Full Elementary Aerodynamics Course by MIT

Sensitivity Analysis
Often in engineering analysis, we are not only interested in predicting the

performance of a vehicle, product, etc, but we are also concerned with the
sensitivity of the predicted performance to changes in the design and/or errors in
the analysis. The quantification of the sensitivity to these sources of variability is
called sensitivity analysis.
Lets make this more concrete using an example. Suppose we are interested in
predicting the take off distance of an aircraft. From Andersons, Intro. To Flight,
an estimate for take-off distance is given by Eqn (6.104):
T
=
max
2
44 . 1
L
LO
SC g
W
S
To make the example concrete, lets consider the jet aircraft CJ-1 described by
Anderson. In that case, the conditions were:
0 . 1
19815
7300
/ 2 . 32
@ / 002377 . 0
318
max
2
3
2
=
=
=
=
=
=
L
C
lbs W
lbs T
s ft g
level sea ft slugs
ft S
Thus, under these nominal conditions:

( )
( )( )( )( )( ) 7300 0 . 1 318 002377 . 0 2 . 32
19815 44 . 1
2
=
LO
S
ft S
LO
3182 = @ nominal conditions
Suppose that we were not confident of the value and suspected that it
might be 0.1 from nominal.
max
L
C
That is,
0.9 1.1
max
L
C
Also, lets suppose that the weight of the aircraft may need to be increased for a
higher load. Specifically, lets consider a 10% weight increase:
1) Linear sensitivity analysis
2) Nonlinear sensitivity analysis (i.e. re-evaluation)
They both have their own advantage and disadvantages. The choice is often
made based on the problem and the tools available. Well look at both options.
Linear Sensitivity Analysis
Linear sensitivity based on Taylor series approximations. Suppose we were
interested in the variation of with w &
LO
S
max
L
C
Then:
W
W
S
C
C
S
C S
W W C C S
LO
L
L
LO
L LO
L L LO
A
c
c
+ A
c
c
+
~ A + A
max
max
max
max max
) (
) , (
That is, the change in is:
LO
S
W
W
S
C
C
S
S
LO
L
L
LO
LO
A
c
c
+ A
c
c
A
max
max
The derivatives
W
S
C
S
LO
L
LO
c
c
c
c
&
max
are the linear sensitivities of to changes in
& , respectively.
LO
S
max
L
C j W
Returning to the example:
max
max
max
2
2
44 . 1
L
LO
L
L
LO
C
S
T SC g
W
C
S
=
=
c
c
W
S
T SC g
W
W
S
LO
L
LO
2
44 . 1 2
max
=
=
c
c
These can often be more information by looking at percent or fractional changes:

W
W
W
S
S
W
C
C
C
S
S
C
W
W
S
S
C
C
S
S S
S
LO
LO L
L
L
LO
LO
L
LO
LO
L
L
LO
LO LO
LO
A
c
c
+
A
c
c
=
A
c
c
A
c
c
~
A
+
max
max
max
max
max
max
1 1
Fractional sensitivities
16.100 2
For this problem:
W
W
S
S
W
S
S
W
C
C
S
S
C
S
S
C
LO
LO LO
LO
L
L
LO
LO
L
L
LO
L
A
~
A
=
c
c
A
~
A
=
c
c
0 . 2 0 . 2
1 0 . 1
max
max
max
max max
C
C
S
S
L
L
LO
LO
A
~
A
1
max
max
W
W
S
S
LO
LO
A
~
A
0 . 2
Thus, a small fraction change in will have an equal but opposite effect on
the take-off distance.
max
L
C
The weight change will result in a charge of which is twice as large and in
the same direction.
LO
S
Thus, is more sensitive to W than changes at least according to linear
analysis.
LO
S
max
L
C
Example
We were interested in varying 0.1 which according to linear analysis
would produce variation in take-off distance:
max
L
C
LO
S 1 . 0
ft S S C
ft S S C
LO LO L
LO LO L
318 1 . 0 1 . 1
318 1 . 0 9 . 0
max
max
= + ~ A =
+ = + ~ A =
For a weight increase of 10% we find
( )( ) ( )
ft
S S lb W
LO LO
636
1 . 0 2 19815 1 . 1
+ ~
+ ~ A =
Nonlinear Sensitivity Analysis
For a nonlinear analysis, we simply re-evaluate the take off distance at the
desired condition (including the perturbations). So, to assess the impact of the
variations we find:
max
L
C
ft C S
C S
L LO
L LO
3535 ) 9 . 0 (
) 7300 )( 9 . 0 )( 318 )( 002377 . 0 )( 2 . 32 (
) 19815 ( 44 . 1
) 9 . 0 (
max
max
2
= =
= =
ft C S
L LO
6 . 353 ) 1 . 0 (
max
+ = = A A
which agrees well with linear result
ft S S C
ft S S C
LO LO L
LO LO L
318 1 . 0 1 . 1
318 1 . 0 9 . 0
max
max
= + ~ A =
+ = + ~ A =
16.100 3
Similarly,
( ) ft C S
L LO
2892 1 . 1
max
= =
( ) ft C S
L LO
290 1 . 0
max
= + = A
Finally, a 10% W increase to 21796lbs gives:
( )
( ) ft W W S
ft lb W S
LO
LO
668 1 . 0
3850 21796
+ = + = A A
= =
16.100 4
Kinematics of a Fluid Element

