You are on page 1of 30

Mechanisms and Efficiency of Wing Flapping

AA200B
Lecture 14
November 29, 2007

AA200B - Applied Aerodynamics II

AA200B - Applied Aerodynamics II

Lecture 14

Introduction
Interest in the aerodynamics of flapping flight has been rekindled with
consideration of micro-air vehicles, autonomous underwater vehicles, and
recent experiments with insect models. Much of the previous and current
research in this area is empirical due to the complexity of the relevant flows,
although some mechanisms have been identified and postulated as being
important to flapping performance. In these notes, we examine some of
the basic mechanisms for efficient flapping flight with analysis suitable for
design. The analysis starts simply with a quasi-steady look at flapping in
forward flight.

AA200B - Applied Aerodynamics II

Lecture 14

Introduction: Flapping Propulsion Myth and Reality


An analysis of bumblebee flight in 1934 [3], concluded that according to
the laws of the resistance of the air applied to insects, for them flight is
impossible. Since that time, and likely long before, people have imagined
that certain mysteries of flapping propulsion elude aerodynamicists. Recent
years have brought a number of articles that seek to explain flapping
flight with complex unsteady viscous effects and vortex dynamics. To be
sure, the hovering of insects involves very complicated low Reynolds number
flow phenomena, but some of the basic concepts can be understood quite
simply and that is the aim of this introduction.
Those studying fish and marine mammal locomotion have their own
version of the bumblebees cannot fly myth:
In 1936 the British zoologist James Gray created a stir by
calculating the power that a dolphin would need to move at 20
3

AA200B - Applied Aerodynamics II

Lecture 14

knots, as some were reported to do. Gray assumed that the resistance
of the moving dolphin was the same as that of a rigid model and
estimated the power that the muscles of the dolphin could deliver.
His conclusion, known as Grays paradox, was that the dolphin was
too weak, by a factor of about seven, to attain such speeds. The
inescapable implication is that there are flow mechanisms at work
around the body of the moving dolphin that lower its drag by a factor
of seven. From [9].
Similarly, the mechanism for swimming propulsion is often considered
a mystery, with recent issues of Nature [1] declaring that One number
explains animal flight.
One number describes the beating of animal wings and tails,
researchers have found. The simple rule of thumb for animal
locomotion could help to design miniature flying machines.
4

AA200B - Applied Aerodynamics II

Lecture 14

Weve described the geometry of the wingbeat, says Graham


Taylor of the University of Oxford, UK. Swimmers and fliers from
insects to whales all cruise at the speed that lets them slip along most
easily, he and his colleagues show.
Wings and tails create eddies as they move. These need to be
left behind, because turbulent air or water is more difficult to travel
through. So limbs shed vortices at the bottom of their downstrokes.
Flap too quickly, and you have to fight this turbulence on the way up.
Too slowly, and turbulence sticks.
This concept is described in a bit more detail in some lectures by
researchers at M.I.T.:
Exploitation of Vortical Wake Dynamics by Live Fish
Evidence suggests that many fish exploit the natural instability of
the flow energetics to assist them in propulsion and maneuvering. By
5

JET

AA200B - Applied Aerodynamics II

Lecture 14

tuning their own kinematics, the fish is able to swim efficiently, to


generate large thrust and turning forces, and to move silently through
the flow with minimal wasted energy. From [10].
MIT Center for Ocean Engineering, 2005

Cylinder wake: Krmn vortex street, induced jet flow


towards the body, causing drag force

Control volume

Fish wake: reverse Krmn vortex street, induced jet


flow away from the body, causing thrust force

Control volume
MIT Center for Ocean Engineering, 2005

Figure 1. Vorticity shedding from cylinder and fish (from Ref. 10)
5

AA200B - Applied Aerodynamics II

Lecture 14

This natural instability of the flow energetics may be described by the


unsteady Navier Stokes equations, but this hardly helps to understand the
mechanisms.
An article in Scientific American [9] describes this idea in more detail:
Any object in a flow, whether it is a wire in the wind or a swimming
swordfish, creates a trail of spinning vortices. The wire obstructs the
flow and leaves a wake, whereas the tail of a fish pushes water
backward, establishing what is more properly known as a jeta column
of moving fluid that includes thrust-producing vortices. We became
convinced that these jet vortices play the central role in the generation
of thrust, and we argued that their optimal formation would increase
efficiency tremendously.
From previous studies we had done on the vortices produced by a
wire in a stream of air, we were well acquainted with a fluid-dynamic
7

