Flexible Flapping Airfoil Propulsion

at Low Reynolds Numbers

S. Heathcote

and I. Gursul

University of Bath, Bath, BA2 7AY England, United Kingdom

DOI: 10.2514/1.25431
Q1 Water tunnel experiments onchordwise-exible airfoils heaving withconstant amplitude have beencarriedout for
Reynolds numbers of 9,000 to 27,000. A degree of exibility was found to increase both thrust coefcient and
propulsive efciency. Measurements of the ow eld revealed stronger trailing-edge vortices corresponding to
higher thrust coefcients, and weaker leading-edge vortices corresponding to higher efciencies. By analogy with a
rigid airfoil in coupled heave and pitch, thrust coefcient and propulsive efciency were found to be functions of the
Strouhal number and pitch phase angle. Propulsive efciency peaks at a pitch phase angle of 95100 deg (consistent
with experimental and computational simulations of rigid airfoils in coupled heave and pitch), and a Strouhal
number of 0.29, which lies in the middle of the range observed in nature. Thrust peaks at pitch phase angles in the
region of 110120 deg, but at higher Strouhal numbers. The results suggest the effect of chordwise exibility is
benecial for purely heaving airfoils at low Reynolds numbers. Q2
Nomenclature
a = amplitude
b = plate thickness
C
P
= time-averaged power-input coefcient
C
T
= time-averaged thrust coefcient
c = chord length
E = modulus of elasticity
h = dimensionless leading-edge amplitude (a
LE
= > c)
K = thin plate bending stiffness (Eb
3
=12)
Re = Reynolds number (U
0
c=)
s = displacement
Sr = Strouhal number (fa=U
0
2)
t = time
T = thrust; period
U
0
= freestream velocity
V = velocity magnitude
v = airfoil leading-edge velocity
x = streamwise position
y = cross-stream position
= effective angle of attack [tan
1
(v=U
0
) ]
= efciency
= bending stiffness coefcient
= viscosity
= kinematic viscosity (=)
= pitch angle
= density
! = heave angular frequency
= vorticity
Subscripts
LE = leading edge
TE = trailing edge
I. Introduction
T
HERE is a wish to build miniature aircraft known as microair
vehicles (MAVs) for which numerous applications have been
cited. One example is to relay video pictures fromthe source of a leak
on an offshore oil rig. Asecond example is to relay sound and picture
in over the hill reconnaissance situations. Some MAVs have
already been constructed [13]. It is thought that it may be possible to
approach the agility and endurance of birds and insects by adopting a
apping wing mechanism [4], and for this reason there is a need to
understand the aerodynamics of oscillating airfoils. The key
aerodynamic parameter is the Reynolds number [5]. Because the
length scale of birds, insect,s and MAVs is of the order of the
centimeter, the Reynolds number is very low, typically
Re =10
3
10
5
, and an understanding of the nature and relative
importance of inviscid and viscous phenomena in this regime is
sought.
The rst scientic explanation of bird ight was given by Knoller
[6] Q3 in 1909. He described howthe apping motion of an airfoil would
give rise to oscillatory lift and thrust forces. The work was repeated
independently by Betz [7], and published shortly afterwards in 1912.
In his ndings, Betz described the potentially positive effect of air
turbulence. Vertical oscillations of the air owing over a gliding
birds wing were shown to generate a force in the direction of ight,
thereby reducing the drag. Wind tunnel experiments by Katzmayr [8]
veried what is nowknown as the KnollerBetz effect. In the manner
of Betz, the wing in Katzmayrs experiment was stationary, and the
oscillations were of the air. In 1936, Garrick [9] published a set of
equations for the thrust and efciency of a at plate oscillating in
inviscid ow. Pure heave, pure pitch, and coupled heave and pitch
motions were modeled, although the equations are valid only in the
limit of small amplitude oscillations. Other key inviscid studies were
by Wu [10], Lighthill [11], and Chopra [12], the last of whom
extended Garricks theory to oscillations of arbitrary amplitude. An
inviscid panel method was developed and applied by Jones et al. [13]
to a heaving NACA0012 airfoil. Jones et al. [13] found a remarkable
similarity between the ow patterns predicted by the panel method
and those observed in experiments, illustrating how the formation of
a reverse von Krmn street for an oscillating airfoil is essentially an
inviscid phenomenon. These inviscid methods are valid in the case of
high Reynolds number and small local (or effective) angles of attack.
At the lowReynolds numbers of birds, insects, sh, and MAVs, ow
separation tends to occur at the leading edge [14], especially at high
angles of attack. Inviscid methods are not capable of predicting such
ow separation effects and, for this reason, NavierStokes codes
have been developed. These methods are able to predict leading-edge
ow separation, vortex formation and shedding, and the consequent
Presented as Paper 1405 at the AIAA Aerospace Sciences Meeting and
Exhibit, Reno, NV, 912 January 2006; received 26 May 2006; revision
received 13 September 2006; accepted for publication 13 September 2006.
Copyright 2006 by Sam Heathcote and Ismet Gursul. Published by the
American Institute of Aeronautics and Astronautics, Inc., with permission.
Copies of this paper may be made for personal or internal use, on condition
that the copier pay the $10.00 per-copy fee to the Copyright Clearance Center,
Inc., 222 Rosewood Drive, Danvers, MA 01923; include the code $10.00 in
correspondence with the CCC.

Department of Mechanical Engineering.

AIAA JOURNAL
1
merger of the leading-edge vortices into the trailing-edge vortex
system. The propulsive efciency of oscillating airfoils has been
shown to depend greatly on these effects, and in general to be rapidly
degraded by the onset of leading-edge ow separation [1518].
These NavierStokes simulations show that efciency is degraded
by the formation of leading-edge vortices. For example, Tuncer and
Platzer [16] observed the transition from weakly separated leading-
edge ow to the shedding of large scale leading-edge vortices to
correspond to a sharp decline in efciency. The diminished
efciency was attributed to both a rise in power input (because of the
suction effect of the leading-edge vortex) and a fall in thrust (because
of the disruption to the reverse vortex street caused by the arrival of
the leading-edge vortex). Peak efciencies occur at angles of attack
sufciently small that such a phenomenon occurs only very weakly.
The airfoils in the majority of experimental studies have been
rigid. Of these studies, the majority have been carried out at low
Reynolds numbers, although a number have modeled hovering ight
[1921] (U
0
=0). A number of experimental techniques [2226]
have been used, including force balance measurements of thrust and
efciency, particle image velocimetry and molecular tagging
velocimetry measurements of the velocity eld, and dye and smoke
visualization. A variety of different oscillating motions have been
investigated, including harmonic heave [13], harmonic pitch [27],
coupled heave and pitch [22], and also alternative displacement
waveforms to the cosine wave (e.g., triangle waveform) [28].
Introducing a pitch motion to the heave of rigid airfoils has been
found to yield potentially higher efciencies. The mechanism has
been found to be the diminishing of the leading-edge vortex [16,29]
through a lower effective angle of attack. The coupled heave and
pitch motion is of interest in studies of nature, particularly in the
thunniform swimming [30] of, for example, tuna, swordsh, and
dolphins.
If we are to be guided by nature in the design of miniature aircraft,
then it becomes necessary to study birds and insects in more detail.
One question is of the stiffness of their wings. Studies of insect wings
by Wootton [31] and Steppan [32] have revealed intricate variations
in their stiffness. However, although areas of the wing have been
identied as being specically exible [31], how this benets the
insect aerodynamically is still not clear [33]. Flexibility is an
interesting subject in the design of MAVs, because in addition to any
aerodynamic benets it is noted that exible wings are inherently
light. The subject of exibility in the arena of oscillating airfoils is
relatively unexplored, however, despite the knowledge that it is of
importance in bird and insect ight, and essential in sh swimming
[34]. Numerical models have indicated higher efciencies for
exible airfoils [3537], though the computational complexity of the
situation requires assumptions either of inviscid ow [35] or of a
predened exing motion [37,38]. Experimentally, there are
ndings of higher thrust at zero freestream Reynolds number (the
special case of hovering ight) [39,40], and of higher efciency at
nonzero Reynolds numbers [41]. A systematic study of the effect of
airfoil deformation on the ow eld and force coefcients is absent
from the literature, however.