Convection: u
K

Rotation rate:
1 1
2 2
i j k
u
x y z
u v w
vorticity

= =

=
K K K
K
K
K

1
2
w v u w v u
i j k
y z z x x y

( (
(
= + +
`
( (
(

)
K K K

Normal strain rates:

x
xx
x
dL
u
dt
L x

= =

y
yy
dL
v
dt z

= =

z
ZZ
dL w
dt z

= =

Shear strain rates:

Angle between edge
1 1
along and along 2 2
j
i
ij ji
j i
u
u d
i j x x dt

| |
| |
= + = =
|
|
|

\ .
\ .

Strain rate tensor:

xx xy xz
yx yy yz
zx zy zz

(
(
(
(

Convection Rotation Compression/Dilation
(Normal strains)
Shear Strain
L
x
L
y
Kinematics of a Fluid Element
16.100 2002 2
Divergence

( )
/
d Volume u v w
u Volume
x y z dt

= + + =

K

Substantial or Total Derivative

u
D
u v w
Dt t x y z

= + + +

K

=rate of change (derivative) as element move through space

Cylindrical Coordinates

x x r r
u u e u e u e

= + +
K K K K

1
x r r
xx rr
u u u u
x r r r
=

= = +

1 1
2
r
r
u u
r
r r r
( | |
= +
|
(

\ .

1
2
r x
rx
u u
x r

(
= +
(

1 1
2
x
x
u u
r x

(
= +
(

( )
1 1
r
x
u
u ru e
r r r

(
=
(

K K 1
x
r
u u
e
r x

(
+
(

K
r x
u u
e
x r

(
+
(

K

( ) 1 1
r x
ru u u
u
x r r r

= + +

K

Stress-Strain Relationship for a Newtonian Fluid
First, the notation for the viscous stresses are:
x
y
z

xx

xy

xz

zx

zy

zz

yx

yy

yz
ij
= stress acting on the fluid element with a face whose normal is in + x
i
direction and the stress is in + x
j
direction.
Common assumption is that the net moment created by the viscous stresses are
zero.
ij
=
ji
Lets look at this in 2-D:
dx
dy
x
y

xx

xy

yx

yy

yy

yx

xy

xx
M
z
|
dx
|
dy
|
xy
2
.
|
dy
yx |
dx
\
2
.
M = Net moment about center=
\
z
|
dx
| |
+
xy |
dy
yx
dy
|
dx
\
2
. \
2
.
|
M = (
xy
)
dxdy
z yx
2
Thus, for M = 0,
xy
=
z yx
Assumptions for Newtonian fluid stress-strain:
1)
ij
is at most a linear function of
ij
.
2) The fluid is isotropic, thus its properties are independent of direction
stress-strain relationship cannot depend on choice of coordinate axes.
3) When the strain rates are zero, the viscous stresses must be zero.
16.100 2002 2
To complete the derivation, we consider the stress-strain relationship in the
principal strain axes (i.e. where
ij
= 0 i for j ).
Thus,

11
= C
11
11
+ C
12
22
+ C
13
33
22
= C
21
+ C
22
22
+ C
23
33

11
33
= C
31
+ C
32
22
+ C
33
33 11
But, to maintain an isotropic relationship:
C
C
11
= C
22
= C
33
12
= C
21
= C
31
= C = C
23
= C
32 13
which leaves only two unknown coefficients.
We define these two coefficients by:
(
= 2 + +
22
+ )
33 ii ii 11
K
u
dynamic viscosity coefficient
2
nd
or bulk viscosity coefficient
For general axes (i.e. not in principal axes):
ij
= 2
ij
+ +
22
+ )
ij
(
11 33
or,
ij
=

x
u
i
j
+

u
x
i
j
|
|
|
+ u
K
ij
\ .
where
0 i j
.
ij
=

1 i = j
16.100 2002 3
Coordination Transformations for Strain & Stress Rates
To keep the presentation as simple as possible, we will look at purely two-dimensional
stress-strain rates. Given an original coordinate system ( x, y ) and a rotated system
( y x, ) as shown below:
x
x'
y
y'

