See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268567100

Unsteady Aerodynamics of Flapping Airfoil in Hovering Flight at Low Reynolds

Numbers

Conference Paper · January 2005

DOI: 10.2514/6.2005-1356

CITATIONS READS
36 1,717

3 authors, including:

Dilek Funda Kurtulus Nafiz Alemdaroglu

Middle East Technical University Atilim University
153 PUBLICATIONS   659 CITATIONS    55 PUBLICATIONS   324 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Verification for Experimental Unsteady Aerodynamic Loads of a Flexible Flapping Wing Actuated with PZT Material View project

Bioinspired flapping wings View project

All content following this page was uploaded by Dilek Funda Kurtulus on 10 September 2016.

The user has requested enhancement of the downloaded file.

 Unsteady Aerodynamics of Flapping Airfoil in Hovering
Flight at Low Reynolds Numbers

D. Funda Kurtulus*
Aerospace Engineering Department Middle East Technical University, 06531 Ankara, Turkey

Laboratoire d’Etude d’Aérodynamique (CNRS UMR6609) Ecole Nationale Supérieure de Mécanique et

d'Aérotechnique, Poitiers, France

Alain Farcy†
Laboratoire d’Etude d’Aérodynamique (CNRS UMR6609) Ecole Nationale Supérieure de Mécanique et
d'Aérotechnique, Poitiers, France

Nafiz Alemdaroglu‡
Aerospace Engineering Department Middle East Technical University, 06531 Ankara, Turkey

The major aim of flapping motion research is based on the understanding of the relation
between the temporal and the spatial changes of the wake structure and the resulting
instantaneous aerodynamic forces over the flapping wings. The essential physics of non-
steady airfoil problems can be observed from simplified two-dimensional experiments, and
the interpretations of the behavior can be supported by theoretical or numerical models. The
aim of this study is to find optimum parameters to generate maximum lift during this
motion, by using numerical methods and analytical models. A great number of cases are
investigated involving the changes in the parameters such as angle of attack, location of start
of change of incidence, location of start of change of velocity, axis of rotation, and Re
number. In addition to the instantaneous aerodynamic forces, pressure distributions and
vorticity contours, the average lift and drag coefficient values are also calculated. Positive lift
values along the motion are obtained for angle of attack greater than 30°. The vortices shed
during the flapping motion generate the lift. A modelization program is developed by use of
Duhamel integral in order to compare with DNS results.

Nomenclature
a = axis of rotation (=0 at leading edge and =1 at trailing edge)
b = semi-chord
CD = drag coefficient
CL = lift coefficient
CN = normal force coefficient
CPPro = mean profile power coefficient
c = chord
D' = drag per unit span averaged during a period
dt = time step
f = feathering parameter
g = determinant of metric tensor
L' = lift per unit span averaged during a period
n = wing beat frequency

*
Ph.D student, Aerospace Eng. Dept., METU Ankara Turkey & LEA ENSMA Poitiers France
†
Prof., LEA ENSMA, BP40109, avenue Clément Ader 86961 Futuroscope Chasseneuil France
‡
Prof., Aerospace Engineering Dept., METU 06531 Ankara Turkey, AIAA Member

1
American Institute of Aeronautics and Astronautics
 ′
Ppro = Profile power per unit span averaged during a period
p = piezometric pressure= ps-ρ0gmxm where ρ0 is reference density, gm gravitational field components and
xm are coordinates from a datum, where ρ0 is defined
ps = static pressure
Re = Reynolds number
s = reduced time
sij = rate of strain tensor
si = momentum source components
sm = mass source
t = time
T = period
ui = absolute fluid velocity component in direction xi
~
uj = uj – ucj, relative velocity between fluid and moving coordinate frame that moves with velocity ucj
V(t) = velocity variation during the flapping motion
V0 = maximum of V(t), initial velocity
xa = x position at the beginning of angle of attack change
xi = Cartesian coordinate (i=1, 2, 3)
xv = x position at the end of constant velocity region
xT/4 = maximum x location, amplitude
α = angle of attack
= stroke plane inclination angle
δij = Kronecker delta (=1 when i=j and zero otherwise)
ρ = density
µ = dynamic viscosity
ν = kinematic viscosity
τij = stress tensor components

