IOP PUBLISHING BIOINSPIRATION & BIOMIMETICS

Bioinsp. Biomim. 6 (2011) 045002 (11pp) doi:10.1088/1748-3182/6/4/045002

Aerodynamics of a bio-inspired flexible

flapping-wing micro air vehicle
T Nakata1 , H Liu1,2,4 , Y Tanaka3 , N Nishihashi1 , X Wang1 and A Sato1
1
Graduate School of Engineering, Chiba University, Chiba, 263-8522, Japan
2
Shanghai-Jiao Tong University and Chiba University International Cooperative Research Center
(SJTU-CU ICRC), Shanghai, People’s Republic of China
3
Fujikura Ltd, Tokyo, 135-8512, Japan
E-mail: hliu@faculty.chiba-u.jp

Received 31 May 2011

Accepted for publication 9 September 2011
Published 29 November 2011
Online at stacks.iop.org/BB/6/045002

Abstract
MAVs (micro air vehicles) with a maximal dimension of 15 cm and nominal flight speeds of
around 10 m s−1 , operate in a Reynolds number regime of 105 or lower, in which most natural
flyers including insects, bats and birds fly. Furthermore, due to their light weight and low flight
speed, the MAVs’ flight characteristics are substantially affected by environmental factors
such as wind gust. Like natural flyers, the wing structures of MAVs are often flexible and tend
to deform during flight. Consequently, the aero/fluid and structural dynamics of these flyers
are closely linked to each other, making the entire flight vehicle difficult to analyze. We have
recently developed a hummingbird-inspired, flapping flexible wing MAV with a weight of
2.4–3.0 g and a wingspan of 10–12 cm. In this study, we carry out an integrated study of the
flexible wing aerodynamics of this flapping MAV by combining an in-house computational
fluid dynamic (CFD) method and wind tunnel experiments. A CFD model that has a realistic
wing planform and can mimic realistic flexible wing kinematics is established, which provides
a quantitative prediction of unsteady aerodynamics of the four-winged MAV in terms of vortex
and wake structures and their relationship with aerodynamic force generation. Wind tunnel
experiments further confirm the effectiveness of the clap and fling mechanism employed in
this bio-inspired MAV as well as the importance of the wing flexibility in designing small
flapping-wing MAVs.
(Some figures in this article are in colour only in the electronic version)

1. Introduction et al 2007). Furthermore, due to their light weight and low

flight speed, the MAVs’ flight characteristics are substantially
Micro air vehicles (MAVs) are now an active and well- affected by environmental factors such as wind gust, which
integrated research area, attracting participation from a wide may lower the flight stability and hence make the flight control
range of talents. With a maximal dimension of 15 cm and a very challenging problem. Like natural flyers, the wing
nominal flight speeds of around 10 m s−1 , MAVs are desired structures of MAVs are often flexible and tend to deform during
to be capable of performing missions such as environmental flight. Consequently, the aero/fluid and structural dynamics
monitoring, surveillance and assessment in hostile situations. of these flyers are closely linked to each other, making the
MAVs normally operate in a Reynolds number regime of entire flight vehicle difficult to analyze (Shyy et al 2010).
104 –105 or lower, in which most natural flyers including
In the past decade, there has been a remarkable increase
insects, bats and birds fly, and the prominent feature of
in research and development of the MAVs and numerous
MAVs’ aerodynamics, in general, is characterized by a large-
vehicle concepts, including fixed wing, rotary wing and
scale vortex flow structure and hence highly unsteady (Shyy
flapping wing, have been proposed (Mueller 2001, Pines
4 Author to whom any correspondence should be addressed. and Bohorquez 2006, Platzer et al 2008, Stanford et al