Recall that the strain rates in the x-y coordinate system are:
u 1 | u v| v
= = =
xx xy yy
x 2
\
y
+
x
.
|
|
y
Or, in index notation:
=
ij
1
2

x
u
i
j
+

u
x
i
j
|
|
|
\ .
Also, we note that the unit vectors for the rotated axes are:
K
i
K

= cosi
K
+ sin j
K
K K
j= sini+ cos j
Thus, the location of a point in ( y x, ) is:
(
cos sin
( (
x x
=
( ( (
y y

sin cos

Similarly, the velocity components are related by:
(
cos sin
( (
u u
=
( ( (
v v

sin cos

For differential changes, we also have
(
cos sin
( (
dx dx
=
( ( (
dy dy

sin cos


16.100 2002 2

Thus, defining T as the rotation matrix, we note that:

cos sin
sin cos
x x
x y
T
y y
x y

(
(
(
(
= =
(

(

(

Inverting this:
1
1
cos sin cos sin
sin cos sin cos

x x
x y
T
y y
x y

(
(
( (
(
= = =
( (

(

(

Thus to find
x
u
in terms of v u, and their derivatives:

cos sin

u u x u y u u
x x x y x x y

= + = +

Then, substituting sin cos v u u + = :

2 2
2 2

2
cos sin ( cos sin )
cos cos sin cos sin sin

cos 2cos sin sin
sin cos ( sin cos )
sin sin cos sin cos

xx xx xy yy
u
u v
x x y
u v u v
x x y y
v
u v
y x y
u v
x x

(
= + +
(

= + + +

= + +
(
= + +
(

=

2
2 2

cos
sin 2sin cos cos
1 1
sin cos ( cos sin ) cos sin ( sin cos )
2 2
yy xx xy yy
u v
y y
u v
u v v
y x x y x y
+

= + +

| | ( (
+ = + + + + +
`
| ( (

\ . )
2 2
sin cos ( ) (cos sin )

xy yy xx xy
= +


16.100 2002 3
If y x are the principal strain directions, then
y y y x x x xy
& , 0 and = are

)
2 2

2 2

( cos sin
cos sin
sin cos
xx yy y x
yy xx y y
yy xx x x

=
+ =
+ =
if 0
xy
=

The next step is to relate the stresses in ( ) ( ) y x to y x , , . Consider a differential surface
with y normal:

The resultant stress is given as the vector
K
and the force on the surface is . ds
K

Decomposing the stress vector into the coordinate axes gives:

( )
( ) ( )

yx yy
xx yx yy xy
dx i j dx
dy dx i dx dy j

= +
= + +
K K
K
K K

Note that:

cos
sin

cos sin

sin cos
dx dx
dy dx
i i j
j i j

=
=
=
= +
K K
K
K K
K
dx
dy
yx
yy
xx
xy
yy
yx
d
x
x
x'
y
y'

16.100 2002 4

Thus, the second line becomes:

( sin cos )(cos sin )

( cos sin )(sin cos ) ( )
xx yx
yy xy yx yy
i j dx
i j dx i j dx

+
+ + = +
K K
K K K K

So collecting all the j i

&
terms (and enforcing

yx xy
=
) gives:

2 2

2 2

( )sin cos (cos sin )
sin cos 2 sin cos
yx yy xx yx
yy xx yy yx

= +
= +

For the principal strain axes,

( )
( )
0
2
2
=
+ + =
+ + =
xy
yy xx yy yy
yy xx xx xx

Plugging this into
y y x y
and gives

( )
( ) ( )

2 2
2 sin cos
2 sin cos
xy
xx yy yy
yx yy xx
yy xx yy xx yy

+
=
= + + +

Thus, we arrive at the known result:

( )

2
2
yx xy
yy yy xx yy

=
= + +

A similar derivation would give:

( )

2
xx xx xx yy
= + +
Compressible
Vi
Compressible
iscid
Incompressible
Vi
Compressible
Compressible
Potential
Incompressible
Invicid
Incompressible
Incompressible
Potential
c
scous
Inv scous Boundary Layer
Boundary Layer
Small Disturbance
Comp. Potential
Small Disturbance
Inc. Potential
Subsonic
Transoni
Supersonic
Note: Other effects could have
been included in this
diagram. e.g. steady vs.
unsteady flows
*Each arrow indicates an additional assumption has been applied
*More than one path to a given approximation
Compressible Viscous Equations

Also known as the compressible Navier-Stokes equations:

Mass: ( ) 0 =
j
j
v
x t

Momentum:
( )
( ) ( )
, 1, 2, 3
i
i j ij
j i j
p
i
t x x x

+ = + =

Energy:
2 2
1 1
( ) ( )
2 2
j
j
e v e v v
t x

(
+ + +
(

( ) ( )
( )
j ij i j
j j j
p q
x x x

= + +

( ) , , ( )
j
i k
ij
ij
j i k
j
x x x
q k e e p state relationship ideal gas
x

| |

= + +
|
|

\ .
= =

Incompressible Viscous Equations

In this case, we assume . const =

Mass: 0 =
j
j
x

Momentum: ( )
(
(
|
|
.
|
\
|
i
j
j
i
j i
j i
j
i
x x x x
p
x t

Usually, ( ) = . Often, when temperature variations are small, =const. is assumed.