The aim of the present study is to examine the ow-structure
interaction, thrust, and efciency of a exible 2-D airfoil oscillating
in heave at low Reynolds numbers. A chordwise-exible airfoil
(Fig. 1) is oscillated in heave with constant amplitude. It is noted that
the shape of the airfoil follows the observations of Wootton [31] that
birds wings have relatively stiff leading edges, and resembles the
airfoil of the MAV of Jones et al. [42], and in some ways the tadpole
in the CFD study of Liu et al. [43]. Unlike experimental studies of
rigid airfoils in coupled heave and pitch, the pitching motion arises
naturally through chordwise exibility.
Four appropriate parameters are Re, h, St, and . The Q4 Reynolds
number is based on the chord length of the airfoil. The dimensionless
heave amplitude h =a=c is based on the amplitude of the leading
edge. The Strouhal number is based on the peak-peak amplitude of
the leading-edge. This denition is found throughout the literature
[28,29,41]. Physically, for an airfoil in pure heave, the Strouhal
number is related to the effective angle of attack. The parameter has
been shown to be signicant in the study of the wake pattern from a
heaving airfoil [24], and in particular whether the wake pattern is
characteristic of net thrust, drag, and lift [13]. It has also been found
to have signicance in nature, where the Strouhal numbers of insects,
birds, bats, and sh are observed to lie within a narrow range [44] of
0:2 <Sr < 0:4. The bending stiffness parameter is dened following
the approach of Murray [45] as
=
K
1
2
U
2
0
c
3
II. Apparatus
Experiments were carried out in water with two airfoils: a 400 mm
span, 100 mm chord, NACA0012 airfoil, inexible in both the
spanwise and chordwise directions, and a 300 mm span, 90 mm
chord, teardrop/at plate design (Fig. 1), exible in the chordwise
direction only. The plate was a sheet of steel of uniform thickness of
modulus of elasticity 2:05 10
11
N=m
2
and length 60 mm. The
teardrop element was of chord 30 mm, thickness 10 mm, and solid
aluminum. The purpose of the experiments on the NACA0012 airfoil
was to validate the force measurements. The second, teardrop/plate,
airfoil, allowed for a study of chordwise exibility by varying the
thickness of the carbonmanganese steel plate. Experiments were
carried out on 7 plates, identical apart from their thickness. The
bending stiffness of the plates relative to the most exible airfoil,
b=c =0:56 10
3
, is shown in Table 1.
Experiments were conducted in a free-surface closed-loop water
tunnel (Eidetics model 1520) with a 381 mm wide 508 mm deep
test section and ow speed range of 00:45 m=s (see Fig. 2).
Experiments were performed for tunnel speeds of 0:1 m=s, 0:2 m=s,
and 0:3 m=s corresponding to Reynolds numbers based on the chord
length of the 90 mm airfoil of 9,000, 18,000, and 27,000,
respectively. The free streamvelocity was measured with a TSI Q5 laser
Doppler velocimetry system in back scatter mode. The airfoil was
mounted vertically with one end attached to a horizontal shaker
(Motovario 0.37 kW three-phase motor, 5:1 worm gear and IMO
Jaguar controller). Perspex splitter plates were positioned 2 mm
above and below the airfoil. The displacement of the leading-edge
was given by s
LE
=a
LE
cos(!t). The displacement of the airfoil was
measured with a position sensor attached to the output shaft of the
worm gear. Tests were carried out for frequencies in the range of
0.292.54 Hz. Experiments on the effect of chordwise exibility
were carried out at constant amplitude, a =17:5 mm. Force
validation experiments on the NACA0012 airfoil were carried out for
amplitudes in the range 12:5 mm < a <20 mm. A binocular strain
gauge force balance [46], machined from aluminum, was used to
Fig. 1 Schematic of the exible airfoil heaving periodically in the
vertical direction.
Table 1 Bending stiffness of plates relative to the most exible airfoil,
b=c 0:56 10
3
.
Plate thickness, b=c Relative stiffness, =
0
0:56 10
3
1
0:84 10
3
3.4
1:12 10
3
8
1:41 10
3
15.6
1:69 10
3
27
2:25 10
3
64
4:23 10
3
422
2 HEATHCOTE AND GURSUL
make direct force measurements in directions parallel and
perpendicular to the freestream velocity.
III. Experimental Methods
A. Force Measurement
The forces applied to the wing in the x and y directions, F
x

and F
y
, were measured with a binocular strain gauge force balance.
The force F
x
is equal to the drag (or thrust) on the wing. The force F
y
is equal to the lift L on the wing, plus a contribution arising from
the inertia of the wing. This contribution is proportional to the mass
m of the wing and the wing acceleration in the y-direction (dv=dt),
where v is the instantaneous plunging velocity. Hence, total force
measured is F
y
=L mdv=dt and the power input can be
calculated as Lv =(F
y
mdv=dt)v. On integrating with time over a
complete heave cycle, the term v dv=dt vanishes because the phases
of v and dv=dt differ by 90 deg. Hence the inertia of the airfoil does
not contribute to the time-averaged power input. This approach of
using F
y
for the input power calculations was validated as will be
discussed later in the paper.
Driving force and thrust force data were collected for 60
oscillations (sample rate 1 kHz) for each test condition. The thrust
coefcient is given by
C
T
=
T
1
2
U
2
0
c
where, T is the time-averaged thrust per unit span. The time-
averaged power input is given by
C
P
=
F
y
v
1
2
U
3
0
c
where, F
y
v is the instantaneous power input, and the overbar
denotes an average over time. The propulsive efciency is given by
=
TU
0
F
y
v
=
C
T
C
P
B. Force Measurement Uncertainty
Since the uid dynamic forces acting on the wing may vary along
the span, it is important that the force balance is sensitive only to
force, and insensitive to the distance from the gauge at which the
force acts. Insensitivity to bending moment is a key feature of a
binocular strain gauge force balance. It was found during the
calibration tests that the uncertainty in F
x
and F
y
induced by a typical
range of bending moments was less than 0.5%. Coupling effects,
ctitious indicated forces in the x direction due to loading in the y
direction, and vice versa, were found to give rise to a 2%uncertainty.
The uncertainty due to torque about the z axis was found to be 1%.
The uncertainty from temperature variations was found to be
negligible. The response of the gauge was found to be linear (1%
uncertainty) over the experimental range. The combined uncertainty
in the strain gauge readings is approximately 3%. Since the efciency
readings depend on readings of both F
x
and F
y
, the uncertainty in the
efciency data is approximately 6%.
C. Force Measurement Validation
To validate the force measurement system, a set of thrust and
power-input measurements were carried out for a 100 mm chord,
400 mm span, NACA0012 airfoil, oscillating with constant
amplitude (h =0:175) between splitter plates. Tests were carried out
for Reynolds numbers of 10,000, 20,000, and 30,000, and for a
frequency range of 0 <Sr < 0:72. The thrust coefcient, power-
input coefcient, and propulsive efciency data are plotted in
Figs. 3a, 3b, and 4, respectively. Also shown for comparison are the
predictions of Garrick theory [9], a panel method [47], and a viscous
NavierStokes code [48] (run for Re =20; 000, M=0:05, laminar
ow). To match the parametric range of the present experimental
data, additional calculations were performed by Q6 Young.

