You are on page 1of 29

Introduction to Unsteady Aerodynamics

AA200B
Lecture 13
November 27, 2007

AA200B - Applied Aerodynamics II

AA200B - Applied Aerodynamics II

Lecture 13

Basic Concepts
Although many applications of interest in aerodynamics involve flows
that may be considered steady at least in some reference frame many
others are fundamentally unsteady phenomena. Here we consider some of
the basic concepts and analysis methods appropriate for dealing with these
cases.
Unsteady aerodynamics may be important in analyzing aircraft stability
and control, in assessing the tendency of a wing or other structure to flutter,
or to understand how birds, insects, or ornithopters propel themselves by
flapping. Although we could analyze these situations by simply specifying
boundary conditions and analyzing the time-dependent flow field with an
unsteady CFD code, it is helpful for the understanding of such flows to
consider the many different aspects of unsteady flow that lead to differences
from steady flow theory.
2

AA200B - Applied Aerodynamics II

Lecture 13

What is unsteady aerodynamics?


Unsteady phenomena may arise due to natural time dependent changes
in the flow itself (such as when flow separates from bluff body or turbulence
is created in a boundary layer) or it may be created by changes in the
position or orientation of a body, sometimes caused by interactions between
the fluid and an elastic structure.
Some phenomena are inherently unsteady, as in acoustic wave
propagation, and in some cases, a flow may be steady on one reference
frame and unsteady in another. It is often useful to consider whether a
particular reference frame leads to simpler steady equations. Although one
can write equations for the unsteady forces on a wing moving at constant
speed relative to a reference frame fixed to the undisturbed atmosphere, it
is much simpler to solve for the steady flow in a frame fixed to the body.
Similarly, a frame that rotates with propeller blades often provides simpler
analyses than an inertial frame with time-dependent blade motion.
3

AA200B - Applied Aerodynamics II

Lecture 13

In many unsteady flows of interest, the important unsteady aspects


involve not only the kinematic changes in boundary conditions caused by
the motion of a body, but the influence of an unsteady wake, and the
changes in the pressure-velocity relationship associated with the unsteady
form of Bernoullis equation.
As an introduction to some of these effects we consider unsteady wing
theory, starting from linear two-dimensional airfoil analysis and exploring
some nonlinear and three-dimensional phenomena.

AA200B - Applied Aerodynamics II

Lecture 13

Unsteady Airfoil Theory (2D)


Perhaps the simplest unsteady aerodynamic analysis involves a twodimensional airfoil free to pitch and plunge as shown in the figure below.
h
!

Figure 1. Definition of terms and geometry for a pitching and plunging airfoil
Some of the initial theory for unsteady airfoils was developed by
Theodorsen, von Karman, and Sears in the 1930s. Theodorsen [1] was
particularly interested in this problem because of its importance to wing
flutter, a coupling of aerodynamics and structural dynamics that can lead
to wing instability and failure.
5

AA200B - Applied Aerodynamics II

Lecture 13

Theodorsen started by assuming that the motion could be described as


simple harmonic and involved only small perturbations.
We start with the expression for unsteady pressure on the airfoil in
irrotational, inviscid flow (See [2]):

or:

d 2
+ V + p p = const
dt 2

(1)

2
d
p = const V
2
dt

(2)

The pressure difference between the upper and lower surfaces of a thin
airfoil is then:

d
p = (Vu2 Vl2) + (u l)
(3)
2
dt
6

AA200B - Applied Aerodynamics II

Lecture 13

For a thin airfoil the upper and lower surface velocities can be written
as: Vu = V + V and Vl = V V where V = (x)/2. And the
potential difference at some point on the airfoil x is related to the circulation
by:
Z x
Z x
(x)dx
(4)
(Vu Vl)dx =
=
0

We can thus write the pressure difference in terms of the vorticity:


Z x
d
(x)dx)
p = V(x) + (
dt 0

(5)

This is expressed in terms of the dimensionless Cp as:


Z x
2 d
2(x)
Cp =
+ 2 (
(x)dx)
V
V dt 0

(6)
7

AA200B - Applied Aerodynamics II

Lecture 13

In this expression the first term is the part associated with the steady
Bernoulli equation, while the second term comprises the unsteady part.
(We note among other things that the first term can include a singularity
at the leading edge, while the second term generally remains finite in the
limit as x 0. This means that there is no leading edge suction due to the
unsteady Bernoulli term: this part of the lift acts normal to the surface.)
Now we need to solve for the actual distribution of vorticity, (x), on
the airfoil. This can be done in a manner very similar to thin airfoil theory.
One major difference, however, is the influence of the unsteady wake and
how the Kutta condition must be modified in unsteady flows.
y

gw
x

Figure 2. Vorticity on airfoil and in wake.


8

AA200B - Applied Aerodynamics II

Lecture 13

We assume that the vorticity on the airfoil is composed of two parts:


one that would be obtained without consideration of the unsteady wake
(the quasi-steady part, 0), and a part that is added in order to compensate
for the induced velocities from the shed wake (1). We can compute
the first part using steady thin airfoil theory. The second part can also
be computed this way by treating the velocities induced by the unsteady
wake as changes in the boundary conditions. After some manipulation the
relationship between the bound vorticity component, 1 and the vorticity
shed into the wake, w is:
1
1(x, t) =

cx
x

Z
c

x0 w (x0, t) 0
dx
x0 c (x0 x)

(7)

The result is that the pressure can be expressed in terms of the quasisteady circulation, its time rate of change, and the vorticity in the wake.
9

AA200B - Applied Aerodynamics II

Lecture 13

The integrated dimensionless force coefficient, Cl, can finally be written


(nontrivial mathematics omitted here) as:

20
2 d
Cl(t) =
2
Vc Vc dt

Z
0

c
1
0(x0, t)(x0 )dx0+
2
V

Z
c

x=V t+c

w (x0, t)
p
dx
0
0
x (x c)
(8)

In the derivation of equation 8, we see that the first term is just the
quasi-steady lift coefficient. The second term is related to the unsteady
part of Bernoullis equation applied to the quasi-steady vorticity, and the
third term is a combination of both wake effects and the unsteady part in
the Bernoulli equation. The second term may also be interpreted as the
unsteady part of the load that would exist if there were no wake, which
10

AA200B - Applied Aerodynamics II

Lecture 13

would happen if there were no net circulation on the airfoil. We often write:
Cl = Cl0 + Cl1 + Cl2
where the first term represents the quasi-steady lift; the second is associated
with apparent mass (sometimes termed noncirculatory lift1); and the third
term represents the effect of the unsteady wake.
If we specify the motion and take Laplace transforms of this, we can
relate the wake strength to the quasi-steady circulation directly (using the
idea of vorticity conservation). This leads to the idea that the unsteady
wake term can be expressed as a factor multiplying the quasi-steady lift.
1

The terminology is actually a bit confusing. It is not correct to associate the apparent mass term
with the noncirculatory lift. It is possible to have no net circulation, = 0 and no quasi-steady vorticity,
0 = 0 = 0 and still have lift arising from the wake term. (If there were never any net circulation, then
the wake term would indeed be 0 when 0 was 0, but in general one might have no quasi-steady circulation,
0 , on the section at a moment in time, and yet have wake effects from previous times.) In this case, all of
the (noncirculatory) lift arises from the wake term, not the apparent mass term.
11

AA200B - Applied Aerodynamics II

Lecture 13

We also note that we could determine the wake strength by keeping track
of the vorticity shed throughout time. If, in a panel code, we included its
effect on the boundary conditions, we would have only part of Cl2 . The
other part would appear in the computation of pressures using the unsteady
Bernoulli equation. The apparent mass term (Cl1) can be computed more
simply because it is not affected by the wake (by construction). It could,
therefore be evaluated using thin airfoil theory. We can write:
2 d
Cl1 = 2
cV dt

Z
0

c
0(x )dx
2

or explicitly in terms of the local normal velocities on the airfoil:


4 d
Cl1(t) = 2
cV dt

Z cp
x0(c x0)wa(x0, t)dx0
0
12

AA200B - Applied Aerodynamics II

Lecture 13

And we note that


Z cr
0 = 2
0

x0 wa(x0, t) 0
dx
0
c x V

If we first consider the case of pure plunging motion, wa is constant with x


so:
Z cr 0
20
x wa(x0, t) 0
Cl0 =
dx
= 4
0
V c
c

x
V

0
4wa(t)
=
V

Z
0

x0
wa(t)
0
dx
=
2
1 x0
V

This is just as expected: the quasi-steady lift coefficient is just 2 times


the instantaneous angle of attack.
13

AA200B - Applied Aerodynamics II

Lecture 13

The apparent mass term in this case is:


Z cp
4w a(t)
0 (c x0 )dx0
Cl1(t) =
x
2
cV
0
4w a(t)c 1 p 0
cw a(t)
0 )dx0 =
=
x
(1

2
2
V
2V
0
This describes a force given by F = ma where m is the mass of air contained
in a circle of diameter equal to the chord. (Of course the airfoil effects
a region of air much larger than a circle around the chord and does not
produce uniform acceleration of this air mass, but the integral does have a
value with this simple interpretation.)
Z

14

AA200B - Applied Aerodynamics II

Lecture 13

At this point the lift (and similarly for moment, Cp, and drag) must
either be computed numerically, or some simple type of motion prescribed.
Theodorsen [1] assumed simple harmonic motion with:
= 0eit and h = h0eit. where complex amplitudes 0 and h0 can be
used to represent phase shifts.
It is convenient to define a dimensionless frequency, sometimes called
c
the reduced frequency, k = 2U
.

The number of chord lengths traveled in one cycle is:


U
U 2

=
=
c
c
k.
The value of k indicates the importance of unsteadiness in the flow,
with values of k less than 0.1 or 0.2 often indicating that unsteady effects
are not very important.

15

AA200B - Applied Aerodynamics II

Lecture 13

When simple harmonic motion is assumed, the previous integrals may


be evaluated in a simple form. In particular, an expression for the wake
vorticity is derived by differentiating the condition that the total circulation
in the field is 0:
Z U t
(t) +
w (x, t)dx = 0
TE

The process of computing w and integrating the previous expressions for


lift is messy, but was done by Theodorsen without Mathematica. Details
are posted on the course website.

16

AA200B - Applied Aerodynamics II

Lecture 13

The result can be written in terms of Bessel functions, easily evaluated


in MatLab or Excel, and lumped together in Theodorsens lift deficiency
factor, C(k) = F (k) + iG(k), with:
F (k) =

J1(J1 + Y0) + Y1(Y1 J0)


(J1 + Y0)2 + (Y1 J0)2

Y1Y0 + J1J0
G(k) =
(J1 + Y0)2 + (Y1 J0)2

17

AA200B - Applied Aerodynamics II

Lecture 13

The final expression for the lift coefficient of an airfoil that is pitching
and plunging is given below. The incidence of the airfoil is and the
rotation axis is ac/2 behind the half-chord. The vertical position of the
rotation axis is h.
h
ac
1 a c
c

Cl = 2C(k)(
+( )
)+
(h + V )
2
V
4 2 V
2V
2
This can be rewritten, with the rotation axis xc behind the leading edge
(a = 2x 1):

h
(0.75 x)c
c

Cl = 2C(k)(
+
)+
(h + V (x 0.5)c)
2
V
V
2V

18

AA200B - Applied Aerodynamics II

Lecture 13

The apparent mass term is just one of the unsteady effects in this
equation. Although this term is directly related to the unsteady term in
Bernoullis equation, some of the difference between C(k) and 1.0 is also
related to the unsteady term in the Bernoulli equation.

19

AA200B - Applied Aerodynamics II

Lecture 13

The following figure shows the value of C(k).


C(k)

1
0.8

F
G

C = F + iG

0.6
0.4
0.2
0
-0.2
0

0.2

0.4

0.6

0.8

Figure 3. Variation of C(k) with reduced frequency.