1748-3182/11/045002+11$33.00 1 © 2011 IOP Publishing Ltd Printed in the UK

Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

2008, Liu 2010). As a vehicle becomes of size smaller the crank is mounted to link and actuate the two pairs of
than 15 cm, normally corresponding to a Reynolds number wings on the 60-tooth final gear. The gearbox system, the
less than 105 , the fixed-wing designs encounter fundamental crank and the wings are connected by a carbon rod of diameter
challenges in a low lift-to-drag ratio and unfavorable flight 0.5 mm with the tail, the rudder, the receiver and the remote-
control. There are merits and challenges associated with controller. The rudder is controlled by a magnetic actuator
rotary- and flapping-wing designs with a smaller size and (HingeAct, PLANTRACO) to move laterally, which weighs
at lower Reynolds numbers. All successful flapping-wing 0.23 g and can provide sufficient control power. The remote-
MAVs developed upto now have flexible and light wings controller with infrared ray offers two channels to control both
as observed in biological flyers in nature (Wootton 1981), the motor frequency and the rudder angle. The rechargeable
which indicates that wing flexibility is likely to have a lithium polymer battery (FR30SC, FULLRIVER) is utilized
significant influence on the resulting aerodynamics as well as the power source. With all parts mounted together our
as the flight stability (Young et al 2009, Mountcastle and flapping MAV weighs less than 3 g in toto and is able to fly
Daniel 2009, Du and Sun 2010, Shyy et al 2010, Liu 2010). with time duration up to 6 min, a maximum height over 10 m
Therefore, flapping flexible wing aerodynamics is of great and a region of 20 m × 20 m.
importance not only in uncovering the novel mechanisms
in insect and bird flights but also in designing efficient 2.2. A high-speed camera setup and flexible wing kinematics
flapping flight vehicles. Recently, we have developed an measurements
integrated computational model for insect flapping flight with
To investigate the flexible wing aerodynamics, we designed
flexible wings, which is established on a fluid–structure
a high-speed camera filming system as depicted in figure 2
interaction approach through an integrated modeling of
and measured the flexible wing kinematics by tracking the
realistic body–wing morphology, realistic flapping-wing and
rigid body motion along with deformation of the flapping
body kinematics, and realistic flexible wing structure (Nakata
wings. The camera system is similar to those in the existing
and Liu 2011). Our results demonstrate the importance of the
literature (e.g. Zheng et al 2009). Three high-speed cameras
inherent flexibility of insect wings in enhancing aerodynamic
(Miro, Vision Research) with an image of 800 × 600 pixels
performance during flapping-wing flight. In this study,
at a frequency of 1000 Hz are synchronized and operated for
we further carry out an extensive study of flexible wing
2 s. Given that the flapping frequency of the MAV normally
aerodynamics of a recently developed hummingbird-inspired
varies over a range of 20 to 35 Hz, the recorded image
flapping-wing MAV. Specific focus is laid on the clap and
sequences are able to provide sufficient temporal resolution
fling mechanism which is achieved by a crank system, not
for the flexible wing kinematics. The camera setup is first
only because such a mechanism is observed in insect flight
calibrated in a three-dimensional global system by a right-
and thought to enhance aerodynamic force generation (Weis-
angled portable calibration frame as shown in figure 2. The
Fogh 1973), but also because such physical interaction can
markers for calibration (the white dots) are arranged in an
affect the in-flight deformation of flexible flapping wing and array with a distance of approximately 100 mm, which is
hence aerodynamic performance. verified to be sufficient to reasonably resolve the flexible wing
kinematics during a wing beat. The recorded image sequences
2. Materials and methods are then downloaded to a computer and the three-dimensional
coordinates of these marked points are reconstructed utilizing
2.1. A hummingbird-inspired MAV a commercial software called DippMotion (Ditect) (figure 2).
Note that the resolution of position tracking is estimated to be
A prototype flapping MAV has been developed recently in
about 5 mm.
Chiba University as illustrated in figure 1(a). The wingspan is
designed to be around 12 cm, of a size as observed in hawkmoth
and hummingbird. The wing has a semi-elliptic planform, 2.3. A computational fluid dynamic model of flapping flexible
which is made of polyethylene film with a thickness of wings
0.03 mm and a carbon rod with a diameter of 0.3 mm at leading 2.3.1. A biology-inspired dynamic flight simulator. In order
edge. The wing length is 60 mm and the wing chord length is to evaluate the aerodynamic performance of the hummingbird-
30 mm at wing base. The mean chord length is calculated to be inspired MAV, we use a biology-inspired, dynamic flight
23.6 mm. Two pairs of the wings are attached to a crank system simulator (Liu 2009, Aono et al 2008), which is designed
(that is similar to that of Delfly—http://www.delfly.nl/, de to integrate the modeling of realistic wing–body morphology,
Croon et al (2009)) that enables the clap and fling mechanism realistic flapping-wing and body kinematics, and unsteady
(Weis-Fogh 1973). As illustrated in figure 1(b), the wings aerodynamics in biological flight. A morphological model
touch thrice in a wing beat: once on the top by upper wings is built based on an effective differential geometric method
and twice on the side by upper and lower wings. The gearbox is for reconstructing geometry and a specific grid generator for
fabricated by cutting the acrylonitrile-butadiene-styrene resin the flapping wings, and a multi-blocked, overset-grid method
so as to ensure a nice match among the motor (MK04S-10, is utilized to deal with complex wing–body geometries and
DIDEL), gears and wing hinges. A 12-teeth pinion gear made complicated flapping movements. A kinematic model is
from polyacetal with a modulus of 0.3 is attached on the motor. constructed to be able to mimic the realistic wing kinematics
With a speed-reduction ratio of 60/12 teeth of the idler gear, of flapping flight, and an efficient analytical method combined

2
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

(a)

Rudder (0.8 g)

Receiver (0.16 g)
Motor Battery
(0.5 g) (0.4 g)
Gear box

(b) Upper wings

Lower
wings
Fling

Clap Clap

Fling
Figure 1. (a) A prototype flapping MAV: hummingbird-inspired MAV. (b) Schematic of wing kinematics in an X-type wing MAV (viewed
from the leading edge).