It is seen fromFig. 3a that the NavierStokes predictions for thrust

coefcient are in very close agreement with the experimental values
over the complete frequency range. Furthermore, close agreement is
found between the measured drag coefcient (C
D
=0:028) and the
values found in previous experiments by Sheldahl and Klimas [49]
(C
D
=0:0245) and Koochesfahani [27] (C
D
=0:027). A number of
additional observations may be made in Fig. 3a. First, the
experimental data show the effect of Reynolds number to be very
small for the range 10; 000 < Re < 30; 000. Second, the panel
method predicts marginally higher thrust coefcients than Garrick
theory, consistent with the ndings of Jones et al. [50]. Both inviscid
methods overestimate the experimentally measured thrust
coefcient.
It is seen in Fig. 3b that close agreement is observed between the
experimental and NavierStokes predictions for power-input
coefcient. Here again, the effect of Reynolds number is observed
to be small for the experimental data. Garrick theory and the panel
method are seen to underestimate the power-input coefcient,
particularly severely in the case of Garrick theory. Young [48]
showed the key limitation of the panel method to be the absence of a
model of leading-edge vorticity shedding; leading-edge vortices in
the NavierStokes model were shown to augment the surface
pressures, leading to a greater power-input requirement. This
limitation is shared by Garrick theory, which is further aficted, as
described by Jones et al. [50], by an overly simplistic model of the
shedding of trailing-edge vorticity.
Propulsive efciency, the ratio of thrust coefcient to power-input
coefcient, is shown in Fig. 4. Although the panel method correctly
predicts a decrease in efciency with increasing frequency (Garrick
theory predicts the efciency to tend to a value of 0.5), both inviscid
methods are seen to signicantly overestimate efciency. It is seen
from Fig. 3 that the inaccuracy of the inviscid methods results
primarily from the power-input coefcient predictions. Reasonable
agreement is observed between the experiments and NavierStokes
simulations. In particular, close agreement is found in the trend
toward drag (negative efciency) at low frequencies, the peak
efciency ( ~30%), and the optimum frequency (Sr ~0:1).
D. Displacement
The oscillating wing was lmed with a 50 frames-per-second
high-shutter-speed video camera (see Fig. 2). Specialist motion-
tracking software (RealViz MatchMover Pro 3.1) returned the
coordinates of the leading edge and trailing edge in each frame. After
Fig. 2 Water tunnel experimental setup for force measurement,
particle image velocimetry, and measurement of the displacements of the
leading and trailing edges.

Young, J., private communication.

HEATHCOTE AND GURSUL 3
the initial transients, the positions of the edges (and the pitch angle
of the airfoil) were recorded over four oscillations. The phase and
amplitude of the leading-edge, the trailing-edge, and the pitch angle,
were found with a cosine t.
E. Particle Image Velocimetry
Particle image velocimetry measurements were taken with a TSI
PIV Q7 system. Nylon particles of 4 mdiameter were usedfor seeding.
The PIV system comprised an 8-bit gray scale digital camera, dual
mini Nd:Yag 120 mJ pulsed laser, remote focusing apparatus,
synchronization unit, SCSI image acquisition card, dual Xeon
processor computer, and Insight 6 image processing software.
Images were captured at a rate of 3.75 pairs per second, and with a
resolution of 2048 2048 pixels. The separation of laser pulses was
500 s. A fast-Fourier-transform algorithm was used to analyze the
image pairs. Awindowsize of 32 32 pixels with 50%overlapping
was chosen, to give 127 127 velocity vectors with spatial
resolution 0.016c. Time-averaged ow elds were obtained by
taking an average of 256 velocity elds. Phase-averaged ow elds
were obtained by taking an average of 200 velocity elds. The
uncertainty in the velocity measurements was estimated to be around
2%.
IV. Results
The following sections present the results for the chordwise-
exible airfoil, where the effect of airfoil stiffness is investigated.
Thrust coefcient and efciency are studied rst as functions of plate
thickness, Strouhal number and Reynolds number and then,
following analysis of the deformation characteristics of the airfoil, as
functions of pitch phase angle, Strouhal number and Reynolds
number. Optimum pitch phase angles for thrust and efciency are
found, and compared with those in the literature. Flow elds are
shown throughout, to illustrate how trends in thrust and efciency
relate to changes in the ow pattern.
It is found that the thrust coefcient of a heaving airfoil increases
rapidly with Strouhal number. According to Garrick [9], the thrust
coefcient of a at plate in pure heave (valid in the limit of small
amplitude oscillations) is given by
C
T
=
3
Sr
2
(F
2
G
2
)
where, Fand Gare the real and imaginary parts of the Theodorsen lift
deciency function. For this reason the quantity C
T
=Sr
2
is
sometimes chosen in preference to C
T
alone [28]. An alternative
method is to dene a thrust coefcient Q8 T=(
1
2
v
2
p
S) C
T
=(Sr)
2
,
where v
p
is the peak heave velocity of the airfoil [15]. This
approach has been taken in studies of hovering ight [19,39]
(U
0
=0), where the conventional formulation of thrust coefcient is
undened. From here on, the quantity C
T
=Sr
2
will be plotted.
A. Thrust Effect of Airfoil Stiffness
The remainder of this paper focuses on the effect of chordwise
exibility. All experimental results fromthis point forward are for the
chordwise-exible airfoil shown in Fig. 1. The variation of thrust
coefcient divided by the square of the Strouhal number C
T
=Sr
2
with plate thickness is plotted in Fig. 5. Plots are shown for
a) Re =9; 000, b) Re =18; 000, and c) Re =27; 000 (the zero
Reynolds number case has been studied before [39]). Measurements
are taken for seven plates of different thickness. For each Reynolds
number, and for each Strouhal number, a peak in the thrust
coefcient is observed. The peak moves to higher values of plate
thickness with increasing Strouhal number. For high Strouhal
numbers, the optimum thickness may be exceeded without a severe
degradation in performance, although a sharp decline in thrust
coefcient occurs if the airfoil is too thin. The results indicate that
exibility improves the thrust performance of the airfoil at low
Reynolds numbers. Note that, although the Strouhal number range
decreases with increasing Reynolds number, the oscillation
frequencies, in Hertz, are the same for parts a), b), and c).
Fig. 3 Comparison of present experimental data with linear theory,
panel method, and NavierStokes solver; NACA0012 airfoil in pure
heave (except Garrick, thin airfoil), h 0:175: a) thrust coefcient;
b) power-input coefcient.
Fig. 4 Comparison of propulsive efciency data with linear theory,
panel method, and NavierStokes solver; NACA0012 airfoil in pure
heave (except Garrick, thin airfoil).
4 HEATHCOTE AND GURSUL
The features of the wake for rigid and exible airfoils may be
illustrated with example velocity and vorticity elds for Re =9; 000,
Re =18; 000, and Re =27; 000. Three velocity elds are shown in
Fig. 6 for Re =9; 000. The outline of the airfoil is drawn, and
velocity vectors in the plane of the midspan of the airfoil are shown.
The position of the leading edge in each part, s =a, corresponds to
a time t=T =0. The three parts in Fig. 6 correspond to three different
plate thicknesses in the Sr =0:56 curve of Fig. 5a. Differences exist
between parts a), b), and c) in the shape of the airfoil and in the
velocity eld. In the case of the stiffest plate (Fig. 6a), the pitch angle
of the airfoil, dened by the angle between the freestream and a
chordline drawn from the leading edge to the trailing edge, is less
than 1 deg; the airfoil is essentially rigid. The pitch amplitudes of the
intermediate [part b)] and highly exible airfoil [part c)] are 6 and
17 deg, respectively. As expected, the pitch amplitude increases with
plate exibility. A more subtle difference exists between the airfoils
in parts b) and c): whereas for the case of intermediate stiffness the
airfoil pitches down at t=T =0, the reverse is true of the most exible
airfoil. In this way it is seen that the stiffness of the plate affects both
the pitch amplitude and the pitch phase angle. It is shown later that a
relationship exists between pitch phase and pitch amplitude. The
thrust coefcient is greatest for the airfoil of intermediate exibility
(Figs. 5a and 6b); the vortices from this airfoil are stronger, and
further apart in the lateral (y) direction. It is noted that vortex pairs
(vortex dipoles) are observed for the stiffest and intermediate airfoils.