Many different approximations to the expression for Cl are often made.
20

AA200B - Applied Aerodynamics II

Lecture 13

These include:
1. Steady (just use the instantaneous incidence as determined from a
snapshot): Cl = 2
2. Quasi-steady (ignore 
unsteady wake and d/dt
 term in unsteady

Bernoulli equation): Cl = 2 Vh + (0.75 x) Vc


3. Unsteady, but ignoring wake effects: Cl = Cl0 + Cl1 As noted below,
this may be worse than quasi-steady.
4. Fully unsteady.
The fully unsteady computation requires either an assumption about
the motion, keeping track of the shed wake, or including it through a
convolution integral.
An analysis of the effect of frequency on Cl1 and Cl2 shows that we
21

AA200B - Applied Aerodynamics II

Lecture 13

should never ignore Cl2 in comparison with the rest of the equation. At low
frequencies, Cl1 is higher order in k (or ) than Cl2. At high frequencies
C(k) goes to 1/2, so Cl2 again cannot be ignored in comparison with the
quasi-steady term, Cl0.

22

AA200B - Applied Aerodynamics II

Lecture 13

Substitution of the expression for C(k) into the equation for Cl shows
how unsteady aerodynamics affects the time history of Cl. The results
depend on the location of the rotation center, the amplitude and phase
of pitching relative to plunging, and the frequency of the motion. In the
simplest case of pure plunging motion ( = 0), the expression for Cl reduces
to:
h
c
Cl = 2C(k)
+
h
2
V 2V
In this case, the unsteady effects appear as an apparent mass effect, a
reduction in magnitude of the first term (more important as the frequency
increases), and an apparent lag in the force due to the negative imaginary
part of C(k).

23

AA200B - Applied Aerodynamics II

Lecture 13

3D Unsteady Aerodynamics
The combination of the usual trailing vorticity from a 3D wing and
the transverse shed vorticity from the unsteady motion leads to vorticity
in various directions in the wake of an oscillating wing. The figure below
shows how this might look for a flapping bird wing, but even a simple rigid
plunging wing has an interesting pattern of vorticity due to these basic
289
Minimum induced power requirements for flapping flight
effects.

FIGURE
2. Top view of bird in flight showing coordinate system and vortex filaments (trailing and
shed vorticity) in unsteady wake. (Harris hawk planform after Tucker 1992).

spanwise circulation varies with time. Nevertheless, as we will show, the Betz criterion
Figure may
4. beUnsteady
3Dmotion
wake
Figure
from Hall
andflight).
Hall [6].
applied to flapping
of wings
(as well as helicopters
in forward
Furthermore, because the resulting theory deals with the far wake and not the details
of the flow about the wing itself, no simplifying assumptions regarding the reduced
frequency or amplitude of flapping motion are required. Therefore, the present theory
is applicable to high-frequency and/or large-amplitude flapping motions.
In $2, we describe the extension of the Betz criterion for M.I.L. propellers to
the case of the flapping motion of wings. In $ 3 , we apply the results of $ 2 to
the case of small-amplitude harmonic flapping motion. We show that the problem
of finding the optimum spanwise circulation distribution can be reduced to a onedimensional integral equation for the unknown circulation. This integral equation is
solved efficiently using numerical quadrature. In $ 4 we describe a three-dimensional

24

AA200B - Applied Aerodynamics II

Lecture 13

Of course, these vortices are not straight, but roll-up under their own
286
K. C. Hall and S. R. Hall
self-induced velocities
to form complex
patterns.

FIGURE
1. (a) Vortex-ring wake. ( b ) Concertina or continuous-vortex wake. (c) Ladder wake.
Sketches after Pennycuick (1988).