Table 1. A summary of all tests conducted in this study.

CFD study Wind tunnel experiment

Type of flapping device Mechanical model Prototype (4 wings) Prototype (2 wings)

Chord length (mm) 23.6
Wing length (mm) 60
Flapping frequency (Hz) 18.5 10 or 20 22
Flight condition Tethered (Uref = 2.22 m s−1 ) Tethered in uniform flow of 1.6 or 3.0 m s−1
Reynolds number 3500 2500 (Uref = 1.6 m s−1 )
4700 (Uref = 3.0 m s−1 )
Reduced frequency 0.62 0.46 (Uref = 1.6 m s−1 , f = 10 Hz) 1.02 (Uref = 1.6 m s−1 )
0.93 (Uref = 1.6 m s−1 , f = 20 Hz) 0.54 (Uref = 3.0 m s−1 )
0.49 (Uref = 3.0 m s−1 , f = 20 Hz)

with three coordinate systems is employed for the dynamic 2.3.2. A mechanical model of the hummingbird-inspired
regridding. A fortified finite-volume method-based Navier– MAV. Demand for reduction in size and weight often leads
Stokes solver for the dynamically moving multi-blocked, to unfavorable increase in body flexibility and decrease of the
overset-grid system (Liu 2009) is developed and verified to be endurance of devices such as motor or battery. In particular,
self-consistent by a variety of benchmark tests, and evaluation some induced body vibration due to such unfavorable body
of flapping energetics is established on aerodynamic forces flexibility must be avoided when one intends to evaluate
and powers. Validation of this integrated insect dynamic flight the flapping-wing aerodynamic performance. In this study,
simulator has been achieved by comparisons of aerodynamic we built a specific mechanical model of the hummingbird-
force production with measurements in terms of the mean inspired MAV with a nearly rigid structure in order to evaluate
aerodynamic force. More details on description of the methods its flexible wing aerodynamics by an accurate computational
can be found in Liu (2009), Aono et al (2008) and Maeda model (figure 3(a), table 1). Note that the crank system and
et al (2010). the wing structure are constructed exactly the same as that

3
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

(a) MAV with markers/ (b)

Calibrator

Trigger
PC switch
(c)

(d)

Figure 2. (a) High-speed camera filming system for wing kinematics measurement. (b) A right-angled portable calibration frame and (c) an
image of the flapping MAV captured by a high-speed camera. (d) Reconstructed coordinates on the wing surface based on a sequence of
images in a complete beat cycle.

(a) (b)
Symmetrical plane

(c)

(d)
Flapping wing
with deformation

Counter
weight Fulcrum
Electronic Flapping
balance mechanism
u i(t, xi, yi)

la lb (xi, yi)
lc

F x Initial wing y
Fz

Fx

Figure 3. (a) A mechanical flapping-wing MAV model, (b) a computational fluid dynamic (CFD) model of MAV wings and a multi-blocked
grid system (single wing grid: 35 × 55 × 19, background grid: 181 × 181 × 181). (c) Force measurement system for the mechanical
flapping-wing MAV model. (d) Definition of the displacement u at point (xi , yi ) on the wing surface at time t.