Time-averaged velocity magnitude contours for the three airfoils
are shown in Fig. 7. The peak jet velocity and jet width are seen to be
highest for the airfoil of intermediate plate thickness, consistent with
Fig. 5 Thrust coefcient as a function of dimensionless plate thickness:
a) Re 9; 000; b) Re 18; 000; c) Re 27; 000.
Fig. 6 Instantaneous velocity elds at t=T 0 for Re 9; 000,
Sr 0:56: a) b=c 4:23 10
3
; b) b=c 1:41 10
3
;
c) b=c 0:56 10
3
. The time-averaged thrust coefcient is greatest in
part b).
HEATHCOTE AND GURSUL 5
the force data. The jets from the intermediate and greatest stiffness
airfoils lie at an angle to the freestream direction. The vortices in
these two cases are shed in pairs. Such deected jets have been found
to occur when the Strouhal number exceeds a critical value [13].
Vorticity elds for three airfoils of different plate thickness are
shown in Fig. 8 for Re =18; 000. The leading edge of the airfoil is
moving upwards through the origin (t=T =
3
4
). Each phase-averaged
vorticity eld was calculated froma series of 200 velocity elds, each
captured at the same point in the oscillation. Clockwise vorticity is
shown white, counterclockwise vorticity black. It is noted that,
whereas vortex pairs were observed in the Re =9; 000 case, a
reverse vortex street, parallel to the freestream, is observed in the
present case. This is attributed to the lower Strouhal number. The
highest thrust coefcient occurs for the airfoil of intermediate
stiffness [part b)]. It is seen that the vortices from the intermediate
airfoil are stronger, and spaced further apart in the lateral direction
than those from the stiffest airfoil. Although a large distance in the
lateral direction separates the vortices fromthe least stiff airfoil, their
strength is considerably lower than for either of the other two airfoils.
Velocity magnitude and vorticity contour elds for the three
airfoils are shown in Fig. 9 for Re =27; 000. Again, the leading edge
is moving upwards through the origin. Gray levels in Fig. 9 indicate
velocity magnitude; white lines indicate contours of vorticity (solid
for clockwise ow, dotted for counterclockwise ow). The highest
thrust coefcient occurs for the airfoil of intermediate stiffness
[part b)]. In this case it is seen that a large lateral distance separates the
regions of opposite vorticity. It is noted that the vortices from the
most exible airfoil are shed almost in a line. The measured thrust
coefcient C
T
=0:04 is correspondingly small.
B. Thrust Effect of Oscillation Frequency
Plots of thrust coefcient against Strouhal number are shown in
Fig. 10 for ve plate thicknesses. Plots are shown for a) Re =9; 000,
b) Re =18; 000, and c) Re =27; 000. With the appropriate choice
of airfoil, benets to exibility are observed over the complete
Reynolds and Strouhal number range. Peaks in thrust coefcient are
observed in some cases, although not as markedly as in the plots of
thrust coefcient against plate thickness. The peaks are seen to move
to higher Strouhal numbers with increasing plate stiffness. For all
Reynolds numbers, a transition from drag to thrust occurs for the
stiffest, essentially rigid, airfoil, at a Strouhal number Sr ~0:17, and
earlier for the exible airfoils.
Instantaneous vorticity contour plots corresponding to the b=c =
0:85 10
3
curve in Fig. 10a are shown in Fig. 11. Note that the
vorticity is normalized with frequency as =!to explore the effect of
Strouhal number. The stiffness of the airfoil is the same for each plot,
and the frequency is incremented from part a) to part c). The leading
edge is moving upwards through the origin in each part. As expected,
the pitch angle of the airfoil at this instant increases with increasing
Strouhal number. The pitch angle of the airfoil undergoing the lowest
frequency oscillation is around 1 deg. The vortices in the
intermediate frequency case (maximumthrust coefcient) are seen to
be stronger, and separated by a greater lateral distance, than the
vortices of the low frequency case. In the high frequency case
(Fig. 11c), vortices are shed in pairs. This is commensurate with the
higher Strouhal number effects [13]. It is seen that the jet is deected
slightly away from the freestream direction.
Q9 Time-averaged velocity magnitude plots for the three Strouhal
numbers in Fig. 11 are shown in Figs. 12b, 12d, and 12f . Additional
Fig. 7 Time-averaged velocity elds corresponding to Fig. 6.
Fig. 8 Phase-averaged vorticity elds for three airfoils; Re 18; 000,
Sr 0:26, 200 capture average: a) b=c 4:23 10
3
;
b) b=c 1:13 10
3
; c) b=c 0:56 10
3
. The thrust coefcient is
greatest in part b).
6 HEATHCOTE AND GURSUL
frequencies are plotted in Figs. 12a, 12c, and 12e. The shading
indicates the velocity magnitude normalized by the peak leading-
edge heave velocity. The difference between the maximum jet
velocity and the velocity of the surrounding owis relatively small at
low Strouhal numbers, but becomes larger with increasing Sr. The
strength of the jet is seen to grow with frequency, reach a maximum
around Sr =0:410:48, and then decay. The frequency for which
the strongest jet is observed corresponds to the peak in thrust
coefcient (Fig. 12d). In Figs. 12e and 12f, a deected jet (seen in the
instantaneous vorticity eld of Fig. 11c) is characteristic of the high
Strouhal number effects mentioned earlier.
C. Efciency
Propulsive efciency is plotted as a function of Strouhal number in
Fig. 13 for a) Re =9; 000, b) Re =18; 000, and c) Re =27; 000.
Cases of negative efciency (drag) are not shown. Curves for ve
plate thicknesses are shown to illustrate the effect of airfoil stiffness.
It is seen that the efciency of the exible airfoils is signicantly
higher than the efciency of the stiffest, essentially rigid airfoil.
Hence exibility is seen to offer signicant efciency benets, in
addition to increases in thrust. For the Re =9; 000 case, the
propulsive efciency at low Strouhal numbers increases with
increasing exibility. It is possible that a still more exible airfoil
Fig. 9 Q13 Phase-averaged velocity magnitude (gray scale) and vorticity
(clockwise: solid white lines; counterclockwise: dotted white lines; rst
level: =!0:85; increment: =!1:9) for three airfoils;
Re 27; 000, Sr 0:17: a) b=c 4:23 10
3
; b) b=c 1:13 10
3
;
c) b=c 0:56 10
3
. The thrust coefcient is greatest in part b).
Fig. 10 Thrust coefcient as a function of Strouhal number:
a) Re 9; 000; b) Re 18; 000; c) Re 27; 000.
HEATHCOTE AND GURSUL 7
may yield even higher efciencies. At the highest Strouhal number, it
is one of the intermediate stiffnesses that yields the greatest
efciency. This trend is observed for all Reynolds numbers. Afurther
trend is for the efciency to move from positive to negative for the
rigid airfoil at a Strouhal number of approximately 0.17 for each
Reynolds number (consistent with the change from negative to
positive thrust), and for the transition to positive efciency to occur
earlier for the exible airfoils.