AA200B - Applied Aerodynamics II

Lecture 14

parameter known as the Strouhal number. It is the product of the


frequency of vortex formation behind an object in a flow and the width
of the wake, divided by the speed of the flow. What the number
indicates, compactly, is how often vortices are created in the wake and
how close they are. Interestingly, the ratio remains constant at about
0.2 for a variety of flow conditions and object shapes.
Although the Strouhal number was invented to describe the wakes
behind flow obstructions, the similarities between wakes and jets are
such that we realized we could use the number to describe jets. For
a swimming fish, we defined the Strouhal number as the product of
the frequency of tail swishing and the width of the jet, divided by the
speed of the fish.
By analyzing data from flapping foils, we found that thrustinducing vortices form optimally when the Strouhal number lies
between 0.25 and 0.35. We anticipated that efficiency should be
at a maximum for these values. Some preliminary experiments at the
8

AA200B - Applied Aerodynamics II

Lecture 14

M.I.T. testing tank confirmed that the efficiency of a flapping foil


does indeed peak when the Strouhal number is in this range.
There is nothing wrong with this approach to understanding flapping
propulsion, but it is a kind of far-field analysis of a phenomenon that may
also be understood in the near field. In the following sections we attempt
to describe flapping propulsion in this way, relying on the more conventional
approaches to aircraft performance analysis.

AA200B - Applied Aerodynamics II

Lecture 14

Basic Mechanism
How flapping is just like gust soaring In a previous chapter, we considered
how animals might extract energy from gusts and propel themselves by
varying the lift of their wings in phase with the ambient air motion. Figure
2 is the picture we used to understand how thrust might be generated from
vertical gusts. If, however, we think of figure 2 as a top view of a fish tail
and the gusts are caused by the tail motion, the situation is completely
equivalent.

Figure 2. Forces on flapping wing (e.g. fish tail viewed from above)
If a 2D wing is moved through a fluid with a velocity of v and the system
10

AA200B - Applied Aerodynamics II

Lecture 14

is moving forward at speed U, it generates a forward thrust given by:


T =

Lv
U

If the lift and lateral velocity are properly phased, the average thrust is
given by:
L0v0

T =
2U
This is the most basic (2D, quasi-steady) way to think of thrust
production due to wing flapping. Many effects are left out here and
can be added in increasingly complex expressions for the average thrust.
The added lift on the wing also generates some unsteady drag, but for
low frequencies the majority of the extra drag is due to 3D quasi-steady
11

AA200B - Applied Aerodynamics II

Lecture 14

effects. If we include these effects:


2
L
v
L
0
0
0
T =

2U
2qb2

The average power required to move the wing is given by:


L0v0

P =
2
so the net efficiency is:
4T
U L0
=1 2 2
=1
2
qv0b
v0 b
Which suggests that for a given thrust, we should move the wing as quickly
as possible.
12

AA200B - Applied Aerodynamics II

Lecture 14

However, if the wing is not articulated, moving the wing very quickly
will produce an angle of attack that is too large and the wing will stall. If
the wing incidence can be changed by, say 30 deg, and the maximum angle
of attack is about 15 deg then the maximum value for v0 is equal to U
(since the total angle is 30 + 15 = 45 deg).
If the motion is generated by oscillating the aft part of the vehicle with
an amplitude, a, the Strouhal number based on overall tail amplitude (2a)
is:
2af
2a
v0
St =
=
=
U
2U
U
So, if v0 = U , the Strouhal number is 0.318. If v0 is restricted by stalling:
St =