0
j
i j
j j
i i
j j i j j j i
x x
i
j j
x x x x x x x
x x
| |
= |
|

\ .
(
(
(
( | |

(
+ = +
( |
|
(
(
\ .

(
(

16.100 2
Usual form of momentum for incompressible flow:

j j
i
i
j i
j
i
x x
v
x
p
v v
x t
v

+
) (

In this case, the energy equation is not needed to find p v
i
& .

Incompressible Inviscid Equations

In this case we assume that the effects of viscous stresses are small compared to
acceleration and pressure forces:

Mass: 0 =
j
j
x

Momentum:
( )
i
i j
j i
p
t x x

+ =

These are known as the incompressible Euler equations.

Incompressible Potential Flow

In potential flow, we assume the flow is irrotational (i.e. 0 V =
K
). This allows the
velocity to be written as the gradient of a scalar potential:

,
i
i
x
= Potential

Mass: 0
j
j j
x x

=

Also written out as:
0
2
2
2
2
2
2
=
z y x

or

2 2
0, Laplacian =

This is a single equation for a single unknown . It is the same for steady and unsteady
flows.

What happens to momentum?

16.100 1
Compressible Equations

Conservation of mass

( )
( )
0
0
0
0,
volume surface
d
dv u ndS
dt
u
t
D
u
Dt
u incompressible

+ =
+ =
+ =
=

K K
K
K
K

Conservation of Momentum

( ) ( ) ( )
( )
( )
Net viscous force in
:
volume surface surface surface
xx yx zx xy yy zy xz yz zz
xx xx zx
d
udv uu ndS pn idS idS
dt
Note
ds dx dy dz i dx dy dz j dx dy dz k
u
p
uu
t x x y z

+ = +
= + + + + + + + +

+ = + + +

K K
K K K K
K K K
K
K
x
yx
xx zx
Du p
Dt x x y z

= + + +

Recall:
2
( ),
3
j
i
ij ij
j i
u
u
u
x x
T

= + +

= =
K

Incompressible Equations in Cartsian Coordinates
2
2
2
0, i.e. 0
1
1
1
u v w
u
x y z
Du p
u
Dt x
Dv p
v
Dt y
Dw p
w
Dt z

+ + = =

= +
= +
= +
K

Compressible Equations
16.100 2002 2

Where

2 2 2
2
2 2 2
, kinematic viscosity
D
u v w
Dt t x y z
x y z

= + + +

= + +

Incompressible Equations for Cylindrical Coordinates

( ) ( ) ( )
( )
2 2
2 2
2
2 2
2
1 1
0
1 1 2
1 2
1
r x
r r
r
r r
x
x
ru u u
r r r x
Du p u u
u u
Dt r r r r
Du u u p u u
u
Dt r r r r
Du p
u
Dt x

+ + =

= +

+ = + +

= +

Where

2 2
2
2 2 2
1 1
, kinematic viscosity
r z
D u
u u
Dt t r r z
r
r r r r z

= + + +

= + +

Equations of Aircraft Motion

Force Diagram Conventions
Definitions

V flight speed
angle between horizontal & flight path
angle of attack (angle between flight path and chord line)
W aircraft weight
L lift, force normal to flight path generated by air acting on aircraft
D drag, force along flight path generated by air acting on aircraft
M pitching moment
T propulsive force supplied by aircraft engine/propeller
T
angle between thrust and flight path

To derive the equations of motion, we apply
= a m F
K
K
(1)

Note: we will not be including the potential for a yaw force.

Applying (1) in flight path direction:
dV
F ma m
dt
= =
& &

and examining the force diagram
cos sin
T
F T D W =
&

cos sin
T
dV
T D W m
dt
= (2)

C
h
o
rd
lin
e
Flight path
V
M
W
D
T
Horizontal

16.100 2002 2
Now applying (1) in direction to flight path

2
c
V
F ma m
r

= =

where
c
r radius of curvature of flight path

sin cos
T
F L T W
= +

2
sin cos
T
c
V
L T W m
r
+ = (3)

Equations (2) & (3) give the equations of motion for an aircraft (neglecting yawing
motions) and are quite general. One important specific case of these equations
is level, steady flight with the thrust aligned w/ the flight path.