I. Introduction

L ast ten years, the numerical and experimental studies of low Re number regime become very important due to
the advances in micro-technologies enabling the development of Micro Air Vehicles (MAV’s). One of the main
concerns of these studies is the flapping motion concept. Although, generally the forward flight regime studies are in
majority in literature; the main objective of MAV applications, i.e. constant position surveillance, reveals the need
for more research on hover mode. Hovering is a flight mode where the body is assumed to be fixed in space and the
freestream velocity is zero. The fluid motions are only due to the wing motions. In hovering, the main effect is to
produce a vertical force in order to balance the weight. Whether a bird can hover or not depends on size, moment of
inertia of the wings, degrees of freedom in the movement of the wings and the wing shape1. Due to these limitations
there are only few species which can hover. Hovering may be symmetrical or asymmetrical. Symmetrical hovering,
also called normal or true hovering, is performed by hummingbirds and insects like Drosophila that hover with fully
extended wings during the entire wing beat cycle which is the case investigated herein. Lift is produced during the
entire wing stroke. The wings are rotated and twisted during the backstroke so that the leading edge of the wing
remains the same throughout the cycle, but the upper surface of the wing during the forward stroke becomes the
lower surface during the backward stroke. For birds which are not able to rotate their wings as in the case of the
hummingbird, the asymmetric hovering is used. In order to avoid large drag forces and negative lift forces, these
birds flex their wings during the upstroke.
Hummingbird and several insects use normal hovering where the wings are moving through a large angle in an
approximately horizontal plane making a figure-of-eight motion with symmetrical half-stroke. The stroke plane in
the case of hovering bats and birds is tilted. This type of motion is named asymmetrical hovering where most of the
lift is generated during downstroke. Table 1 shows the physical and motion characteristics of 4 insects and
hummingbird performing normal hovering, calculated through the use of the actuator disk theory.

2
American Institute of Aeronautics and Astronautics
 Table 1 Dimensions and parameters calculated by actuator disk theory for some birds and insects
performing normal hovering2

Body mass Wing Disk loading Stroke Feathering

Normal Hovering
[kg] semi-span [m] [Nm-2] Period [s] Parameter, f
Fruit Fly,
2×10-6 0.003 0.69 0.004 0.0137
Drosophila virilis
Crane fly,
2.8×10-5 0.0173 0.29 0.018 0.0036
Tipula paludosa
Hover fly,
1.5×10-4 0.0127 2.90 0.0055 0.0056
Eristalis tenax
Bumble bee,
8.8×10-4 0.0173 9.18 0.0064 0.0130
Bombus terrestris
Hummingbird,
5.1×10-3 0.059 4.57 0.0285 0.0111
Amazilia fimbriata

Feathering parameter f in Table 1 is defined as the square ratio between the induced velocity on the wing disk
and the mean tip velocity 2:

2
wg
f = 2
(1)
ut

where the mean tip velocity, defined here as ut=6c/(T/2) is the ratio of the distance covered by the wing tips during
the entire motion of the wings to the time during which the animal is supported by a single vortex ring.
The objectives of the present investigation are to understand the aerodynamics phenomena of flapping motion in
hover and to optimize the maximum lift coefficient value. In persuite of the forgoing, the numerical simulations
have been performed and the results compared with
those obtained through the use of Duhamel Integral with
Wagner and Küssner functions. The previously written
fortran program, has been further developed to handle all
of the cases investigated. The total force is obtained by
superposition of a circulatory part, found from the
indicial method, and a non-circulatory part. The latter
accounts for the pressure forces required to accelerate
the fluid in the vicinity of the airfoil. Results for the
unsteady cases have been formulated in both the time
domain and the frequency domain, primarily by Wagner a) Definition of motion with a 2-D cut on a wing section,
(1925), Theodorsen (1935), Küssner (1935), and von
Kármán & Sears (1938). These solutions are based on
the unsteady thin airfoil theory. The results have shown
positive values of lift along the motion for the angles of
attack greater than 30°. Average lift values are compared
for different parameters.