4
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

of the MAV while the body including the gear box is made
MAV Lift
from aluminum. The mechanical model is attached onto an
aluminum beam mounted on a fulcrum for the measurement Drag
of the flapping-wing aerodynamic forces. The adjustment of U
the balance using a counterweight allows the movement of the
opposite side of the fulcrum to rest gently on an electronic Body angle Strut
balance, the resolution of which is 10 mg. The forces that act
on the electronic balance can then be measured in response to
the aerodynamic resultant force by the flapping wing because Bottom of wind tunnel
of the utilization of the seesaw mechanism (figure 3(c)).
Load cell
for drag
2.3.3. Morphological and kinematic models. A realistic
morphological model of the MAV’s wing for computational Load cell for lift
fluid dynamic (CFD) analysis is constructed by tracing the
outline of the wing planform. A uniform thickness is taken
but with elliptic smoothing at the leading and trailing edges as
well as at the tip. To deal with the complexity of the wing Figure 4. Lift and drag force measurement system for wind tunnel
deformation and wing kinematics we use a multi-blocked experiments.
overset-grid method, in which the wing grid is clustered to
(a)
the wing surface with the minimum grid spacing adjacent √ to z
the wing surface controlled by a formula δmin = 0.1cm / Re
where cm is the mean chord length of the wing and Re the Flapping wing
cm -based Reynolds number. The wing block has an outside
boundary of 0.5 times the mean chord length cm . The Cartesian
grid with 12 times the chord length on a side, which is clustered Marker for measurement
around the flapping wing, is employed as the background grid.
φ
Figure 3(b) illustrates computational geometric models and y
grid systems of the MAV.
The kinematic model of the MAV’s wing is constructed x
by interpolating the reconstructed coordinates of the markers
on the flapping wings. The displacements u(t, x, y) at some (b) -0.5
point of the wing (x, y) are interpolated using a function of
Fourier series, such as Lower wing Upper wing
0

nx  
ny
n LE
u(t, x, y) = (α(l, m, n)x l y m cos(nωt) LE
x/cm

l=0 m=0 n=0 0.5

l m
+ β (l, m, n)x y sin(nωt)), (1)
TE
where terms α and β are derived by the least-squares method. 1
The wing surface grids are translated by u and the grid is TE
regenerated on the basis of the hyperbolic grid generation
1.5
scheme (Steger and Rizk 1985). Note that the CFD analysis -1 0 1 2
is performed by introducing a symmetric plane as depicted in φ /rad
figure 3(b) under the assumption that the left and right wings
Figure 5. (a) Definition of the angle position φ. (b) Time courses of
move and deform symmetrically.
both upper and lower wings at a wing cross-section of 0.7R from
The CFD study is performed under the assumption of wing base.
hovering flight condition. Given the mean chord length cm as
the reference length Lref , the mean wing tip velocity in hovering here is proportional to the wing beat amplitude, , the flapping
flight as the reference velocity Uref , which is proportional to frequency, f , the square of the wing length, R2 , but inversely
Uref = ωR, where R is the wing length and ω is the mean proportional to the aspect ratio of the wing, AR.
angular velocity of the wing (ω = 2f , where  is the wing The reduced frequency that normally characterizes
beat amplitude and f is the flapping frequency), the Reynolds rotational versus translational speeds, is defined in the case
number in hovering flight can be reformed as of hovering flights as
  πf Lref π cm π
Uref Lref 2f Rcm f R 2 4 k= = = . (3)
Re = = = , (2) Uref 2R AR
ν ν ν AR
Note that the reduced frequency k is inversely proportional to
where the aspect ratio AR is in the form of AR = (2R)2 /S, the beat amplitude  and the aspect ratio AR of the wing.
with a wing area of S = 2Rcm . Note that the Reynolds number According to the measured data of the MAV’s mechanical

5
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

TEV Symmetrical plane

TV

LEV

TV

y x
TEV

TV LEV

TEV

y x LEV

00.0
0 00.5
5 11.0
0 11.5
5 22.0
0 -3 0
-3.0 15
1.5 00
0.0 15
1.5 30
3.0
Absolute flow velocity (non-dimensional) Pressure (non-dimensional)

Figure 6. Instantaneous streamlines, iso-vorticity surface and pressure contours on the upper surface of flapping wings at each half stroke.

2.4. Wind tunnel experiment setup

Symmetrical plane
Aerodynamic performance of the MAV in forward flight
condition is evaluated by a force measurement setup based
on wind tunnel experiments (table 1). The MAV is equipped
with a load cell for the measurements of mean lift and drag
forces via strut and sting from backward as illustrated in
figure 4, and is located at the centre of a low-speed wind tunnel
with a test section of 0.36 m × 0.36 m (Akita National College
of Technology). The load cell is made from an aluminum block
with strain gauge. The body angle of the MAV, θ b , varies from
0◦ to 70◦ through adjustment of the sting. The free-stream
velocity is set to be 1.6 and 3.0 m s−1 . The flapping frequency
is controlled by a dc power source. A two-winged MAV
that has a similar flapping-wing system as the lower wing
Pressure
(non-dimensional) -1.0 -0.5 0.0 0.5 1.0
of the four-winged MAV is also built for comparison. Note
that the two-winged and four-winged MAVs have the same
Figure 7. Pressure contours on a virtual cylindrical surface at 2.0 cm wing structure. A different two-winged model made from thin
from wing base.
Mylar film is also tested. The Re and k of the four-winged MAV
model (cm = 23.6 mm, R = 60 mm,  = 1 rad, f = 18.5 s−1 , for forward flight condition, in which the reference velocity is
ν = 1.5 × 10−5 m2 s−1 ), Re and k are calculated to be about defined by the free-stream velocity, are calculated to be 2500
3400 and 0.59, respectively. and 0.93, respectively, when Uref = 1.6 m s−1 .