1
tan (max + i)

With an incidence change restricted to 20 deg, the Strouhal number is


0.21. These are just in the range suggested for efficient swimming. (No
13

AA200B - Applied Aerodynamics II

Lecture 14

mysterious unsteady vortex interaction is needed to explain this, its just


kinematics.)
The Strouhal number is just one dimensionless frequency measure.
Another is the classical reduced frequency, k, which is defined as: k =
c
c
2U = a St . For a wing with a motion amplitude of 5 times the chord
length at a Strouhal number of 0.3, the reduced frequency is about 0.2,
meaning that unsteady effects are not insignificant, but that the quasi-steady
approximation should be quite good for initial design purposes.

14

AA200B - Applied Aerodynamics II

Lecture 14

Nonlinear Effects
Of course, with v0 U we need to include viscous drag increments
associated with higher dynamic pressure. When this is done, some algebra
confirms that the maximum efficiency occurs at high L/D and at v0 U .
Thus, the Strouhal number may be limited to values less than 1 = .318
due to kinematics, but otherwise should approach this value. (As an
approximation = CL 1/v
1+v with  = CD /CL . Even with low L/D values,
the optimal v0/U is only somewhat larger, confirming the significance of
the 0.3 Strouhal number using simple performance arguments.)
To see how this result comes about, consider the thrust produced by
flapping:
Lv
DU
Fx = L sin D cos =

2
2
v +U
v2 + U 2
15

AA200B - Applied Aerodynamics II

Lecture 14

.
= 0.5CLv

v 2 + U 2 0.5CD U

v2 + U 2

.
p
Cx = v2 + 1 (CLv CD )
.
Cz =

v2 + 1 (CL + vCD )

. The efficiency of the system is the average value of Cx (with the zero lift
drag coefficient removed if the wing must be there for other reasons) divided
by the average value of Cz v. This is easily computed in a spreadsheet with
results shown in figure 3. Additional nonlinear effects not considered in the
above include nonplanar wake motion and non-quadratic profile drag polars,
which could be important for certain designs.
16

AA200B - Applied Aerodynamics II

Lecture 14

Efficiency vs. v/U


0.9

Efficiency

0.85
0.8
0.75
0.7
0.65
0.6
0

0.5

1.5

v/U

Figure 3. Efficiency of an oscillating wing based on quasi-steady aerodynamics


but with nonlinear effects of viscous drag. CD0 = 0.01, CL0 = 1, AR = 5

17

AA200B - Applied Aerodynamics II

Lecture 14

Unsteady Effects
In 1936 I. E. Garrick published NACA Report 567, Propulsion of a
Flapping and Oscillating Airfoil, in which he extended Theodorsens 2D
simple harmonic results to include the horizontal force as well as the vertical
force of a pitching and plunging thin airfoil. He did this in two ways (and
showed that the answers are the same): by using thin airfoil theory to
estimate the leading edge suction force and by computing the energy left
in the wake due to the shed vorticity. The details are provided in Garricks
paper (posted on the class website), but may be summarized as follows.
If an airfoil is pitching, (t), about an axis located at ac/2 behind the
mid-chord and moving vertically with the displacement of the rotation axis
given by h(t), the we can write:
(t) = 0eit+0
h(t) = h0eit+2

18

AA200B - Applied Aerodynamics II

Lecture 14

From Theodorsen (with h positive downward and Fz positive upward):


ba)
+ 2U bC(k)Q
Fz = b2(U + h


1
1

M = b2 U ( a)b + b2( + a2) abh


2
8
with:

1
Q = U + h + b( a)
2

From Garrick, assuming small angles:


Fx = bS 2 + Fz
with:
S=

r
2C(k)Q

1
b
2
19

AA200B - Applied Aerodynamics II

Lecture 14

The work done to maintain the pitching and plunging is:


W = Fz h + M
So the efficiency over one cycle is:
FxU
=

Fz h + M
this
In the simple case of pure plunging motion ( = 0, Q = h),
becomes:
2U
bS2U
b2(C(k)Q)
F 2 + G2
=
=
=