As discussed earlier, the efciency of oscillating airfoils is closely
related to the ow near the leading edge. Flow separation has been
shown to lead to signicantly reduced propulsive efciencies. It is
therefore of interest to study the ow at the leading edge of the
teardrop/plate airfoil, and to make a comparison between the rigid
and exible cases. The ow pattern over the airfoil for one complete
oscillation is shown in Fig. 14 for times t=T =0,
1
4
,
1
2
,
3
4
. The method is
of dye owvisualization, and dye is released into the owone chord
length upstream of the leading edge. It is noted that the Reynolds
number in Fig. 14 is chosen to yield the clear ow visualization
images, and is lower than the Reynolds number of the thrust and
efciency measurements. However, the Strouhal number is typical of
those in the force measurements. The airfoil is moving upwards
through the origin in part a), reaches the highest displacement in
part b), returns through the origin in part c), and reaches the lowest
displacement in part d). The formation of both leading-edge and
trailing-edge vortices is visible. Leading-edge vortices are observed
to be shed in pairs, to convect downstream and be swept into the
trailing-edge vortex pattern. Similar vortex shedding patterns have
been simulated numerically by Lewin and Haj-Hariri [18], who
found that although leading-edge vortices may be shed in pairs, the
vortices of the pair tend to be unequal in strength. The manner in
which the leading-edge vortices affect the efciency has been
explained in terms of the perturbation to the pressure distribution
over the surface of the airfoil [48] and the disturbance they cause to
Fig. 11 Instantaneous vorticity elds for an airfoil oscillating at three
frequencies; Re 9; 000, b=c 0:85 10
3
: a) Sr 0:27; b) Sr 0:48;
c) Sr 0:97. The thrust coefcient is greatest in part b).
Fig. 12 Time-averaged velocity elds for a single-stiffness airfoil
oscillating at six frequencies; Re 9; 000, b=c 0:85 10
3
:
a) Sr 0:20; b) Sr 0:27; c) Sr 0:41; d) Sr 0:48; e) Sr 0:88;
f) Sr 0:97.
8 HEATHCOTE AND GURSUL
the trailing-edge vortex system [18]. It has been shown that
minimizing the strength of the leading-edge vortices is necessary to
achieve high efciency, and this is achieved through minimizing the
effective angle of attack [15,17,29]. However, such small angles of
attack tend to yield low thrust coefcients and a balance may
therefore be sought between high efciency and high thrust
[17,29,51]. For a given Reynolds number and Strouhal number, it is
interesting to compare the vortex shedding patterns for the airfoil
stiffness that maximizes efciency with the stiffness that maximizes
thrust coefcient. Such a comparison is made in Fig. 15, where the
vorticity patterns of two airfoils are shown over one heave cycle, for
Re =9; 000, Sr =0:34. The two columns of images show vorticity
magnitude elds at the same points in the cycle as Fig. 14. The rst
Fig. 13 Propulsive efciency as a function of Strouhal number:
a) Re 9; 000; b) Re 18; 000; c) Re 27; 000.
Fig. 14 Dye ow visualization of vortices; Re 1; 800, Sr 0:29,
b=c 4:23 10
3
: a) t=T 0; b) t=T
1
4
; c) t=T
1
2
, d) t=T
3
4
.
HEATHCOTE AND GURSUL 9
column of images is for the airfoil that experiences the greatest thrust
coefcient. The second column of images is for a more exible
airfoil, which experiences the greatest efciency. The formation and
shedding of leading-edge vortices is visible in both cases, although
those of the more exible airfoil (column II) are observed to be
weaker, both at the point of formation and further downstreamwhere
they are swept into the trailing-edge vortex pattern. This is consistent
with the higher efciency. In contrast, the trailing-edge vortex
pattern is moderately stronger for the stiffer airfoil (column I). It is
also noted that, as observed in the simulations of Lewin and Haj-
Hariri [18], leading-edge vortices may be shed in pairs, with one
dominant in strength.
D. Airfoil Shape Characteristics
A possibly more physically meaningful approach is to plot thrust
coefcient against a shape characteristic of the airfoil, rather than
against plate thickness. Pitch amplitude, pitch phase angle, and
Fig. 15 Instantaneous vorticity; Re 9; 000, Sr 0:34.
I) b=c 0:85 10
3
, maximizes thrust; II) b=c 0:56 10
3
, max-
imizes efciency: a) t=T 0; b) t=T
1
4
; c) t=T
1
2
; d) t=T
3
4
.
Fig. 16 Leading-edge displacement, trailing-edge displacement, and
deformation as a function of time; Re 9; 000, b=c 0:56 10
3
,
Sr 0:34.
Fig. 17 The variation with Strouhal number of a) pitch amplitude;
b) pitch phase angle; and c) tailing-edge amplitude, Re 9; 000.
10 HEATHCOTE AND GURSUL
trailing-edge amplitude are possible parameters. A displacement-
time plot for the single case Re =9; 000, b=c =0:56 10
3
, Sr =
0:34 is plotted in Fig. 16. The y coordinates (see Fig. 1) of the leading
and trailing edges, and of the difference between them, s
LE
s
TE
, are
plotted as functions of time. It is seen that the trailing-edge trails the
leading-edge in phase, whereas s
LE
s
TE
is seen to lead the leading
edge. FromFig. 1, it is seen that the tangent of the pitch angle is equal
to s
LE
s
TE
divided by the streamwise separation of the leading and
trailing edges. The pitch phase angle is found by nding the phase of
this new quantity relative to the leading edge. In the particular case
illustrated in Fig. 16, the pitch phase angle is found to be 92.2 deg.
The nding of a pitch phase angle of this value is of interest because
phase angles of 90 deg have been found in the literature to optimize
efciency [15,17].
Pitch amplitude, pitch phase angle, and trailing-edge amplitude
are plotted against the Strouhal number in Fig. 17. In addition to the
specic case just discussed (Re =9; 000, b=c =0:56 10
3
),
curves are also drawn for b=c =0:85 10
3
, b=c =1:13 10
3
,
and b=c =1:41 10
3
. It is seen from Fig. 17a that, as a
consequence of the inertia of the uid, the pitch amplitude increases
with increasing oscillation frequency and increasing exibility. Pitch
phase angle is plotted in Fig. 17b. It is seen that the pitch phase angle
is always positive, indicating that the pitching motion of the airfoil
always leads the heaving motion. The pitch phase angle is seen to
increase with increasing plate thickness. The amplitude of the trailing
edge is plotted against oscillation frequency in Fig. 17c. Trailing-
edge amplitude is not an independent quantity, but is a function of
pitch amplitude and pitch phase angle. With reference to Fig. 1, a
pitch phase angle of zero degrees gives maximum positive
(s
TE
<s
LE
) pitch angle when s
LE
=a. A pitch phase angle of
180 deg gives a maximum negative pitch angle when s
LE
=a. A
180 deg pitch phase angle is therefore favorable if a high trailing-
edge amplitude is sought. At low Strouhal numbers the pitch phase
angle is favorable. However, the pitch amplitude is very small,
leading to a trailing-edge amplitude approximately equal to the
leading-edge amplitude. At high Strouhal numbers the pitch
amplitude is favorable and the pitch phase angle is unfavorable.
Peaks in trailing-edge amplitude are observed at intermediate
Strouhal numbers.
As stated above, the thrust coefcient may be plotted against a
shape characteristic of the airfoil, rather than against plate thickness.
It appears at rst that three possibilities exist: pitch amplitude, pitch
phase angle, and trailing-edge amplitude. When these quantities are
plotted against each other an interesting pattern emerges. Pitch
amplitude is plotted against pitch phase angle in Fig. 18a. Data for
airfoils with b=c =0:56 10
3
, 0:85 10
3
, 1:13 10
3
,
1:41 10
3
, and 4:23 10
3
are plotted for Re =9; 000, 18,000,
and 27,000, and for all Strouhal numbers. The direction of increasing
exibility and oscillation frequency is indicated. It is seen that pitch
amplitude and pitch phase angle are dependent. Pitch amplitude
varies almost linearly with pitch phase, with a gradient of
0:18 0:02. It is noted that the experimental scatter is greatest in
the small pitch-amplitude region. This is because of the difculty in
measuring the phase of small amplitude oscillations accurately. The
trailing-edge amplitude is plotted against pitch phase angle in
Fig. 18b. It is seen that the two quantities are dependent. For high
pitch phase angles the pitch amplitude is very small (Fig. 18a(, and
the trailing-edge amplitude tends to the limit of unity. At low pitch
phase angles the pitch amplitude is high but the phase is such that the
airfoil pitches most steeply at an unfavorable point in the cycle; the
trailing-edge amplitude is low again. At an intermediate pitch phase,
of approximately 100 deg, the trailing-edge amplitude reaches a
peak. In summary, it is seen that the three quantities, pitch amplitude,
pitch phase angle, and trailing-edge amplitude, are dependent.