0, , 0, 0
c T
dV
r
dt
= = =

Level, steady flight
T D
L W
=
=

Moment definitions
The pitching moment must be defined relative to a specific location. The two
typical locations are:
leading edge

1
,
4
c quarter of mean chord

16.100 2002 3

Force & Moment Coefficients
Typically, aerodynamicists use non-dimensional force & Moment coefficients.

2
2
1
2
3D Drag/Lift coefficients
1
2
L
D
L
C
V S
D
C
V S

where
is freestream density
V
is freestream velocity (flight speed)

S is a reference area (problem dependent)

2
1
Freestream dynamic pressure
2
q V

The moment coefficient requires another length scale:

2
1
2
M
ref
M
C
V S

A

ref
A reference length scale (problem dependent)

For 2-D problems, such as an airfoil, the forces are actually forces/length. So, for
example

3D force 2D force/length
L L
D D
Similarly, M M . The non-dimensional coefficients for 2-D are defined:


16.100 2002 4
2
2
2 2
1
2
1
2
1
2
l
ref
d
ref
m
ref
L
C
V c
D
C
V c
M
C
V c

where
ref
c is a reference length such as the chord of an airfoil.

Forces on Airfoils

The forces & moments on airfoils are normalized by the chord length. So,

2
,
1
2
l
L
C
V c

2 2 2
,
1 1
2 2
d m
D M
C C
V c V c

Force coefficients data is generally plotted in 2 forms:

Lift curve

C
l
max
C
l

16.100 2002 5

Drag polar

Forces on Wings

) ( y c chord distribution
b wing span
S planform area =
2
2
b
b
cdy
A aspect ratio
S
b
2

We can think of the 3-D or total lift on the wing as being the sum (i.e. integral) of
the 2-D lift acting on the wing.

( )
2
2
b
b
L L y dy
=

where ( ) L y = lift distribution

C
l
C
d
C
d
min
C
l
C
d
C
d
min
x
y
b
c(y)
Wing planform

16.100 2002 6
The average 2-D lift on the wing L can be defined:

2
2
1
b
b
L
L L dy
b b

Plugging that into
L
C :

2
2
2 2 2
1 1 1
2 2 2
b
b
L
L dy
L L b
C
V S V S V S
= =

But, the average chord or mean chord can be defined as:

2
2
1
b
b
S
c cdy
b b

2 2
1 1
2 2
L
L L
C
V S V c

=

In other words, we can think of the 3-D lift coefficient as the mean value of the 2-
D lift coefficient on the wing. The same is true for drag and moment:

2 2
2 2 2
1 1
2 2
1 1
2 2
D
M
ref
D D
C
V S V c
M M
C
V S V c

=
A

where
ref
c = A is used.

A Closer Look at Drag

The drag coefficient can be broken into 2 parts:

N
N
2
,
L
D D e
parasite
induced
drag
drag
C
C C
eA
= +


16.100 2002 7

where e = span efficiency factor (more on this when we get to lifting line).

The parasite drag contains everything except for induced drag including:
skin friction drag
wave drag
pressure drag (due to separation)

It is a function of , thus, we can also think of
, D e
C as being a function of
L
C .
The parasitic drag can be well-approximated by:

0
2
, D e D L
C C rC = +

where
0
D
C drag at 0
L
C = , r = empirically determined constant.

0
2
1
D D L
C C r C
eA

= + +

Finally, we can re-define e to include r :

0
2
1
D D L
C C C
eA
= +

where
1
e
e
r eA
+
.

This re-defined e is known as the Oswald efficiency factor.

We will refer to
0
D
C as the parasite drag coefficient from now on.

16.100 2002 1
Brequet Range Equation

f
i
t
t
Range Vdt =

In level flight at const speed:

L W = , Lift = Weight
T D = , Thrust = Drag

The aircraft weight changes during flight due to use of fuel. Relate weight
change to time change:

dW
dW dt
dt
fuel weight
dt
time
fuel weight T
dt
time T
=
=
=

The quantity:

1 fuel weight
time T

is known as the specific fuel consumption or sfc. It has units of:

sfc units =
( ) fuel
lb force
or
lb hr force time

1
f
i
W
W
dW sfc T dt
V
Range dW
sfc T
=
=
i

But since T D = and L W = we have:

f
i
W
W
V L dW
Range
sfc D W
=

Range
V
t=t
i
W=W
i
t=t
f
W=W
f

This is the general form of the range equation. The Breguet range equation is
found by assuming that
D
L
sfc
V
is constant for the entire flight. In that case:
f f
i i
W W
W W
V L dW V L dW
Range
sfc D W sfc D W
= =

Breguet range equation:

log
i
f
V L W
Range
sfc D W
=
Aerodynamic Center
1

Suppose we have the forces and moments specified about some reference
location for the aircraft, and we want to restate them about some new origin.
ref
M = Pitching moment about
ref
x
new
M = Pitching moment about
new
x
ref
x = Original reference location
new
x = New origin
N = Normal force L for small
= A Axial force D for small

Assuming there is no change in the z location of the two points:

( )
ref new ref new
M x x L M = +

Or, in coefficient form:

N
. .
ref new
new ref
m L M
mean
a c
x x
C C C
c

= +

The Aerodynamic Center is defined as that location
ac
x about which the pitching
moment doesnt change with angle of attack.