6
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

Upper-right Lower-right
wing wing

z y

Downwash by previous
x half stroke

Upper-right Lower-right
wing wing

z y
Downwash by previous
half stroke
x

Downward flow velocity

(non-dimensional) -1.0 -0.5 0.0 0.5 1.0

Figure 8. Wake topologies at the end of each half stroke. Velocity vectors and contours are visualized at a virtual cylindrical surface at
2.0 cm from wing base.

3. Results beat cycle and reconstructed the flexible wing kinematics by

means of the interpolation methods given in section 2.3. The
3.1. Flexible wing kinematics flexible wing kinematics with wing deformations can then be
defined in a three-dimensional manner through the Fourier
Based on the methods described in section 2.2 we measured the series-based function (1). Figure 5(b) shows the time courses
flapping motions and deformations of the wings in a complete of both upper and lower wings at a wing cross-section of 0.7R

7
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

from wing base (figure 5(a)), in which the cross-sections at the

stroke reversal are shown by the dotted lines. Note that those
markers where the coordinates cannot be reconstructed due to
the overlapping of the upper and lower wings are excluded
(figure 5(b)). It is seen that since the crank system can only
generate the flapping motion, i.e. one-axial motion, the wing
rotation or the feathering motion is induced passively by the 10 : Total
wing flexibility. However, the angles of attack of both upper (a) : Upper
and lower wings are generated somehow in a good timing : Lower
0
contributing to the lift force production during each half stroke.
It is found that the attitude of the upper wing is nearly vertical

mN
to the horizontal axis at both stroke reversals, and hence, the

Fx /m
-10
10
rotational phase of the upper wing is ‘symmetric’ (Dickinson
et al 1999), which is very likely a result of the clap and fling
at each stroke reversal. On the other hand, the lower wing -20
keeps the attitude for the previous half stroke at stroke reversal
without the clap and fling, rotating to prepare the forthcoming
half stroke, and hence displays some ‘phase delay’ at the -30
stroke reversal, which may lead to lowering of the aerodynamic 20
performance of flapping wings (Dickinson et al 1999). (b)

3.2. CFD modeling of flexible wing flapping 10

CFD modeling of unsteady flows around the MAV’s flexible

N
Fz /mN

wings undergoing flapping is performed for a single flexible 0

wing model in which realistic geometric and kinematic models
are utilized as described in the preceding sections. The CFD-
based results show that a leading edge vortex (LEV) and hence -10
a strong negative pressure region are generated on the upper
and lower wings during both of the half strokes (figure 6). As 0.15 0.6
observed in insect flapping flight (Ellington et al 1996), this -20
LEV very likely plays a crucial role in the lift and/or thrust 0 0.25 0.5 0.75 1
force production in the MAV flight. The vortex rings that are Stroke cycle, t/ T
formed from the LEV, the tip vortex (TV) and the trailing edge
Figure 9. Time courses of force components (a) Fx and (b) Fz on
vortex (TEV) are also observed, showing a similar pattern right wings of the four-winged MAV.
to those of insect flight (Liu and Aono 2009). Obviously,
the strong negative pressure regions are detected between the
next stroke reversal (vertical dashed line in figure 9) while the
upper right and left wings as illustrated in figure 7, which is
upper wing shows a larger peak again at the half stroke. Since
likely induced by the clap and fling mechanism. the present flapping-wing mechanism generates large forces at
Figure 8 illustrates the wake topology on a virtual each half stroke, one peak in the total Fx is observed at each
cylindrical surface at the end of each half stroke. An intense half stroke.
downwash is generated behind the path of the wings, identical Furthermore, the mean aerodynamic force is calculated to
to the centre of the vortex rings (Liu 2009). The downwash be 23.3 mN, which is in good agreement with the measurement
disks are connected by the clap and fling with a pitching of a value of 26.46 mN acting on the electronic balance. The
up motion, resulting in a pronounced downwash in far wake mean force components of Fx and Fz generated by the upper
below. In contrast, it is interesting to see that the downwash wing are −4.2 and 0.2 mN, and, by the lower wing are −3.8
generated by the lower wings without the clap and fling is and 2.0 mN, respectively. The mean aerodynamic powers
rather weak and there is almost no high-speed zone visible consumed by the upper and lower wings are estimated to be
downward in the far wake. 14.2 and 14 mW, respectively.
Figure 9 shows the time courses of aerodynamic forces
generated by the upper and lower wings in a wing beat.
3.3. Wind tunnel experiments: lift and drag forces, and
Clearly, the upper wing generates very few aerodynamic forces
flexible wing kinematics
until 0.15 T after the clap and fling (vertical dashed line in
figure 9) while the aerodynamic forces generated by the lower The aerodynamic performance of the present prototype MAV
wing show a stable increase after the stroke reversal. However, is further evaluated by means of wind tunnel experiments.
the horizontal force Fx by the upper wing shows a rapid rise As described in section 2.4 the forces acting upon the MAV
subsequently, having a higher peak than the lower wing. Then undergoing tethered fight are measured by utilizing the load
the force Fx keeps decreasing to a low level until 0.6 T after the cells. Figure 10 shows the mean lift and drag forces plotted