The question arises as to which parameter is most suitable to
replace the plate thickness. Of the three possibilities, pitch phase
Fig. 18 a) Pitch amplitude as a function of pitch phase angle for the
complete data set. The data points for all Reynolds numbers, airfoil
thicknesses, and oscillation frequencies fall onto a line. b) Trailing-edge
amplitude as a function of pitch phase angle. The data points fall onto a
curve.
Pitch Phase Angle, deg.
P
r
o
p
u
l
s
i
v
e
E
f
f
i
c
i
e
n
c
y
,

0 30 60 90 120 150 180

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Present experiment, Flexible airfoil, h=0.194, Sr=0.31, 1<
0
<18, Re=27000
Tuncer et al., NS, NACA0012, M=0.3, h=1, Sr=0.318,
0
=10, Re=100,000
Isogai et al., NS, NACA0012, M=0.3, h=1, Sr=0.318,
0
=10, Re=100,000
Pedro et al., Incomp. NS, NACA0012, h=0.5, Sr=0.25,
0
=30, Re=1100
Ramamurti & Sandberg, Incomp. NS, NACA0012, h=1, Sr=1.8,
0
=30, Re=1100
Jones, Panel method, NACA0012, h=0.2, Sr=0.032,
0
=4
Garrick, Flat plate, h=0.2, Sr=0.032,
0
=4
Fig. 19 Propulsive efciency as a function of pitch phase angle.
Comparison of present experimental results with linear theory, panel
method, and both compressible and incompressible NavierStokes
codes. The pitch amplitude
0
is variable for the present experiments,
andxedfor all other data series. Optimumphase angles inthe vicinity of
90 deg are found over a wide range of parameters.
HEATHCOTE AND GURSUL 11
angle, has been found throughout the literature to be of consistent
signicance. Furthermore, the value of the pitch phase angle that
maximizes efciency has been found across a broad range of studies
to lie in the region of 90 deg. For example, the rigid-airfoil Navier
Stokes analyses of Tuncer and Kaya [17], Isogai et al. [15], Tuncer
et al. [52], Ramamurti and Sandberg [53], and Pedro et al. [29]
predict optimum pitch phase angles of 86, 90, 90, 90100, and 90
110 deg, respectively; in nature it is observed that pitch oscillations
of the n lead the heave motion by an angle close to 90 deg [54];
experimentally, Anderson [22] found high efciencies for pitch
phase angles of 75 and 90 deg, whereas Read et al. [55] found high
efciencies in the region of 90100 deg. The common nding in the
NavierStokes studies is that pitch phase angles in this range tend to
lower the effective angle of attack amplitude, thus lessening the
degree of leading-edge owseparation. Propulsive efciency curves
froma selection of these analyses, as well as the prediction of Garrick
theory and a panel method from Jones [50], are plotted in Fig. 19.
Note that, although the Reynolds number, Strouhal number, and
amplitudes differ greatly, the optimum pitch phase angle is around
90 deg for each data set. Also shown in the gure is a curve from the
present set of data, where each data point corresponds to a different
plate thickness. The similarity of the shape of the efciency curve for
the present data set to those of the studies of rigid airfoils in coupled
heave and pitch, and the closeness of the optimum pitch phase angle
to those in the other studies, suggests that the pitch phase angle is a
suitable alternative parameter to the plate thickness. The
performance of the airfoil may now be analyzed in terms of the
Strouhal number and pitch phase angle.
E. Effect of Pitch Phase Angle and Strouhal Number
When thrust coefcient and propulsive efciency are plotted
against pitch phase angle an interesting pattern emerges: curves of
constant Strouhal number tend to fall onto each other. Example plots
are shown in Fig. 20. In part a), thrust coefcient curves for Sr =
0:33 and Sr =0:49 are drawn; efciency curves are shown in part b).
It is seen that the effect of Reynolds number is small, as seen earlier
for the NACA0012 airfoil.
The complete data set is represented as contour plots in Fig. 21 for
thrust coefcient [part a)] and efciency [(part b)]. Contours of thrust
coefcient are drawn on a pitch phase angleStrouhal number plane
in part a). The white curve indicates the optimum pitch phase angle
(in terms of thrust) for a given Strouhal number. For a Strouhal
number of Sr =0:29, for example, the optimumpitch phase angle is
110 4 deg. An airfoil of this exibility would experience a higher
thrust coefcient than a less exible airfoil (the region to the right of
the white curve), or a more exible airfoil (the area to the left). The
optimum pitch phase angle is observed to decrease as the Strouhal
number increases. From a design aspect, the plot indicates the
optimum pitch phase angle for a given ight Strouhal number. The
optimum pitch phase angle may be compared with values from the
literature. Tuncer and Kaya [17], Pedro et al. [29], Ramamurti and
Sandberg [53], and Isogai et al. [15] observed optimum (in terms of
thrust) pitch phase angles of 100, 115, 120, and 120 deg,
respectively. It is seen from Fig. 21a that these values are in
agreement with those found in the present experiment. The
corresponding plot for propulsive efciency is shown in Fig. 21b.
The contour map exhibits a peak in efciency at a Strouhal number of
Sr =0:29. It is noted that this lies within the range of Strouhal
Fig. 20 The variation with pitch phase angle of a) thrust coefcient and
b) propulsive efciency. Curves are shown for Sr 0:33 (open symbols)
and Sr 0:49 (solid symbols).
Fig. 21 Contours of a) thrust coefcient and b) propulsive efciency, in
the Strouhal numberpitch phase angle plane. Black dots indicate
experimental data points. The complete set of data is plotted (all
stiffnesses, frequencies, and Reynolds numbers). The solid white line
indicates the peak thrust coefcient for a given Strouhal number.
12 HEATHCOTE AND GURSUL
numbers of 0:2 <Sr < 0:4 found in nature [44]. It may also be
compared with the optimum Strouhal number Sr =0:30 found
experimentally by Anderson [22]. The reason for the decrease in
efciency for lower or higher Strouhal numbers may be attributed to
the increase in ow separation at higher Strouhal numbers (higher
effective angles of attack), and a transition to drag at lower Strouhal
numbers. The optimumpitch phase angle is seen fromFig. 21b to be
100 4 deg, consistent with the values found in studies of rigid
airfoils in coupled heave and pitch, and those found in nature. It is
noted that the angles found to optimize the thrust coefcient, both in
the literature and in the present study, are higher than those found to
maximize the efciency. This is consistent with the observation [17]
that it is not generally possible to achieve maximum thrust and
efciency simultaneously.
V. Conclusions
Water tunnel experiments on chordwise-exible airfoils heaving
with constant amplitude have been carried out for Reynolds numbers
of 9,000 to 27,000. The results revealed peaks in thrust coefcient for
intermediate stiffness at constant Strouhal number. These indicate
that a degree of exibility is benecial from a thrust aspect. In the
cases of higher thrust, the vortices are stronger and farther apart in the
cross-stream direction. Also, the time-averaged owelds showed a
stronger jet. The effect of Strouhal number is similar to that of
exibility. Peaks in thrust coefcient are observed, which move to
higher Strouhal numbers with increasing stiffness. Stronger vortices
and time-averaged jets are observed for these optimum cases.
Efciency benets were also observed over a broad range of
chordwise exibility, with the optimum efciency exceeding the
efciency of the inexible airfoil by approximately 15%. Flow
visualization and PIV data experiments revealed weaker leading-
edge vortices in the cases of optimum efciency.