II. Definition of Motion

The flapping motion is divided into 4 regions with
the first region corresponding to the half of the
downstroke where the leading edge is pointing in
positive direction and second one to the half-upstroke
[Fig. 1]. While the third and fourth regions, are the b) Flapping motion definition of NACA 0012 airfoil
mirror images of these two regions, corresponding to the profile with the axis of rotation at 1/4c location
second half of upstroke and downstroke respectively.
Each region is composed of a translational phase and a Figure 1: Definitions of parameters.
rotational phase. In the translational phase, the airfoil

3
American Institute of Aeronautics and Astronautics
translates with a constant velocity until a predefined time, where a rotational motion around a point on the chordline
is superimposed to the translational motion. Each half cycle starts from rest and comes to a stop. The rotation is such
that the leading edge stays as leading edge during all phases of the motion. A symmetrical NACA 0012 airfoil
section is chosen for this study, so that the symmetry of the motion both during upstroke and downstroke is
maintained.

III. Numerical Method

The unsteady, incompressible, 2D, laminar flow equations are used to compute the flowfield using a Direct
Numerical Simulation (DNS) code4. The analyses are done in terms of flowfield parameters and aerodynamic loads.
For transient calculations efficient, optimized finite-volume solution algorithms are used. The mass and momentum
conservation equations solved for general incompressible and compressible flows and a moving coordinate frame.
The Navier-Stokes equations in Cartesian tensor notation are given by:

1 ∂
g ∂t
( )
gρ +
∂
∂x j
(ρu~ j ) = sm (2)

1 ∂
g ∂t
( g ρ ui +) ∂
∂x j
(ρu~ j ui − τ ij ) = − ∂p + si
∂xi
(3)

where for Newtonian fluid, the constitutive relation is given by Eq.4, with sij being the rate of strain tensor (Eq.5)

2 ∂uk
τ ij = 2µsij − µ δ ij (4)
3 ∂xk

1 ∂ui ∂u j (5)
sij = +
2 ∂x j ∂xi

The mesh can be made to translate, rotate or distort in any prescribed manner, by specifying time-varying
positions for some or all of the cell vertices. For this case, an additional equation called the ‘space conservation law’
is solved for the moving coordinate velocity components.
The program use implicit methods to solve the algebraic finite-volume equations. The transient calculations are
done with PISO algorithm using predictor-corrector scheme. As a result of the decoupling of the equation for each
dependent variable and subsequent linearization, large sets of linear algebraic equations are obtained. The equations
are solved either by Conjugate Gradient (CG) type solvers or the Algebraic Multi-Grid (AMG) approach4.Each run
required approximately 10 hours on a HP 4000 workstation.

A. Computational Domain
O-type grids are used around the airfoil, with a rectangular region close to the profile. The flapping motion is
implemented by user-defined subroutines by moving the computational domain. The grid domain consists of 57500
cells with a domain of 15c length diameter. The arbitrary mesh-interface option is used to join dissimilar mesh
structures, in order to decrease the number of cells at the farfield locations. A finer grid domain is used close to the
airfoil and a coarse gridding is implemented at farfield.

B. Boundary Conditions
On the airfoil surface, the instantaneous translational and rotational velocities are prescribed by user-defined
subroutines; imposing no-slip, no-penetration boundary conditions. At the farfield, the pressure boundary conditions
are applied. In hover condition, the farfield pressure is assumed to be known and taken to be the standard air
pressure. The velocities at the corresponding cell faces are linked to the local pressure gradients by special
momentum equations, whose coefficients are equated to those at the cell centre. These equations, together with the
continuity constraint, effectively allow the magnitude and direction of the local flow (which may be inwards or
outwards) to be calculated4. For 2-D calculations, the front and back side of the grid domain are defined as
symmetric boundary conditions.

4
American Institute of Aeronautics and Astronautics
 IV. DNS Results
Different cases are investigated by changing the parameters such as the angle of attack, location of the start of
change of incidence, location of start of change of velocity and the axis of rotation. Instantaneous forces, pressure
distributions and vorticity contours obtained for these cases are compared with each other. The different effects of
these parameters on instantaneous aerodynamic force coefficients are investigated in this section. Mean
aerodynamic coefficients (Eq.6-Eq.8) are calculated as the time average of instantaneous forces throughout the
cycle. The 7th stroke is taken as due to the fact that the effects of the impulsive start are negligibly small. The drag is
defined as the force opposing the direction of the airfoil translational motion.

1 t =7 T
CL = C L (t )dt (6)
T t =6T

1 t =7 T
CD = C D (t )dt (7)
T t = 6T

The total mechanical power required to flap the wings is dominated by profile power, which is the power
required to overcome the drag on the flapping wings1. This power does not include that required for the rotational
motion of the wings. The coefficient of profile power and profile power per unit span are calculated as in Eq.8 and
Eq.9 respectively, where n=1/T is the wing beat frequency. The results are tabulated in Table 2 and Figure 3.