8
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

(a) 40 (b) 30
: f = 20 H
Hz
30 : f = 10 Hz 20
: f = 0 Hz
20 10

10 0

0 -10
U = 1.6
1 6 m/s
/
-10 -20
40 30
: U = 3.0
3 0 m/s
30 : U = 1.6 m/s 20
: U = 0.0 m/s

Drag/mN
N
Lift/mN
N

20 10
L

D
10 0

0 -10
f = 20 Hz
H
-10 -20
40 30
: 4 wing
30 : 2 wing 20
: 2 wingg ((soft))
20 10

10 0
U = 1.6 m/s
0 f = 20 Hz ((4 wing)
g) -10
f = 22 Hz (2 wing)
-10 -20
0 20 40 60 80 0 20 40 60 80
Body angle/deg Body angle/deg

Figure 10. (a) Lift and (b) drag forces plotted against body angles for four-winged and two-winged MAVs. Gray lines with filled triangles
or squares represent twice the lift and drag forces generated by the two-winged MAV.

against the body inclination angle of θ b , which varies from a significant variation spanwise from wing base to wing tip,
0◦ to 70◦ in this study. The lift force is seen to increase which points to the existence of some passive wing twists. In
with increasing body angle, the flapping frequency (with two the four-winged MAV, the feathering angle approaches zero
frequencies of 10 and 20 Hz) and the wind velocity (when at stroke reversal (broken line in figure 11), which implies
θ b is greater than 40◦ ), which reaches a value of 25 mN at a that the phase difference between positional and feathering
flapping frequency of 20 Hz and a wind velocity of 1.6 m s−1 , angles is approximately a quarter-period, T/4. While the phase
very close to the MAV’s weight. The drag force also shows an difference of the two-winged MAV with the thin Mylar wing
increase with increasing body angle but a negative drag force, is observed to be around a quarter-period, T/4, the feathering
i.e. the thrust force, is observed when the body angle is less angle of the two-winged MAV with polyethylene obviously
than 50◦ . Note that the thrust force is generated merely in shows an advance in phase.
the case of no wind velocity and turns to the drag force at the
larger wind velocity of 3.0 m s−1 .
4. Discussion
One key reason that the present MAV utilizes the X-type
wing (figure 1) with four wings is because it can generate 4.1. Effect of clap and fling on force production
more lift and/or thrust forces than a two-winged MAV does.
Comparison of the lift forces between the two types of MAVs The clap and fling of flapping wings at dorsal stroke
indicates that the present four-winged MAV generates more reversal is found in various insects such as wasps (Weis-Fogh
than twice the lift force of the two-winged MAV in the case 1973), Diptera (Ellington 1984) and damsefly (Wakeling and
when θ b is greater than 40◦ . However, twice the lift force of Ellington 1997), which is explained as an effective mechanism
the two-winged MAV with the thin Mylar wing shows a larger to strengthen the development of circulation (Weis-Fogh
value than that of the present four-winged MAV. 1973). Ellington (1984) proposed another variation of this
Since the wing flexibility may influence the wing mechanism termed as clap and peel, in which the fling or peel
deformation and hence the aerodynamic performance phase during downstroke is suggested to be responsible for
differently between the two-winged and four-winged MAVs, enhancing the circulation due to the flow of air into the opening
we further investigate the wing deformation in terms of gap and hence an attached LEV (figure 6). The strong negative
positional and feathering angles associated with the lower pressure region (figures 6 and 7) between the upper wings of
wing of the four-winged MAV and a single wing of the two- the present four-winged MAV is likely induced by the fling or
winged MAV (figure 11). A common feature observed here peel of the wings, and, as a result, leads to a rapid increase in
is that the feathering angles in the two cases obviously show aerodynamic force production (figure 9).

