticity close t o the
cylinder surface, and then releases it into the flow as a ‘vortex’. It is plausible that
this forcing mechanism will continue t o operate for much higher Reynolds numbers,
i.e. a t least as high as the drag-critical Reynolds number, Re NN RvCr> lo5, and
possibly higher.
The results reported here were recorded a t moderate Reynolds numbers, i.e. Elu =
1.5x lo4.I n an effort t o assess Reynolds-number effects, some flow visualization and
wake measurements were also performed on a cylinder whose surface was roughened
by means of 0.017 in. diameter glass beads. This roughness had no discernible effect
on thc wake flow visualization or velocity mcasurcmcnts. In addition, some flow
visualization was also performed on a smooth-surface cylinder, at a lower Reynolds
number of 3.3 x 103. The resulting flows were found to be very similar. It may be
interesting to note, however, that the effwts a t the higher forcing Strouhal numbers
were somewhat less pronounced a t this lower Reynolds number.

3.3. A note about the interpretation of linear stability analyses

Thrrc has been much interest in the use of linear stability analyses to describe
unforced bluff body flows. See Koch (1985),Triantafyllou & Chryssostomidis (1986),
Monkewitz & Nguyen (1987),Provansal et al. (1987), Chomaz, Huerre & ttcdekopp
(1988), Monkewitz (1988), and Karniadakis & Triantafyllou (1989). Using lincar
theory, it has been noted that the response to periodic forcing of the absolutely
unstable near-wake region of a bluff body will be overwhelmed by the ‘intrinsic.’
vortex shedding mode observed in the absence of external forcing. See Huerre &
Monkewitz (1985), Monkcwitz & Nguyen (1987), Chomaz rt al. (1988). and
Karniadakis & Triantafyllou (1989). This argument is based on a linearized system
and suggests that the periodic response will be overwhelmed because the transient
response grows exponentially in time while the forcing is only periodic in time. LS’ince
thc system cannot be considered linear for large excursions from the ‘base‘ state,
88 P. T. Tokumaru and P. E . Dimotakis
however, this argument can only be relied upon for infinitesimal forcing amplitudes
and short times. It is not surprising perhaps, that this interpretation of the results
of linear theory is not in accord with experiments and simulations which show that
finite-amplitude forcing of the wake of a bluff body can result in a wake structure
that is synchronous with the forcing, over a range of forcing frequencies and
amplitudes, e.g. Koopman (1967), Provansal et al. (1957), and Karniadakis &
Triantafyllou (1989).
Provansal p t al. (1987),examined the (nonlinear) Landau stability model (Landau
1944; Stuart 1958, 1960) and showed how it could be used to qualitatively describe
the conditions necessary for synchronization a t small but finite forcing amplitudes,
and Reynolds numbers in the vicinity of the onset of vortex shedding. See also
Landau & Lifshitz (1987, @26,30).Evidently, while the linear stability analysis, and
hence the notion of the bluff-body wake as a globally unstable flow with a region of
absolute instability, is useful in describing the system dynamics, given an a priori
knowledge of the final averaged flow (Karniadakis & Triantafyllou 1989), prediction
of the final dynamic state, especially in the presence of finite external forcing requires
a different approach.

4. Conclusions
In these experiments, we examined the efficacy of oscillatory cylinder rotation as
an actuation mechanism for actively controlling the cylinder wake. We conclude that
i t is indeed possible t o exercise considerable control over the cylinder wake via
oscillatory rotary forcing. Not surprisingly, our experiments show that working in a
control domain in which the structures shed are synchronous with the forcing
provides the greatest control authority over the wake structure. I n addition, while
these results were obtained for moderate Reynolds numbers, i.e. R e = 1.5 x 104,
preliminary evidence suggests that the description of the flow phenomena presented
here is qualitatively the same over a large range of Reynolds numbers. The similarity
of our results with those of others (e.g. Koopman 1967; Roberts 1985; Roberts &
Roshko 1985; Ongoren & Rockwell 1988a, b ; Williams & Amato 1988, and
Karniadakis & Triantafyllou 1989),using various bluff bodies and forcing techniques,
supports the proposition that the mechanisms by which the dynamics in the wake
can be controlled are largely generic and have more to do with the ejection of
circulation into the flow, rather than with the behaviour of the flow observed in the
absence of forcing.
Finally, we note that the type of active programme control described in this paper
relies on the regulated injection of circulation into the separated flow region, and the
subsequent large-scale vortical interactions. This type of control which exploits the
natural unsteadiness in the flow instead of reacting to i t is in contrast to the feed-
forward control employed by Licpmann, Brown & Nosenchuck (1982),and Liepmann
& Nosenchuck (1O82),and thcfeedback control employed by Ffowcs-Williams & Zhao
(1989), for example. We believe that to control high-Reynolds-number, fully
developed, separated turbulent flows, i t may prove the more efficient and effective
to programme the inevitable large-scale structures that dominate these flows. rather
than to attempt to suppress thcm. Thc large reductions in the drag coefficient and
the normalized wake displacement thickness that were achieved by the simple
control scheme employed here supports this approach.