P ' pro t = 7T V 3 (t )
C Ppro = =n C D (t ) ⋅ dt (8)
1 ρ cV0 3 t =6T
V0
3
2
t =7 T
P ′ pro = n D′(t ) ⋅ V (t )dt (9)
t = 6T

Symmetrical hovering is not necessarily done in horizontal plane. There exists a plane named stroke plane
making β angle relative to horizontal axis [Fig.2]. All the calculations are done relative to this stroke plane. But, if
we take into consideration the hover of an insect or bird, there is a resultant force averaged in a stroke period which
must balance the weight of the animal.

Table 2 DNS average force and force coefficient results for different cases
Stroke
α P′pro
xv xa Period D′ [N/m] L′ [N/m] CD CL CFtotal β CPpro
[°] [W/m]
[s]
5 2c 1c 0.09824 3.6856E-3 2.9341E-3 0.2847 0.2266 0.3639 51.48 3.9707E-3 0.1584
5 2c 1.5c 0.09824 3.2672E-5 2.1516E-3 0.2524 0.1662 0.3022 56.64 3.3701E-3 0.1328
5 2c 2c 0.09824 2.9542E-3 1.2171E-3 0.2282 0.0940 0.2468 67.61 2.8790E-3 0.1142
5 2c 2.5c 0.09824 2.5797E-3 -6.48E-6 0.1993 -5.01E-4 0.1993 90.14 2.3889E-3 0.1017
30 2c 1c 0.09824 9.2442E-3 0.009416 0.7140 0.7273 1.0192 44.47 0.01117 0.4936
30 2c 1.5c 0.09824 8.3148E-3 0.009170 0.6422 0.7083 0.9561 42.20 0.01004 0.4492
30 2c 2c 0.09824 8.0837E-3 0.009328 0.6244 0.7205 0.9534 40.91 0.00963 0.4344
30 2c 2.5c 0.09824 8.3534E-3 0.009666 0.6452 0.7466 0.9868 40.83 0.00955 0.4300
45 2c 1c 0.09824 1.5617E-2 0.010747 1.2062 0.8301 1.4642 55.46 0.019439 0.8758
45 2c 1.5c 0.09824 1.4618E-2 0.011052 1.1291 0.8536 1.4155 52.91 0.018385 0.8428
45 2c 2c 0.09824 1.3954E-2 0.011401 1.0778 0.8806 1.3918 50.75 0.0177426 0.8322
45 2c 2.5c 0.09824 1.5206E-2 0.013047 1.1745 1.0078 1.5476 49.37 0.01931 0.9211
60 2c 1c 0.09824 2.2510E-2 0.010496 1.7386 0.8107 1.9183 65.00 0.028533 1.3494
60 2c 1.5c 0.09824 2.1817E-2 0.01099 1.6852 0.8490 1.8870 63.26 0.0283408 1.3384
60 2c 2c 0.09824 2.1933E-2 0.01144 1.6941 0.88345 1.9106 62.46 0.028558 1.3558
60 2c 2.5c 0.09824 2.3036E-2 0.012302 1.7793 0.9502 2.0171 61.90 0.02979 1.4163

5
American Institute of Aeronautics and Astronautics
The resultant force coefficient is:

2 2
C Ftotal = C D + C L (10)

The angle β corresponding to the resultant force relative to the horizontal plane axis is called the stroke plane
inclination angle and is given by Eq.11.

CD
β = tan −1 ( ) (11)
CL

The value of stroke-plane inclination is assumed to be between

20° and 40° for a dragonfly who is performing asymmetrical
hovering3. It is observed that the drag coefficient is approximately
the same order of magnitude as lift coefficient. By performing a
flapping motion in required stroke plane, considerably high
resultant force coefficients can be obtained to overcome the weight
of the whole body. As will be discussed later, the drag coefficient
here is calculated with respect to the motion of the airfoil, the force
opposing to the motion. In some circumstances, with respect to the
body, this force turns out to be a thrust for the whole body. The first
aim is to estimate the lift or the normal force coefficient with
respect to the airfoil position during the flapping motion.
Figure 2 Stroke plane definition for hover