was seeded5.
withRolled-up
neutrally buoyant soap
bubbles
filled with helium. from
Trained birds
and
Figure
wake
geometries
birds.
bats then flew through the seeded air, and the resulting flow structure in the wake
of the animals was captured on film using stroboscopic photography. Rayner (1991)
surveyed the available experimental data, and found that the structure of the wakes
behind birds and bats falls into one of two distinct patterns. In slow flight, the wake
appears to be composed of a series of vortex rings, one ring for each downstroke
of the wings (see figure 1). In fast flight, the wake is composed of two undulating
vortices which trail behind the animal. Furthermore, no transverse vorticity is
observed. Quoting Rayner, The absence of transverse vortices is not surprising, since
the interaction of transverse vortices with the vortex on the wing can dramatically
increase induced drag.
Based on these experimental observations, Rayner (1991, 1993) has proposed that
birds use two distinct gaits: the vortex-ring gait and the continuous-wake gait. The
vortex-ring gait is used in slow flight. According to Rayner, during the upstroke,
the wing is flexed so that the span of the wing is reduced. Furthermore, the wing is
aerodynamically inactive (little or no circulation is generated along the span of the
wing). During the downstroke, the wing is fully extended and nearly flat, and the wing
circulation along the span of the wing generates thrust and lift. The continuous-wake
gait, on the other hand, is used in fast flight. The wing is aerodynamically active
during both the downstroke and the upstroke with constant total circulation. During
the downstroke, the wing is fully extended. During the upstroke, however, the wing is
flexed or swept slightly to shorten the span of the wing, reducing the instantaneous
lift while maintaining constant circulation.
Rayner (1991) has asserted that all vertebrates use one of these two gaits in forward

These wakes and those shed from the tails of swimming animals have
been studied extensively, but often more significance is attributed to the
complex geometry than may be necessary to understand some of the basic
flow physics. In particular, just as one does not need to compute the details
of the 3D wake roll-up process to very accurately compute induced drag,
much of the important unsteady aerodynamics may be understood with
much simpler models of the wake motion. This is suggested by Hall and
25

AA200B - Applied Aerodynamics II

Lecture 13

Hall [6] in their analysis of flapping flight, and in the case of small amplitude
unsteady motion, the analysis is even simpler.
Just as we argued in the case of induced drag, the velocity induced by
the wake becomes less important as it is convected farther downstream.
And since the near wake geometry is simpler than the far wake, it is often
a very good approximation to ignore the self-induced wake motion and
compute the pressures on the lifting surface itself. Just as in the case of
induced drag we must be careful if we compute forces based on far-field
velocities and simplified wakes, while near-field calculations are less sensitive
to the assumed wake geometry.
In fact, many interesting cases can be studied with a simple vortex panel
code in which the wake propagates downstream in a simple lattice. Trailing
vorticity trails aft in the freestream direction while transverse vorticity is
shed in proportion to the change in circulation on the wing at the specified
spanwise station.
26

AA200B - Applied Aerodynamics II

Lecture 13

This is the standard approach used by codes such as NASTRAN


for aeroelastic flutter analysis.
Of course this is a very simple
approximation, ignoring, not just the effects of wake deformation, but
nonlinear compressibility effects, separation and other viscous effects. A
discussion of certain of these phenomena, of particular interest in helicopter
aerodynamics, is given in Ref. [7].
Even simpler approximations are also common, including the use of
unsteady 2D results in a kind of modified strip theory in which 3D steady
results are used in place of the 2D steady lift curve slopes in the previouslydescribed derivation. This is less commonly used these days, as true 3D
codes are more computationally tractable.

27

AA200B - Applied Aerodynamics II

Lecture 13

References
[1] Theodorsen T., General Theory of Aerodynamic Instability and the
Mechanism of Flutter,NACA Report No. 496, 1935.
[2] von Karman, T., and Sears, W. R., Airfoil Theory for Non-Uniform
Motion, Journal of the Aeronautical Sciences, Vol. 5, No. 10, August,
1938, pp. 378-390.
[3] Sears, W. R., A Systematic Presentation of the Theory of Thin Airfoils
in Non-Uniform Motion, PhD. Thesis, California Institute of Technology,
1938.
[4] Garrick, I. E., A Review of Unsteady Aerodynamics of Potential
Flows, Applied Mechanics Review, Vol. 5, No. 3, March 1952, pp. 89-91.
[5] Bisplinghoff, R., Ashley, H., Halfman, R., Aeroelasticity, Addison
Wesley, 1955, pp. 288-293.
28

AA200B - Applied Aerodynamics II

Lecture 13

[6] 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.
[7] W. J. McCroskey, W.J., Unsteady Airfoils, Ann. Rev. Fluid Mech.
1982. 14:285-311

29

You might also like