We would like to thank Hans Hornung, Tony Leonard, Richard Miakc-Lyc, Herb
 Rotary oscillation control of a cylinder wake 89
Gaebler, Harry Hamaguchi, and Pave1 Svitek of GALCIT, as well as John Doyle of
the Dept. of Electrical Engineering at Caltech, for their help and discussions. This
research was sponsored by the Air Force Office of Scientific Research, URI AFOSR
Grant No. F49620-86-C-0134.

REFERENCES
BISHOP,R. E. D. & HASSAN,A. Y. 1964 The lift and drag on a circular cylinder oscillating in a
flow field. Proc. R . SOC.Lond. A 277, 51-75.
BRATT,J. B. 1953 Flow patterns in the wake of a oscillating aerofoil. Aero. Res. Counc. R . & M .
2773.
CHOMAZ, J. M., HUERRE,P. & REDEKOPP, L. G. 1988 Bifurcations to local and global modes in
spacially developing flows. Phys. Rev. Lett. 60,25-28.
CIMBALA, J. M., NAGIB, H. M. & ROSHKO,A. 1988 Large structure in the far wakes of two-
dimensional bluff-bodies. J . Fluid Mech. 190, 265-298.
DIMOTAKIS, P. E. 1978 Laser-Doppler velocimetry momentum defect measurements of cable drag
at low t o moderate Reynolds numbers. NCBC Rep. Contract N62583/77-M-R541.
FFOWCS-WILLIAMS, J. E. & ZHAO,B. C. 1989 The active control of vortex shedding. J . Fluids
Struct. 3, 1 1 5 1 2 2 .
GUCKENH EIMER, J. & HOLMES, P. 1983 Nonlinear Oscillations, Dynamical Systems, and Bifurcations
of Vector Fields. Springer.
HUERRE,P. & MONKEWITZ, P. A. 1985 Absolute and convective instabilities in free shear layers.
J. Fluid Mech. 159, 151-168.
KARNIADAKIS, G. E.& TRIANTAFYLLOU, G. S. 1989 Frequency selection and asymptotic states in
laminar wakes. J . Fluid Mech. 199, 4 4 1 4 6 9 .
KOCH,W. 1985 Local instability characteristics and frequency determination of self-excited wake
flows. J. Sound Vib. 99, 53-83.
KOOCHESFAHANI, M. M. 1987 Vortical pattern in the wake of an oscillating airfoil, 25th AZAA
Aerospace Sciences Meeting, 12-15 January 1987, Reno, Nevada, AZAA Paper 87-01 1 1 .
KOOCHESFAHANI, M. M. & DIMOTAKIS, P. E. 1988 A cancellation experiment in a forced turbulent
shear layer. In Proc. First Natl Fluid Dynamics Congr. 25-28 July 1988, Cincinatti, Ohio, vol.
TI, pp. 1204-1208; AZAA-88-3713-CP.
KOOPMAN, G. H. 1967 The vortex wakes of vibrating cylinders a t low Reynolds numbers. J . Fluid
Mech. 28, 501-512.
KUROSAKA, M.. CHRISTIANSEN, W. H., GOODMAN, J. R., TIRRES,L. & WOHLMAN, R. A. 1988
Crossflow transport induced by vortices. AZAA J. 26, 1403-1405.
LANDAU, L. D. 1944 On the problem of turbulence. Dokl. Acad. Nauk. SSSR 44,31 1-314.
LANDAU, L. D. &, LIFSHITZ,E. M. 1987 Fluid Mechanics, 2nd edn. Pergamon.
LIEPMANN, H. W., BROWN,G. L. & NOSENCHUCK, D. M. 1982 Control of laminar-instability
waves using a new technique. J. Fluid Mech. 118, 187-200.