2 2
1.8 1.8
1.6 1.6
Drag Coefficient

1.4 1.4
Lift Coefficient

1.2 1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
5 30 45 60 5 30 45 60
angle of attack [deg] angle of attack [deg]

a) Lift coefficient CL b) Drag coefficient CD

2 2
1.8 1.8
Profile Power Coefficient
Total Force Coefficient

1.6 1.6
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
5 30 45 60 5 30 45 60
angle of attack [deg] angle of attack [deg]

c) Total Force coefficient CFtotal d) Profile power coefficient CPpro

Figure 3 Aerodynamic force coefficients for xv=2c, Re=1000, a=1/4 axis of rotation averaged during
the 7th stroke

6
American Institute of Aeronautics and Astronautics
 A. Effects of the Position of Axis of Rotation and xa location:
Three different position for axis of rotation is investigated namely ¼, ½ and ¾ chord locations. At ½ chord
rotation, two positive peak values are observed at the beginning and at the end of each stroke (Fig. 4b). For ¼ chord
position, the lift peak is at the end of a stroke and for the ¾ c position it is at the beginning of the stroke. For the last
case (at the end of the stroke) a negative peak forms. The parameter xa is defined to be the position where the
variation of the angle of attack starts beyond a specific time ta. On Figure 4, in addition to the different position for
axis of rotation , xa variations are also demonstrated to show their influence on aerodynamic force coefficients for
xv=2c. Fig. 5 and Fig. 6 show the vorticity contours with velocity vectors for ¼ chord and ½c chord position rotation
respectively, during the 7th stroke. Warm tones (reds) correspond to counter-clockwise vorticity; cool tones (blues)
correspond to clockwise vorticity. Different types of vortices are observable such as leading edge vortex (LEV),
translational starting vortex (TSV), rotational starting vortex (RSV) and shear layer vortex (SLV).

a) a=¼c b) a=½c

c) a= ¾ c

Figure 4: Lift and drag coefficients for α=30°, xv=2c, Re=1000 with different position of axis of rotation during 7th
stroke

7
American Institute of Aeronautics and Astronautics
0.5924 0.5949 0.5973 0.5998

0.6023 0.6047 0.6072 0.6096

0.6121 0.6145 0.6170 0.6194

0.6219 0.6244 0.6268 0.6293

0.6317 0.6342 0.6366 0.6391

0.6415 0.6440 0.6465 0.6489

0.6514 0.6538 0.6563 0.6587

0.6612 0.6636 0.6661 0.6686

0.6710 0.6735 0.6759 0.6784

0.6808 0.6833 0.6857 0.6876

Figure 5: Vorticity contours with velocity vectors for α=30°, xv=2c, xa=2c, Re=1000 with rotation axis at ¼c during 7th
stroke. Each frame is (0.001*T) second apart and the numbers in the lower left-hand corners represent the time in second.

8
American Institute of Aeronautics and Astronautics
0.5924 0.5949 0.5973 0.5998

0.6023 0.6047 0.6072 0.6096

0.6121 0.6145 0.6170 0.6194

0.6219 0.6244 0.6268 0.6293

0.6317 0.6342 0.6366 0.6391

0.6415 0.6440 0.6465 0.6489

0.6514 0.6538 0.6563 0.6587

0.6612 0.6636 0.6661 0.6686

0.6710 0.6735 0.6759 0.6784

0.6808 0.6833 0.6857 0.6876

Figure 6: Vorticity contours with velocity vectors for α=30°, xv=2c, xa=2c, Re=1000 with rotation axis at ½c during 7th
stroke. Each frame is (0.001*T) second apart and the numbers in the lower left-hand corners represent the time in second.
9
American Institute of Aeronautics and Astronautics
B. Effects of the translational acceleration/deceleration position xv
In Figure 7, the lift and drag coefficients for different xa locations are shown for comparison of the cases for
xv=2c and xv=2.5c. The same definition for xa has been done used for velocity variation where the parameter is
denoted as xv. As the acceleration position gets closer to the maximum amplitude location, for the same Reynolds
number, the period gets smaller in order to provide the same velocity at the end of acceleration/deceleration period
in the desired location. The gray regions correspond to the quarter period of the xv=2.5c cases and the vertical
dashed lines correspond to the quarter-periods of xv=2c cases. The signs on the x-axis are the positions of the
quarter-chord (circular signs) or time tv location (diamond signs) for cases xv=2c (blue) and xv=2.5c (red)

a) xa=1c b) xa=1.5c

c) xa=2c d) xa=2.5c

Figure 7: Lift and drag coefficients comparing different xv locations for α=45°, Re=1000, 1/4c rotation
during 7th stroke. Each graph corresponds to a different xa position.