F
Fz h
2U bC(k)Qh
Here, F and G are the real and imaginary components of C(k).
20

ULSION OF A FLAPPING AND OSCILLATING AIRFOIL


AA200B - Applied Aerodynamics II

Lecture 14

sent). It is observed that a propelling force exists in


The resultthefor
thisrange
caseofisl/k,
plotted
below. being
The50case
for pitching and
entire
the egciency
percent
d+ (a3+c3> Po2
100 percent and
for is left as an
Cor infinitely
oscillations
and amplitudes
plunging
depends
on the rapid
relative
phases and
O -k2
6.5)
POh,
%POI
exercise.(29) Lnfinitely slow flapping.
d (29) agree we must

ge propulsive force is

.80

.60

Pz V

T O

.zo

'

'

"

'

f0

'

'

"

'

f6

f8

educed to an identity,
m (15)) (23)) and (26)

l/k

I4

fZ

20

FIGURE3.-The ratio-of energy of propulsion to the energy required to maintain the

( 5 )as a function
of l/k for the case of pure flapping.
Note that even for relatively
low frequency
motion (k = 0.2) substantial
decreases in efficiency
20%)
appear.
For the (>
special
case
of angular oscillations about a

ing relation must hold

hat

oscillations

PZV

(a

alone (h=O, p=O) the horizontal force is

+G2)[; +

p,=Tpbp2b2a02{
(P

+;(

-a )

-a ) ]

-;
(+a);]

21

(34)

AA200B - Applied Aerodynamics II

Lecture 14

3D Considerations
Additional considerations apply in 3D. For a useful discussion of some
of these including the effects of large amplitude motion, see Ref. [11].
Even for small motions the 3-dimensionality of the flow makes 2D results of
limited use. Here we mention two fundamental issues.
The first is that, just as in steady flow, the wing sheds vorticity that
trails downstream, induces downwash on the wing, and creates drag. Even
when there is no average lift on the wing, induced drag is created by the 3D
unsteady trailing vorticity. However, in this case the quasi-steady analysis
is pessimistic, since the time-varying vorticity leads to reduced downwash
compared with the constant strength sheet. As the frequency increases, the
extra induced drag caused by flapping is actually reduced, while the losses
due to transverse vorticity increase. (Note that in the linear theory, these
effects of these vorticity components may superimposed.)
22

AA200B - Applied Aerodynamics II

Lecture 14

The second, very important 3D consideration is associated with optimal


loadings for flapping flight. This problem was addressed in a simple, quasisteady analysis by R.T. Jones [13], and with a more refined study described
in Ref. [11]. To illustrate some of these ideas, consider the question of
optimal loading in the linearized, quasi-steady case. Just as described in the
discussion of minimum induced drag, we can use the method of restricted
variations to determine the optimal load distribution for flapping.
Consider arbitrary, small simple flapping motion where the local motion
of the wing is given by wf (y), and for which the additional loading during
the flapping cycle produces a downwash distribution of ww (y). Assume also
that we add a small increment to the loading at a station y1 and y2. In this
case, the change in total thrust is given by:
T = wf 11 + wf 22 ww 11 ww 22
23

AA200B - Applied Aerodynamics II

Lecture 14

The change in power required to flap the wings is:


P = U 1wf 1 + U 2wf 2
If we hold the power constant (P = 0) and seek the optimal loading,
then T = 0. Substituting the second expression into the first and setting
it to 0 shows that for arbitrary (but small) flapping motions, the optimal
loading should be that associated with a downwash distribution ww that
is proportional to the local motion due to flapping, wf . So for simple
plunging, the optimal loading is elliptic, while for rotation about a central
hinge, the optimal downwash is linear. This then would allow us to compute
the lift distribution and then solve for the required wing twist.