LIEPMANN, H. W. & NOSENCHUCK, D. M. 1982 Active control of laminar-turbulent transition. J.
Fluid M w h . 118, 201-204.
MONKEWITZ, P. A. 1988 The absolute and convective nature of instability in two-dimensional
wakes a t low Reynolds numbers. Phys. Fluids 31, 99S1006.
MONKEWITZ,P. A. & NQUYEN,L. N. 1987 Absolute instability in the near-wake of two-
dimensional bluff bodies. J. Fluids Struct. 1, 165-184.
OKAJIMA, A,, TAKATA, H. & ASANUMA,T. 1975 Viscous flow around a rotationally oscillating
circular cylinder. Inst. Space & Aero. Sci. (University of Tokyo), Rep. 532.
ONGOREN, A. & ROCKWELL, D. 1988a Flow structure from an oscillating cylinder. Part 1 .
Mechanisms of phase shift and recovery in the near wake. J. Fluid Mech. 191, 197-223.
ONGOREN, A. & ROCKWELL, D. 1988b Flow structure from an oscillating cylinder, Part 2. Mode
competition in the near wake. .I. Fluid Mech. 191, 2 2 5 2 4 5 .
PROVANSAI,, M., MATHIS,C. & BOYER,L. 1987 BBnard-von Karman instability: transient and
forced regimes. J. Fluid Mech. 182, 1-22.
90 P . T . Tokumaru und I-’. E . Dimotakis
ROBERTS,F. A. 1985 Effects of a periodic disturbance on structure and mixing in turbulent shear
layers and wakes. Ph.11. thesis, California Institute of Technology.
ROHEKTS, F. A. & ROSHKO.A. 1985 Effects of periodic forring on mixing in turbulent shear layers
and wakes. A I A A Shear Flow Control C‘onf. 12-14 March 1985, Bvuldrr, C‘O, A I M Paper 85-
0570.
STTART.,J. T. 1958 On the non-linear mechanirs of hydrodynamic^ stability. .I. Flitid M w h . 4.
1-21.
STCART. J. T. 1960 On the non-linear mechanirs of wavc disturbancw in stable and unstable
parallel flows. J . Fluid Xech. 9, 353-370.
TABEDA. S. 1978 Visual observations o f t h e flow past a circular cylinder performing a rotatory
oscillation. J . Phys. Sot. Japan, 45. 1038-1043.
TRIAXTAFYLLOC.. ( i . S..‘rRIANTAFYLI,OI:, $1. s. & CHRYSSOSTOMIDIS. (’. 1986 On the formation Of
vortex streets behind stationary cylinders. J . Fluid Jfc~ch.170. 1 6 1 4 7 7 .
THITTON, D. .J. 1959 Experiments on the flow past a vircwlar cylinder a t low Reynolds numbers.
J . Fluid Mech. 6,547-567.
WEIIIS,D. 1972 Semi-infinite vortex trails, and their relation to oscillating airfoils. J . Fluid N e c h .
54. 679-690.
WILLIAMS. D. R. & AMATO,C. W. 1988 I‘nateady pulsing of cylinder wakes. In Proe. First Xatl
Fluid Dynamics Congr., 25-28 J u l y 1.988. Cincinatti. Ohio. 1111. 731-537 (AIAA4-88-3532-CP).
WILLIAMSON, C. H. K. & RosirKo. A. 1988 Vortex formation in the wake of an oscillating
cylinder. J . Fluids Struct. 2. 355-381.
W u , J., Mo, J. & VAKILI, A. 1989 On the wake of a cylinder with rotational oscillations. S L 4 A
2nd Shear Flow Conf., 73-16 March 1.989, Trmpc. AZ, AlL-l,4 Paper 8!)-1024.