9
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

reversal (figure 5(a)). This implies that the clap and fling
IV III II I
(a) of a flexible wing can adjust the feathering angle near the
wing tip at stroke reversal so as to avoid some unfavorable
Wing phase delay during wing rotation. As a result, it is found that
base with the clap and fling at both stroke reversals the upper wing
Wing that is perpendicular to the stroke plane of flapping wings
tip generates 10% more mean aerodynamic force of Fx than
the lower wing, which claps and flings only on one side of
stroke reversal, with a marginal increase in mean aerodynamic
power. Furthermore, the downwash in the far wake is
clearly generated by the half-stroke preceding wing clap
(figures 7 and 8); the fling-induced additional circulation and
80 Positional the passive deformation-based flexible wing kinematics in toto
(b) Feathering at I are very likely responsible for augmenting the aerodynamic
60 Feathering at II force production effectively in the present four-winged MAV.
40 Feathering at III
Feathering at IV
20 4.2. Which performs better: a four-winged MAV or a
two-winged MAV?
0
-20 The flapping-wing mechanism utilizing an X-type wing with
four wings is employed for the present four-winged MAV,
-40
which is inspired by a smart two-winged natural flyer, the
-60 hummingbird, but obviously has crucial differences. Although
there is a great variety of biological flyers, insects, bats and
80
(c) birds in nature, most of them fly with a pair of wings. The
60 present flapping-wing mechanism is quite unique because of
40 the multiple physical interactions in a wing beat cycle. Here
Angles/deg

we find that a four-winged MAV is able to generate twice as

20
high a lift force as that of a two-winged MAV at body angles
0 larger than 40◦ where the MAV actually flies (figure 10). This
-20 implies that the present flapping-wing mechanism very likely
utilizes the clap and fling mechanism effectively and hence
-40
enhances the force production over the two-winged flapping
-60 mechanism. On the other hand, we also find that the two-
winged MAV with a softer Mylar wing generates more lift
80
(d) forces even than the four-winged MAV in the case of body
60 angles larger than 30◦ (figure 10), which further points to the
40 importance of choice of an appropriate wing structure in the
MAV design (Heathcote and Gursul 2007, Heathcote et al
20 2008).
0
-20 5. Conclusion
-40
In this study, flexible wing aerodynamics of a hummingbird-
-60 inspired flapping MAV is evaluated by combining a CFD
0 0.25 0.5 0.75 1 method and wind tunnel experiments. A CFD model of
Stroke cycle, t/T the realistic flapping flexible wing is successfully established
Figure 11. (a) Markers used in measuring flapping angles. for the prototype flapping MAV, which provides a detailed
Positional and feathering angles of (b) a four-winged MAV, (c) a and quantitative prediction of the unsteady aerodynamics
two-winged MAV and (d) a two-winged MAV with Mylar wing. of the four-winged MAV in terms of the vortex and wake
structures and their relationship with aerodynamic force
Wing deformations due to wing flexibility may also extend generation. Combined with wind tunnel experiments, our
influence on the effect of the clap and fling because the wings CFD analysis further offers a comprehensive understanding of
in our MAV are made mainly from highly flexible films. In the aerodynamic effect of the clap and fling on the flapping
this study, we find that with the wing clap the rotational four-winged mechanism, confirming that the clap and fling
phase of both upper and lower wings at stroke reversal is mechanism observed in insect flight is very likely utilized by
nearly symmetric, while without the wing clap the rotation the present four-winged MAV. Furthermore, an extended study
of the lower wing obviously exhibits a phase delay at stroke of the aerodynamic performance of the MAV is performed to