How do we find it?

1
Anderson 1.6 & 4.3

x
z
x
ref
x
new
M
new
M
ref
A A
N
~
L ~ N
~
L ~
Aerodynamic Center

16.100 2002 2

Let
new ac
x x =

Using above:
( )
ref ac
ac ref
M L M
x x
C C C
c
= +

Differential with respect to :

ref
ac
M M ac ref
L
C C x x
C
c

= +

By definition:

0
ac
M
C

Solving for the above

,
ref
ref
M
ac ref
L
M
ac ref ref
L
C
x x
C
or
c
C
x x x
c c C

Example:

Consider our AVL calculations for the F-16C
0
ref
x = - Moment given about LE
Compute
ref
M
L
C
C
for small range of angle of attack by numerical

differences. I picked
0 0
3 to 3 = = .
This gave 2.89
ac
x
c
.
Plotting
M
C vs about
ac
x
c
shows 0
M
C
.

Note that according to the AVL predictions, not only is 0 @ 2.89
M
ac
C
x
, but
also that 0
M
C = . The location about which 0
M
C = is called the center of
pressure.

Aerodynamic Center

16.100 2002 3
Center of pressure is that location where the resultant forces act and about which
the aerodynamic moment is zero.

Changing new to cp and ref to NOSE to correspond to AVL:

NOSE
NOSE cp
NOSE
cp
M L M
M cp
L
x x
C C C
c
C x
c C

=

+
=

So if:
NOSE NOSE
L
M M
L
C C
C C
,
this will be true. This means that

0 at 0
M L
C C = .

C
m

v
s

A
l
p
h
a

f
o
r

F
1
6
C

f
r
o
m

A
V
L

(
M
=
0
)
-
3
.
5
-
3
-
2
.
5
-
2
-
1
.
5
-
1
-
0
.
5 0
0
.
5 1
1
.
5
-
1
0
-
5
0
5
1
0
1
5
2
0
A
l
p
h
a
C m
X
r
e
f

=

0
X
r
e
f

=

X
a
c

=

2
.
8
9
W
i
n
d

T
u
n
n
e
l

T
e
s
t

A
e
r
o
d
y
n
a
m
i
c

C
e
n
t
e
r

C
h
a
r
a
c
t
e
r
i
s
t
i
c
s

f
o
r

W
a
s
h
o
u
t

a
n
d

R
i
g
i
d

W
i
n
g
s

(
A
l
t
i
t
u
d
e

=

1
0
,
0
0
0

f
t
.
)
1
5
2
0
2
5
3
0
3
5
4
0
4
5
5
0
0
.
5
0
.
6
0
.
7
0
.
8
0
.
9
1
1
.
1
1
.
2
1
.
3
M
a
c
h

N
u
m
b
e
r
A . C . ( % M A C )
R
i
g
i
d

W
i
n
g
W
a
s
h
o
u
t

W
i
n
g
Notes on CQ #1

( )
( )
1
2
1 1
2 2
2
Re
Re
( )
d
d
d
D q cC
D q cC
Drag scales withV C V

=
=

=

=
= =
=
V
C V
V
C V
q V p q
V p q
2
2
1
1
1
2
2
2
2
1
1
Re
Re
4
2
1
2
1

N
N
1
2
2
1
2
1
2

1 13 3
7 7 2
(Re)
Fromdata, Re Laminar flow behavior
Re
, . . Turbulent Re ,
d
d
V
V
d
D q cC
C
D q c
D V c f C D V
D withV
=

Note dependence on chord
6 1
7 2
, c.f. Turbulent D c D c

Drag Polar

N N
or
L l
ref ref
usually usually
wing chord
planform
L L
C C
q S q

A

C
L
C
D
C
D
min
C
D
0
0
C
L
min
increasing
stall
Notes on CQ
16.100 2
For many aircraft,
( )
min min
2
min D D L L
C C k C C +

Also, since
min min
& 0
o
D D L
C C C
2
O
D D L
C C kC +

The first option will be slightly more accurate, but both are reasonable
approximations.

Notes on CQ #2

(1) First, we note that &
O
D
C k will almost certainly depend on the Reynolds
number. But, this dependence is probably weak since the b.l. flow will be
turbulent. So, we assume &
O
D
C k remain constant to good approximation.
Also important is that for a general aviation aircraft, we expect no wave drag
since the flight is subsonic.