An analogy was made between a exible airfoil oscillating in
heave and a rigid airfoil oscillating in pitch and heave. The three
shape characteristics, pitch phase angle, pitch amplitude, and
trailing-edge amplitude, were found to be interdependent. Data from
measurements for all Reynolds numbers, plate thicknesses, and
heave frequencies, were found to lie on a line (pitch amplitude vs
pitch phase angle) or a curve (trailing-edge amplitude vs pitch phase
angle). The phase angle of the pitch was found to lead the heaving
motion.
Thrust coefcient and propulsive efciency were found to be
functions of Strouhal number and pitch phase angle. Contour plots of
efciency in the Strouhal numberpitch phase angle plane showed a
distinct peak for a pitch phase angle of 95100 deg (consistent with
experimental and NavierStokes analyses of rigid airfoils in coupled
heave and pitch), and a Strouhal number of 0.29 (within the range
0:2 < Sr <0:4 observed in nature). Contours of thrust coefcient
showed thrust to peak for pitch phase angles in the region of 110
120 deg, but at higher Strouhal numbers. The results suggest the
effect of chordwise exibility is benecial for purely heaving airfoils
at low Reynolds numbers.
Acknowledgment
This work has been supported by an EPSRC (Engineering and
Physical Sciences Research Council) studentship.
References
[1] Pornsin-sirirak, T. N., Tai, Y. C., Nassef, H., and Ho, C. M., Titanium-
Alloy MEMS Wing Technology for a Micro Aerial Vehicle
Application, Sensors and Actuators A (Physical), Vol. 89, No. 12,
2001, pp. 95103.
[2] Peterson, B., Erath, B., Henry, K., Lyon, M., Walker, B., Powell, N.,
Fowkes, K., and Bowman, W. J., Development of a Micro Air Vehicle
for Maximum Endurance and Minimum Size, AIAA Paper 2003-416,
2003.
[3] Ifju, P. G., Jenkins, D. A., Ettinger, S., Lian, Y., Shyy, W., and Waszak,
M. R., Flexible-Wing-Based Micro Air Vehicles, AIAA Paper 2002-
0705, 2002.
[4] Spedding, G. R., and Lissaman P. B. S., Technical Aspects of
Microscale Flight Systems, Journal of Avian Biology, Vol. 29, No. 4,
1998, pp. 458468.
[5] Q10 Mueller, T. J., and DeLaurier, J. D., Aerodynamics of Small Vehicles,
Annual Review of Fluid Mechanics, Vol. 35, 2003, pp. 89111.
[6] Knoller, R., Die Gesetze des Luftwiderstandes (The Laws of Wind
Resistance), Flug-und Motortechnik (Wien), Vol. 3, No. 21, 1909,
pp. 17.
[7] Betz, A., Ein Beitrag zur Erklaerung des Segeluges (A Paper to
Explain Glider Flight), Zeitschrift fuer Flugtechnik und Motorluft-
schiffahrt, Vol. 3, No. 21, 1912, pp. 269272.
[8] Katzmayr, R., Effect of Periodic Changes of Angle of Attack on
Behaviour of Airfoils, NACA 147, 1922.
[9] Garrick, I. E., Propulsion of a apping and oscillating airfoil, NACA
567, 1936.
[10] Wu, T. Y., Hydromechanics of Swimming Propulsion, Part 2: Some
Optimum Shape Problems, Journal of Fluid Mechanics, Vol. 46,
No. 3, 1971, pp. 521544.
[11] Lighthill, M. J., Aquatic Animal Propulsion of High Hydromechanical
Efciency, Journal of Fluid Mechanics, Vol. 44, No. 2, 1970, pp. 265
301.
[12] Chopra, M. G., Large Amplitude Lunate-Tail Theory of Fish
Locomotion, Journal of Fluid Mechanics, Vol. 74, No. 1, 1976,
pp. 161182.
[13] Jones, K. D., Dohring, C. M., and Platzer, M. F., Experimental and
Computational Investigation of the KnollerBetz Effect, AIAA
Journal, Vol. 36, No. 7, 1998, pp. 12401246.
[14] Bhaskaran, R., and Rothmayer, A. P., Separation and Instabilities in
the Viscious Flowover Airfoil Leading Edges, Computers and Fluids,
Vol. 27, No. 8, 1998, pp. 903921.
[15] Isogai, K., Shinmoto, Y., and Watanabe, H., Effects of Dynamic Stall
on Propulsive Efciency and Thrust of Flapping Airfoil, AIAA
Journal, Vol. 37, No. 10, 1999, pp. 11451151.
[16] Tuncer, I. H., and Platzer, M. F., Computational Study of Flapping
Airfoil Aerodynamics, Journal of Aircraft, Vol. 37, No. 3, 2000,
pp. 514520.
[17] Tuncer, I. H., and Kaya, M., Optimization of Flapping Airfoils for
Maximum Thrust and Propulsive Efciency, AIAA Journal, Vol. 43,
No. 11, 2005, pp. 23292336.
[18] Lewin, G. C., and Haj-Hariri, H., Modelling Thrust Generation of a
Two-Dimensional Heaving Airfoil in a Viscous Flow, Journal of Fluid
Mechanics, Vol. 492, 2003, pp. 339362.
[19] Freymuth, P., Thrust Generation by an Airfoil in Hover Modes,
Experiments in Fluids, Vol. 9, 1990, pp. 1724.
[20] Sunada, S., Kawachi, K., Matsumoto, A., and Sakaguchi, A.,
Unsteady Forces on a Two-Dimensional Wing in Plunging and
Pitching Motions, AIAAJournal, Vol. 39, No. 7, 2001, pp. 12301239.
[21] Lai, J. C. S., and Platzer, M. F., Characteristics of a Plunging Airfoil at
Zero Freestream Velocity, AIAA Journal, Vol. 39, No. 3, 2001,
pp. 531534.
[22] Anderson, J. M., Streitlien, K., Barrett, D. S., and Triantafyllou, M. S.,
Oscillating Foils of High Propulsive Efciency, Journal of Fluid
Mechanics, Vol. 360, 1998, pp. 4172.
[23] Bohl, D. G., and Koochesfahani, M. M., MTV Measurements of the
Flow Structure Downstream of an Oscillating Airfoil, AIAA
Paper 2003-4017, 2003.
[24] Lai, J. C. S., and Platzer, M. F., Jet Characteristics of a Plunging
Airfoil, AIAA Journal, Vol. 37, No. 12, 1999, pp. 15291537.
[25] Heathcote, S., and Gursul, I., Jet Switching Phenomenon for a
Plunging Airfoil, AIAA Paper 2004-2150, 2004.
[26] Ihara, A., and Watanabe, H., On the Flow Around Flexible Plates:
Oscillating with Large Amplitude, Journal of Fluids and Structures,
Vol. 8, No. 6, 1994, pp. 601619.
[27] Koochesfahani, M. M., Vortical Patterns in the Wake of an Oscillating
Airfoil, AIAA Journal, Vol. 27, No. 9, 1989, pp. 12001205.
[28] Hover, F. S., Haugsdal, O., and Triantafyllou, M. S., Effect of Angle of
Attack Proles in Flapping Foil Propulsion, Journal of Fluids and
Structures, Vol. 19, 2004, pp. 3747.
[29] Pedro, G., Suleman, A., and Djilali, N., A Numerical Study of the
Propulsive Efciency of a Flapping Hydrofoil, International Journal
for Numerical Methods in Fluids, Vol. 42, 2003, pp. 493526.
[30] Sfakiotakis, M., Lane, D. M., and Davies, J. B. C., Review of Fish
Swimming Modes for Aquatic Locomotion, IEEE Journal of Oceanic
Engineering, Vol. 24, No. 2, 1999, pp. 237251.
[31] Wootton, R. J., Support and Deformability in Insect Wings, Journal
of Zoology: Proceedings of the Zoological Society of London, Vol. 193,
1981, pp. 447468.