10
Bioinsp. Biomim. 6 (2011) 045002 T Nakata et al

investigate the effects of the body angle, the flapping frequency Liu H (section editor) 2010 Micro air vehicles Encyclopedia of
and the forward flow velocity on lift and drag force generation Aerospace Engineering ed R Blockley and W Shyy
(Chichester: Wiley)
with a specific focus on the discrepancy compared with a two-
Liu H and Aono H 2009 Size effects on insect hovering
winged MAV, which points to the effectiveness of the four- aerodynamics: an integrated computational study Bioinsp.
winged flapping-wing mechanism as well as the importance Biomim. 4 015002
of the wing flexibility in designing small flapping-wing MAVs. Maeda M, Gao N, Nishihashi N and Liu H 2010 A free-flight
simulation of insect flapping flight J. Aero Aqua Biomech.
1 71–9
Acknowledgments Mountcastle A M and Daniel T L 2009 Aerodynamic and functional
consequences of wing compliance Exp. Fluids 46 873–82
The authors thank Dr Y Kawamura (Fukuoka Institute of Mueller T J ed 2001 Fixed and Flapping Wing Aerodynamics for
Technology) for his advice and suggestions on the design Micro Air Vehicle Applications (Progress in Astronautics and
Aeronautics vol 195) (Reston, VA: AIAA)
and development of the flapping MAV, and Mr I Yano and
Nakata T and Liu H 2011 Aerodynamic performance of a hovering
Mr K Terada (Fujikura Ltd) for their help with designing and hawkmoth with flexible wings: a computational approach
building the mechanical model of the MAV. Special thanks Proc. R. Soc. B at press (doi:10.1098/rspb.2011.1023)
go to Dr M Okamoto (currently at Kanazawa Institute of Pines D J and Bohorquez F 2006 Challenges facing future
Technology) for his great help when performing wind tunnel micro-air-vehicle development J. Aircr. 43 290–305
Platzer M, Jones K, Young J and Lai J 2008 Flapping wing
experiments at Akita National College of Technology. This
aerodynamics: progress and challenges AIAA J. 46 2136–49
work is partly supported by the grants-in-aid for Scientific Shyy W, Aono H, Chimakurthi S K, Trizila P, Kang C-K,
Research no 18656056 and no 18100002, JSPS, Japan. HL is Cesnik C E S and Liu H 2010 Recent progress in flapping
also funded by a CKSP scholarship. TN is funded by JSPS wing aerodynamics and aeroelasticity Prog. Aerosp. Sci.
Research Fellowships for Young Scientists. 46 284–327
Shyy W, Lian Y, Tang J, Viieru D and Liu H 2007 Aerodynamics of
Low Reynolds Number Flyers (New York: Cambridge
References University Press)
Stanford B, Ifju P, Albertani R and Shyy W 2008 Fixed membrane
Aono H, Liang F and Liu H 2008 Near- and far-field aerodynamics wings for micro air vehicles: experimental characterization,
in insect hovering flight: an integrated computational study numerical modeling, and tailoring Prog. Aerosp. Sci. 44 258–94
J. Exp. Biol. 211 239–57 Steger J L and Rizk Y M 1985 Generation of three-dimensional
de Croon G C H E, de Clercq K M E, Ruijsink R, Remes B and body-fitted coordinates using hyperbolic partial differential
de Wagter C 2009 Designing, aerodynamics, and vision-based equations NASA TM-86753
control of the DelFly Int. J. Micro Air Veh. 1 71–97 Wakeling J M and Ellington C P 1997 Dragonfly flight: II.
Dickinson M H, Lehmann F-O and Sane S P 1999 Wing rotation Velocities, accelerations and kinematics of flapping flight
and the aerodynamic basis of insect flight Science 284 1954–60 J. Exp. Biol. 200 557–82
Du G and Sun M 2010 Effects of wing deformation on aerodynamic Weis-Fogh T 1973 Quick estimates of flight fitness in hovering
forces in hovering hoverflies J. Exp. Biol. 213 2273–83 animals, including novel mechanisms for lift production J. Exp.
Ellington C P 1984 The aerodynamics of insect flight: IV. Biol. 59 169–230
Aerodynamic mechanisms Phil. Trans. R. Soc. Lond. B Wootton R J 1981 Support and deformability in insect wings
305 79–113 J. Zool. 193 447–68
Ellington C P, van den Berg C, Willmott A P and Thomas A L R Young J, Walker S M, Bomphrey R J, Taylor G K and
1996 Leading-edge vortices in insect flight Nature 384 626–30 Thomas A L R 2009 Details of insect wing design and
Heathcote S and Gursul I 2007 Flexible flapping airfoil propulsion deformation enhance aerodynamic function and flight
at low Reynolds number AIAA J. 45 1066–79 efficiency Science 325 1549–52
Heathcote S, Wang Z and Gursul I 2008 Effects of spanwise Zheng L, Wang X, Khan A, Vallance R R, Mittal R and Hedrick T L
flexibility on flapping wing propulsion J. Fluids Struct. 2009 A combined experimental-numerical study of the role of
24 183–99 wing flexibility in insect flight 47th AIAA Aerospace Sciences
Liu H 2009 Integrated modelling of insect flight: from morphology, Meeting Including New Horizons Forum and Aerospace
kinematics to aerodynamics J. Comput. Phys. 228 439–59 Exposition (Orland, FL, 5–8 January) AIAA-2009-382

11