(2)
( )
0
2 2
2
2 2
2
2 2
2
1
2
1 1
1
2 2
2
1 1
1
2
2
O
O
O
L
D L
D
D
D
D
D V S C kC
W
V SC k V S
V S
k
D SC V W
V
S
= +

= +

= +

So we see that
2
0 2
1
&
L
D V D
V

(3) Often at cruise,
0
C C
L
D D for prop-aircraft.
0 0
2
C C C
C L
D D D D = + =

At approach:
0 0 0
1 1
4 4
4 4
C C C
A
D D D D = + =
Note:
A C
D D >
Quick Visit to Bernoulli Land

Although we have seen the Bernoulli equation and seen it derived before, this
next note shows its derivation for an uncompressible & inviscid flow. The
derivation follows that of Kuethe &Chow most closely (I like it better than
Anderson).
1

Start from inviscid, incompressible momentum equation

1 u
u u p
t
+ =

There is a vector calculus identity:

( )
2
,
1
2
vorticity
u u u u u
| |
=
|
\ .

2 1 1
2
u
u p u
t

| |
+ + =
|
\ .

From here, we can make the final re-arrangement:

2 1
2
u
p u u
t

| |
+ =
|

\ .

Two common applications:
1. Steady irrotational flow
0 0
Irrotational
Steady
u
t

= =

2
2
1
0
2
1
.
2
p u
p u const for entire flow
| |
+ =
|
\ .
+ =

1
Kuethe and Chow, 5
th
Ed. Sec 3.3-3.5
16.100 2002 2

2. Steady but rotational flow
0 0
Rotational
Steady
u
t

2 1
2
p u u
| |
+ =
|
\ .

This is a vector equation. If we dot product this into the streamwise
direction:

u
u
s

streamwise direction
( )
( )
2
0,
2
2
1
2
1
0
2
1
.
2
u u
s p u s u
d
p u
ds
p u const along streamline
=
| |
+ =
|
\ .
| |
+ =
|
\ .
+ =

Vortex Panel Methods
2

Step#1: Replace airfoil surface with panels

Step #2: Distribute singularities on each panel with unknown strengths

In our case we will use vortices distributed such that their strength varies linearly
from node to node:

Recall a point vortex at the origin is:

1
tan
2 2
y
x

| |
= =
|
\ .

2
Kuethe and Chow, 5
th
Ed. Sec. 5.10

1 2
3 4
m
m+1
i-1
i
i+1
Original airfoil m-panels (m+1 nodes)
16.100 2002 3
A point vortex at y x , is:

1
tan
2
y y
x x

| |
=
|
\ .

Next, consider an arbitrary panel:

At any
j
s , we will place a vortex with strength
( )
j
s ds :
( )
1
( )
, tan
2
j j
j
s ds y y
d x y
x x

| |
=
|
|
\ .

where
( )
1
1
( )
j
j j j j
j
j
j j j j
j
s
x x x x
S
s
y y y y
S
+
+
+
+

Thus, the potential at any ( , ) x y due to the entire panel j is:

( )
( )
1
0
, tan
2
j
S
j j
j
j
s y y
x y ds
x x

| |

|
|
\ .

We will assume linear varying on each panel:
( ) ( )
1
j
j j j j
j
s
s
S

+
= +

s
j
S
j
(s
j
)
ds
j
Vortex of strength
(s
j
)ds
j+1
j
(x
j
,y
j
)
(x
j+1
,y
j+1
)
expanded
view
Midpoints will
be (x
j
,y
j
)
16.100 2002 4
With this type of panel, we have m+1 unknowns =
1 , , 1 ... 3 , 2 , 1 + m m m
, so we
need m+1 equations.

Step#3: Enforce Flow Tangency at Panel Midpoints

The next step is to enforce some approximation of the boundary conditions at the
airfoil surface. To do this, we will enforce flow tangency at the midpoint of each
panel.

Panel method lingo: control point is a location where o n u =

is enforced.

To do this, we need to find the potential and the velocity at each control point.

The potential has the following form:
#

panels
freestream individual panel
potential potential
| | | |
= +
| |
\ . \ .

Suppose freestream has angle :

( ) ( )
( )
1
1
0
, cos sin tan

2
j
S
m
j j
j j
s y y
x y V x y ds
x x
=
| |
= +
|
|
\ .

The required boundary condition is
( )
,
0 1
i i
i
x y for all i m
n

= =

So, lets carry this out a little further:

( ) ( )
1
,
0
,
component of freestream normal
to surface of panel i
normal velocity due to panel j
at control poi
( )
cos sin tan
2
j
i i
S
j j
i i i
j i j
x y
s y y
x y V i j n ds
n n x x

(
| |

= +
( | `
|

(
\ .
)
0
nt of panel i
0
m
j=
=

And recall ( ) ( )
j
j
j j j j
S
s
s + =
+1
.
We can re-write these integrals in a compact notation:

( )
1
1
1 2 1
0
,
Influence of panel j due to
node j on control point of panel i
tan
2
j
ij ij
i i
n
ij
S
j j
n j n j
i j
x y
C
s y y
ds C C
n x x

+
=

(
| |

= +
( | `
|

(
\ .