[32] Steppan, S. J., Flexural Stiffness Patterns of Buttery Wings
(Papilionoidea), Journal of Research on the Lepidoptera, Vol. 35,
2000, pp. 6177.
HEATHCOTE AND GURSUL 13
[33] Maxworthy, T., The Fluid Dynamics of Insect Flight, Annual Review
of Fluid Mechanics, Vol. 13, 1981, pp. 329350.
[34] Triantafyllou, M. S., Triantafyllou, G. S., and Yue, D. K. P.,
Hydrodynamics of Fishlike Swimming, Annual Review of Fluid
Mechanics, Vol. 32, 2000, pp. 3353.
[35] Katz, J., and Weihs, D., Hydrodynamic Propulsion by Large
Amplitude Oscillation of an Airfoil with Chordwise Flexibility,
Journal of Fluid Mechanics, Vol. 88, No. 3, 1978, pp. 485497.
[36] Murray, M. M., and Howle, L. E., Spring Stiffness Inuence on an
Oscillating Propulsor, Journal of Fluids and Structures, Vol. 17,
No. 7, 2003, pp. 915926.
[37] Miao, J.-M., and Ho, M.-H., Effect of Flexure on Aerodynamic
Propulsive Efciency of Flapping Flexible Airfoil, Journal of Fluids
and Structures (to be published).
[38] Bozkurttas, M., Dong, H., Mittal, R., Madden, P., and Lauder, G. V.,
Hydrodynamic Performance of Deformable Fish Fins and Flapping
Foils, AIAA Paper 2006-1392, 2006.
[39] Heathcote, S., Martin, D., and Gursul, I., Flexible Flapping Airfoil
Propulsion at Zero Freestream Velocity, AIAA Journal, Vol. 42,
No. 22, 2004, pp. 21962205.
[40] Jones, K. D., and Platzer, M. F., Experimental Investigation of the
Aerodynamic Characteristics of Flapping-Wing Micro Air Vehicles,
AIAA Paper 2003-0418, 2003.
[41] Q11 Prempraneerach, P., Hover, F. S., and Triantafyllou, M. S., The Effect
of Chordwise Flexibility on the Thrust and Efciency of a Flapping
Foil, Proceedings of the Thirteenth International Symposium on
Unmanned Untethered Submersible Technology, 2003.
[42] Jones, K. D., Duggan, S. J., and Platzer, M. F., Flapping-Wing
Propulsion for a Micro Air Vehicle, AIAA Paper 2001-0126, 2001.
[43] Liu, H., Wassersug, R. J., and Kawachi, K., A Computational Fluid
Dynamics Study of Tadpole Swimming, Journal of Experimental
Biology, Vol. 199, 1996, pp. 12451260.
[44] Taylor, G. K., Nudds, R. L., and Thomas, A. L. R., Flying and
Swimming Animals Cruise at a Strouhal Number Tuned for High Power
Efciency, Nature (London), Vol. 425, 2003, pp. 707711.
[45] Murray, M. M., Hydroelasticity Modeling of Flexible Propulsors, Ph.
D. Dissertation, Mechanical Engineering and Materials Science
Department, Duke Univ., Durham, NC, 2000.
[46] Frampton, K. D., Goldfarb, M., Monopoli, D., and Cveticanin, D.,
Passive Aeroelastic Tailoring for Optimal Flapping Wings, Fixed and
Flapping Wing Aerodynamics for Micro Air Vehicle Applications,
edited by Mueller, T. J., Progress in Astronautics and Aeronautics,
Vol. 185, AIAA, Reston, VA, 2001, pp. 473482.
[47] Young, J., Numerical Simulation of the Unsteady Aerodynamics of
Flapping Airfoils, Ph.D. Dissertation, School of Aerospace, Civil and
Mechanical Engineering, New South Wales Univ., Canberra, 2005.
[48] Young, J., and Lai, J. C. S., Oscillation Frequency and Amplitude
Effects on the Wake of a Plunging Airfoil, AIAA Journal, Vol. 42,
No. 10, 2004, pp. 20422052.
[49] Q12 Sheldahl, R. E., and Klimas, P. C., Aerodynamic Characteristics of
Seven Symmetrical Airfoil Sections Through 180-Degree Angle of
Attack for Use in Aerodynamic Analysis of Vertical Axis Wind
Turbines, Sandia National Laboratories, 1981.
[50] Jones, K. D., Lund, T. C., and Platzer, M. F., Experimental and
Computational Investigation of Flapping Wing Propulsion for Micro
Air Vehicles, Fixed and Flapping Wing Aerodynamics for Micro Air
Vehicle Applications, edited by Mueller, T. J., Progress in Astronautics
and Aeronautics, Vol. 195, AIAA, Reston, VA, 2001, pp. 307339.
[51] Platzer, M. F., and Jones, K. D., Flapping Wing Aerodynamics:
Progress and Challenges, AIAA Paper 2006-500, 2006.
[52] Tuncer, I. H., Walz, R., and Platzer, M. F., A Computational Study of
the Dynamic Stall of a Flapping Airfoil, AIAA Paper 1998-2519,
1998.
[53] Ramamurti, R., and Sandberg, W. C., Simulation of Flow About
Flapping Airfoils Using Finite Element Incompressible Flow Solver,
AIAA Journal, Vol. 39, No. 2, 2001, pp. 253260.
[54] Rozhdestvensky, K. V., and Ryzhov, V. A., Aerohydrodynamics of
Flapping-Wing Propulsors, Progress in Aerospace Sciences, Vol. 39,
No. 8, 2003, pp. 585633.
[55] Read, D. A., Hover, F. S., and Triantafyllou, M. S., Forces on
Oscillating Foils for Propulsion and Maneuvering, Journal of Fluids
and Structures, Vol. 17, 2003, pp. 163183.
F. Coton
Associate Editor
14 HEATHCOTE AND GURSUL
Queries
Q1. Please provide a position title such as Professor, Research Assistant, Lecturer, etc., for each of the authors to be used in author
footnotes. Also if AIAA membership is applicable, please indicate as, for example, Member AIAA.
Q2. Please check that the conference information at the beginning of the rst footnote is correct (AIAA Aerospace Sciences Meeting
and Exhibit, Reno, NV, 912 January 2006).
Q3. Please note that bibliographic information such as titles must be conned to the reference section. Both the 1909 and 1912 titles
were removed here and added to Refs. [6] and [7]. Please check that these references appear correctly.
Q4. Please note that it is an AIAA style guideline that when denitions are given in the Nomenclature they should not be repeated in
text. We have deleted some of these repeated denitions in this paragraph and throughout the paper. Please check to conrm
meaning was not altered in text where these deletions occurred. If there are other redundant denitions that we have overlooked in
text, please remove them.
Q5. Please supply a denition for TSI and IMO.
Q6. Please note that private communications must be cited in footnote form rather than as a reference. We have removed reference
[49] from the reference list and placed its content here (reference order has changed from this point on). If there is a date (year)
available for this private communication, please add it at the end of the footnote.
Q7. Please supply a denition for PIV.
Q8. Is bracketing correct in this expression?
Q9. Please check that all of the gure citations are identied correctly in this paragraph.
Q10. Please provide an issue number (or month) for references 5,18,19,28,29,31-34,43,44, 56
Q11. Please provide the name of the publisher, city of publication, and date of publication (not the city where the conference was held
or the date the conference took place) for reference 41.
Q12. Can you provide a report number for reference [49]?
Q13. AIAA prefers gure captions to be 12 words or less in length. If possible, please shorten the following captions accordingly
(present length is in parentheses): Figs. 9 (38), 18 (47), 19 (61). Note that equations count as one word. One option would be to
incorporate some of the caption wording into text, or as sublabels in the gure itself (in which case a revised gure would need to
be submitted).
Q14. Please note that all tables must have a caption. We have tentatively added one here, but feel free to reword it as preferred.
HEATHCOTE AND GURSUL 15