10
American Institute of Aeronautics and Astronautics
C. Effect of the angle of attack α

Leading edge vortex (LEV), translational starting vortex (TSV), rotational starting vortex (RSV) and shear layer
vortex (SLV). All the vorticity intensities increases as the starting angle of attack α increases. The rotational starting
vortex is higly strong for high angle of attack values. For 30°, the leading edge and translational vortex are smooth
and attached during the translational region but for 60° angle of attack, new vortex formation is observed during this
region (Fig.8).

TSV LEV (clockwise) TSV

LEV (clockwise)
(counter-clockwise) deattached during (counter-clockwise)
translational
motion

SLV
(counter-clockwise)_
Trace of TSV
Trace of TSV (clockwise)
(clockwise)
t= 0.5924 s t= 0.5924 s
LEV LEV
LEV LEV Trace of TSV
(counter- (clockwise) (counter- (clockwise) (counter-clockwise)
clockwise) clockwise)
RSV
(counter-clockwise)

RSV
TSV (counter-clockwise)
TSV
(clockwise) (clockwise)

t= 0.6194 s t= 0.6194 s

Trace of TSV LEV LEV

Trace of TSV (counter-
(clockwise) (counter-
(clockwise) clockwise)
clockwise)

RSV
(clockwise) RSV
(clockwise)

t= 0.6636 s t= 0.6636 s
a) α=30° b) α=60°
Figure 8: Vorticity contours for xv=2c, xa=1c, Re=1000 with rotation axis at ¼c during 7th stroke.

11
American Institute of Aeronautics and Astronautics
 V. Analytical Model

A. The estimation of Force Coefficients using the Duhamel Integral with Wagner and Küssner Functions by
implementing induced velocities from DNS results
The indicial functions can be used to calculate the lift and moment on a wing undergoing an arbitrary motion by
means of Duhamel’s integral assuming attached flow conditions and based on the linear superposition assumption4.
The indicial lift on a 2-D wing in incompressible flow was first derived by Wagner5. Figure 9 shows the Wagner
indicial function φw given below as Eq.13 and the Küssner function given as Eq.14 (see, Refs. [6] to [10]). Arbitrary
variations in free-stream 10, 11 are mostly implemented by use of Eq.12.

s
dw3 / 4 c
Lcirculatory = πρV (t ) ⋅ S ⋅ w3 / 4 c (0)φ ( s) + (σ ) ⋅ φ ( s − σ )dσ (12)
0
dσ

In this study, φw is taken to be the Wagner Function with

φw ( s) = 1.0 − 0.165e −0.0455s − 0.335e −0.3s (13)

and ψ is the Küssner Function which is approximated with exponential form as:

ψ ( s ) = 1.0 − 0.5e −0.13s − 0.5e −1.0 s (14)

t
where s is the reduced time defined as s = 1 Vdt .
b 0

The non-circulatory or apparent mass terms arise

from the ∂Φ / ∂t term contained in the unsteady
Bernoulli equation and account for the pressure forces
required to accelerate the fluid in the vicinity of the
airfoil. It is the results of instantaneous local
Figure 9: Wagner and Küssner Functions
accelerations (Eq.15).

∂
L NC = πρb 2 (V sin α ) + πρb 2 ∂ θ b (15)
∂t ∂t 2

Since the angle of attack increases during the rotational region (which approaches to 90° at the end of the
stroke), the method does not respond too well in these regions. It is also observed that the multiplication of this
region with V(t) is incorrect as V(t) 0 and 90°. The first part of the circulatory lift is the steady state value at
t=0. The Duhamel Integral calculates the lift using V(t) and (t) in the region of time dependent flow for given
initial values.

C Lα s
dw3 / 4 c
Lcirculatory = ρV (t ) ⋅ S ⋅ w3 / 4 c (0)φ ( s ) + (σ ) ⋅ φ ( s − σ )dσ + L3 (16)
2 0
dσ

Since the lift is multiplied by V(t), the steady state lift at s=0 is lost for V(t)≠Vo. Thus, this value (called L3 in
Eq.16) is added to the total circulatory lift.