24

AA200B - Applied Aerodynamics II

Lecture 14

Initial Conclusions
The conclusion is that a well-designed flapping propulsion system may
be better or worse than a propeller in terms of efficiency, depending on
the constraints on the many parameters. The disadvantages of flapping
propulsion for fish and AUVs include:
Low frequencies require high gear ratios and heavy or complex
transmissions
A dorsal fin may be required to counteract the periodic sideforce
Time-dependent incidence changes are required, but might be achieved
with passive hydroelastic tailoring
The advantages of flapping propulsion include:
25

AA200B - Applied Aerodynamics II

Lecture 14

Elimination of a rotary bearing may be useful at high pressures


The fin area may be needed for lateral stability/control as well as
propulsion
Small angular deflections may reduce possibility of fouling
When the flapping wing is also used for lift, as in birds and insects, the
surfaces must be located near the c.g. and cannot be translated so easily.
In this case a flapping motion in which the plunging rate varies over the
span is generally used. The actual motion varies greatly from v-like motions
to much more complex motions, but may be analyzed in an analogous
manner. The paper in these notes entitled Optimal Wing Flapping shows
how linear quasi-steady theory is applied in this case.
If we consider flapping for propulsion in forward flight, the reduced
26

AA200B - Applied Aerodynamics II

Lecture 14

frequency is related to the Strouhal number by:


k=

c
f c f A c
c
=
=
= St
2U
U
U A
A

So with flapping amplitude of about c, the two frequency parameters


are equal. With Strouhal numbers in the range of 0.3, the quasi-steady
assumption may reasonable when the amplitude is larger than about 3
chords, but not if it is much smaller. The paper in these notes Unsteady
Wing Flapping addresses some of the issues in these cases.

27

AA200B - Applied Aerodynamics II

Lecture 14

References
1. John Whitfield, One number explains animal flight, Science Update,
Nature, 16 October 2003
2. Taylor, G. K., Nudds, R. L., Thomas, A. L. R. Flying and swimming
animals cruise at a Strouhal number tuned for high power efficiency. Nature,
425, 707-711, 2003.
3. August Magnan, Le Vol Des Insects, Hermann and Cle, Paris, 1934,
and Can bumblebees fly: an internet discussion.
4. Michael H. Dickinson, Fritz-Olaf Lehmann, Sanjay P. Sane, Wing
Rotation and the Aerodynamic Basis of Insect Flight, 18 JUNE 1999 VOL
284 SCIENCE, www.sciencemag.org
5. Unsteady aerodynamic forces of a flapping wing, Jiang Hao Wu and
28

AA200B - Applied Aerodynamics II

Lecture 14

Mao Sun, The Journal of Experimental Biology 207, 1137-1150 Published


by The Company of Biologists 2004 doi:10.1242/jeb.00868.
6. Magic number revealed for flying and swimming,18:00 15 October
03, NewScientist.com news service
7. Michael Dickinson, Solving the Mystery of Insect Flight: Insects use a
combination of aerodynamic effects to remain aloft, ScientificAmerican.com,
June 17, 2001
8.
Robert Dudley, BIOMECHANICS: Enhanced:
Aerodynamics, 18 JUNE 1999 VOL 284 SCIENCE.

Unsteady

9. Triantafyllou, Michael S., Triantafyllou, George S., An efficient


swimming machine, Scientific American; Mar95, Vol. 272 Issue 3, p64.
10. Alexandra H. Techet, Fish Swimming, Lecture 1, Biological
and Medical Engineering, Center for Ocean Engineering, Department of
29

AA200B - Applied Aerodynamics II

Lecture 14

Mechanical Engineering, Massachusetts Institute of Technology, Cambridge,


MA, Spring 2005
[11] Garrick, I. E., A Review of Unsteady Aerodynamics of Potential
Flows, Applied Mechanics Review, Vol. 5, No. 3, March 1952, pp. 89-91.
[12] Hall, K.C., and Hall, S.R., Minimum Induced Power Requirements
for Flapping Flight, J. Fluid Mech. (1996), vol. 323, pp. 285-315,
Cambridge University Press.
[13] Jones, R. T. 1980 Wing flapping with minimum energy. Aero. J.
84, 214-217.

30