V (0) − V (t ) (17)
L3 = Lss ⋅
V (0)

12
American Institute of Aeronautics and Astronautics
where the steady state lift at initial position t=0 is

C Lα C
Lss = ρ ⋅ ( − V (0) ) ⋅ S ⋅ [w3 / 4 c (0)] = Lα ρ ⋅ (− V (0) ) ⋅ S ⋅ [− V (0) ⋅ sin α (0)] (18)
2 2

The coefficient CLα implemented in Eq.16 is

the averaged normal force coefficient (Fig.10)
found from impulsive motion analysis, divided by
sinα. (It is demonstrated that the starting location
for averaging the normal force coefficient
corresponds to a time value where the influence of
the impulsive start on aerodynamic forces
diminishes.) The slopes obtained by DNS
calculations are incorporated into the circulatory
lift equation rather than the value of 2π since for
high angle of attacks the said curve slope is no
longer valid.
The hovering flight problem is commonly
Figure 10: Lift, drag and normal force coefficients of solved in the literature by the actuator disk and its
steady state values obtained from impulsive start motion associated Rankine-Froude momentum jet. In the
with a 1m/s velocity using a DNS code for different angle literature the actuator disk theory is mostly used to
of attack values. Data averaging for steady state values predict the induced fluid velocities predicted via
from unsteady impulsive result solution is done in the the momentum jet theory and the results are
interval [T, 5T] where T corresponds to a stroke period of combined with other theories such as blade-
8, 9, 10
the flapping motion. element theory . For an insect or
hummingbird the usual momentum-jet estimate
may be from 10 to 15% too low2. Thus, in
addition to the Wagner impulsive function, the
sudden gust velocity change, Küssner function (Eq.14), is implemented into the program developed where the
induced velocities for each case are approximated using the actuator disc theory. The weight is taken to be average
lift during a period per unit span obtained from the DNS code and a semi-empirical correlation involving feathering
parameter could be implemented in the model.

s
C Lα dw3 / 4c
Lcirculatory = ρ ⋅ (− V (t ) ) ⋅ S ⋅ w3 / 4c (0)φ ( s) + (σ ) ⋅ φ ( s − σ )dσ + L3
2 0
dσ (19)
C Lα s
dw3 / 4c s
dw g _ 3 / 4c
+ ρ ⋅ (− V (t ) ) ⋅ S ⋅ w g _ 3 / 4c (0)ψ ( s) + (σ ) ⋅ φ ( s − σ )dσ + (σ ) ⋅ψ ( s − σ )dσ + L3k
2 0
dσ 0
dσ

Mg L′
wg = = (20)
2 ρA 2 ρ ⋅ 6c

where the area A is taken to be rectangular region covered by the airfoil during its overall motion. The calculated wg
value is taken to be constant all over the motion, where it is assumed that the airfoil enters into a constant gust with a
velocity of wg (Table 3). The effect of this velocity in normal direction of the airfoil at ¾c location is
wg_3/4c=wgcosα(t) and this gust velocity is implemented in circulatory lift calculations with Küssner function. Thus,
the noncirculatory lift is defined as:

LNC = πρb 2
∂
∂t
(V sin α ) + πρb 2 ∂ θ b + πρb 2 ∂ wg cos α
∂t 2 ∂t
( ) (21)

13
American Institute of Aeronautics and Astronautics
It must be noted that even though the gust velocity, Table 3 Induced velocity calculated by actuator
where the airfoil is assumed to be entering during the disk theory using DNS average lift results for
motion, is taken to be a constant throughout the motion, different cases for Re=1000 and wing beat
the value of wg_3/4c changes as the angle of attack frequency, n=10.18 Hz
changes. This results gives rise to a correction in α
xv xa wg [m/s] f
noncirculatory lift which is also implemented in the [°]
program written in Fortran90 (Eq. 21). But, it is 5 2c 1c 0.1413 0.01338
observed that, order of magnitude of this correction is 5 2c 1.5c 0.1210 0.00981
very small and can be neglected since the angle of attack 5 2c 2c 0.091 0.00555
goes to 90°. All other times, the angle of attack remains 5 2c 2.5c 0 0
mostly constant thus angular velocity is zero or very 30 2c 1c 0.2531 0.04293
30 2c 1.5c 0.2498 0.04181
small.
30 2c 2c 0.2519 0.04253
Figure 11 represent the analytical model obtained
30 2c 2.5c 0.2564 0.04407
using Duhamel Integral approach implementing the 45 2c 1c 0.2703 0.04900
curve slope values obtained via the DNS code. The 45 2c 1.5c 0.2742 0.05039
results are highly comparable with DNS solution. The 45 2c 2c 0.2785 0.05198
comparison was done for normal force coefficients but it 45 2c 2.5c 0.2979 0.05949
is observed that the analytical model is closer to the lift 60 2c 1c 0.2672 0.04786
coefficient results especially at the peak values during 60 2c 1.5c 0.2734 0.05011
rotational phase. More detailed explanation of the 60 2c 2c 0.2790 0.05215
analytical model developed is given in Ref. [3]. 60 2c 2.5c 0.2893 0.05610

a) xa=1c b) xa=1.5c

c) xa=2c d) xa=2.5c
Figure 11: Lift coefficient, normal force coefficient and Duhamel integral solution with Wagner and
Küssner functions for α=30°, xv=2c, Re=1000 at 1/4c rotation during 7th stroke using dCN/dα
α values

14
American Institute of Aeronautics and Astronautics
 VI. Conclusion
In this study, the flapping motion aerodynamics is considered for a symmetrical hovering case in the use of
future Micro Air Vehicle applications. Different parameters are investigated and discussed in order to compare the
lift, drag, normal force values in addition to the profile power requirements. It is observed that after an angle of
attack 30° the positive lift values are obtained throughout the motion. The analytical model developed through the
use of Wagner and Küssner functions and Duhamel Integral (with Rankine-Froude momentum jet theory) gives
comparable results to those obtained by the DNS code. This was implemented here in order to investigate the effect
of different parameters and their influences on the aerodynamic force coefficients. Comparison with experimental
results is deferred to a future paper.

Acknowledgments
Financial support provided by the French Government as a scholarship which made it possible for the first
author, Miss D. Funda Kurtulus, to realise these research activities as part of a joint PhD programme between
ENSMA and METU is greatly appreciated. The financial support provided by METU and scholarship provided by
TUBITAK during this PhD programme is also greatly acknowledged.

References
1
Shy W., Berg M., Ljungqvist D., "Flapping and Flexible Wings for Biological and Micro Air Vehicles," Progress in
Aerospace Sciences, Vol.35, 1999, pp. 455-505
2
Rayner J. M. V., “A Vortex Theory of Animal Flight. Part 1. The vortex wake of a hovering animal,” J. Fluid Mech.,
Vol.91, part 4, 1979, pp. 697-730
3
Kurtulus D. F., Farcy A., Alemdaroglu N., “Numerical Calculation and Analytical Modelization of Flapping Motion in
Hover,” First European Micro Air Vehicle Conference and Flight Competition, Braunschweig Germany, 13-14 July 2004
4
Star-CD Version 3.10A, “Methodology Manual,” Computational Dynamics Limited, 1999
5
Sane P. S., Dickinson M. H., “The Control of Flight Force by a Flapping Wing: Lift and Drag Production,” The Journal of
Experimental Biology, No. 204, 2001, pp.2607-2626.
6
Lomax H., “Indicial Aerodynamics,” Manual on Aeroelasticity, edited by W. P. Jones, Nov. 1960, Part II, Chap.6, pp.1-58
7
Drela, M.,Youngren H., “Xfoil 6.9 User Guide,” 11 Jan 2001, pp. 1-33
8
Weis-Fogh, T., “Energetics of hovering flight in hummingbirds and Drosophila,” J. Exp. Biol., No. 56, 1972, pp. 79-104
9
Weis-Fogh, T., “Quick estimates of flight fitness in hovering animals, including novel mechanisms for lift production,” J.
Exp. Biol., No. 59, 1973, pp. 169-230
10
Norberg, U. M., Kunz T. H., Steffensen, J. F., Winter Y., Helversen Y., ‘The cost of Hovering and Forward Flight in a
Nectar-Feeding Bat, Glossophaga Soricina, Estimated From Aerodynamic Theory”, J. Exp. Biol., No. 182, 1993, pp. 207-227
11
Katz, J., Plotkin, A., “Low Speed Aerodynamics,” Cambridge University Press, Cambridge, 2002, Chap. 13

15
American Institute of Aeronautics and Astronautics

View publication stats