)

i.e.
1
ij
n j
C =normal velocity from panel j due to node j on control point of panel i .
i i
y x ,
16.100 2002 5
2
ij
n
C =
Influence of panel j due to node 1 j + on control point at panel i

Total normal velocity at control point of panel i due to panel j
1 2 1
ij ij
n j n j
C C
+
= +
So, lets look at the control point normal velocity

So, for panel i , flow tangency looks like:

( ) ( )
1 2 1
1
cos sin , for all 1
ij ij
i
m
n j n j i
j
V n
C C V i j n i m
+
=
+ = + =

We can write this as a set of m equations for m+1 unknowns.

Question: What can we do for one more equation?

Step#4: Apply Kutta condition

We need to relate Kutta condition to the unknown vortex strengths
j
. To do this,
consider a portion of a vortex panel.

ds
16.100 2002 6
Put a contour about differential element ds

Recall: | |
( ) ( )
2 2 1 1
1 2 1 2
u ds
ds V dn U ds V dn U ds
V V dn U U ds
=
= +
=

Now let & 0 dn ds :
( )
1 2
2 1
,
, or
, in general
top bottom
dn O ds U U ds
U U
U U
=
=
=

So, since the Kutta condition requires
top bottom
U U = at TE:

. .
0, Kutta condition
t e
=

For the vortex panel method, this means:

1 1
0
m

+
+ =

Step#5: Set-up System of Equations & Solve

ds V
2
V
1
U
2
U
1
ds
dn
16.100 2002 7
1
2
8
11 12 19 1
21 22 2
31 32 33 3
41
51
61
71
8 81
9
0@ 1
0@ 2
.
. .
. .
.
0@
1 0 0 0 1 0
n
n
n
I
V
I I I
u n i
V
I I u n i
I I I
I
I
I
I
V u n i m I
Kutta
( = = (
( (
= =

( (
( (
( (
( (
( (
=
( (
( (
( (
( (
= =
( (
( (

i

i

i

n
V
(
(
(
(
(
(
(
(
(
(
(
(
(
(

Where =
ij
I total influence of node j at control point i

For example:
36 37
2 1 37 n n
C C I + =

The problem thus reduces to = Ax b , or, using our notation

n
I = V ,

which we solve to find the vector of s!

1
2
3 4
6
5
8
7
9 C
n1
37
C
n2
36
Nodes
Control points
16.100 2002 8
Step #6: Post-processing

The final step is to post-process the results to find the pressures and the lift
acting on the airfoil.

airfoil
L V V ds

= =

So, for our method, this reduces to:

( )
1
1
1
2
1
1
2
1
2
m
i i i
i
m
i i i
l
i
L V S
L S
C
V V c
V c

+
=
+
=
= +
| |
= = +
|
\ .

Vortex Panel Method Summary

In practice, the vortex panel method used for airfoil flows is a little different than
the strategy used in the windy city problem. Heres a summary:

Step #1: Replace airfoil surface with panels

Note: the trailing edge is double-numbered 1 points, panels m m + .

Step #2: Distribute vortex singularities with linear strength variables on each
panel
( )
1
( )
j
j j j j
j
s
s
S

+
= +

1 2
3 4
m
m+1
ds
S
j
j
j+1
(j)
(j+1)
(s
j
)
s
j
ds
16.100 2002 9
This means we have m+1 unknowns:

1 ,
.. ..........
, 3 , 2 , 1 + m m

Step #3: Enforce flow tangency at panel midpoints

0 u n =

at midpoint of every panel

equations m

Step#4: Apply Kutta condition

Kutta condition becomes:

. . 1 1
0 0
1 equations & 1 unknowns
t e m
m m

+
= + =
+ +

Kutta Condition

Thought Experiment
1

Suppose we model the flow around an airfoil using a potential flow approach.

We know the following:
L V

= u =

0 D = 0 =

0 u n =

i
Bernoulli applies

Question: How many potential flow solutions are possible?
Answer: Infinitely many!

For example:
Both of these flows have circulation which are not all equal

1 2 1 2
L L

1
Anderson, Sec. 4.5

V
8
p
8
L'
Flow #1 Flow #2
Kutta Condition
16.100 2002 2
Another difference can be observed at the trailing edge:

As a result of this and the physical evidence, Kutta hypothesized:

In a physical flow (i.e. having viscous effects), the flow will smoothly leave a
sharp trailing edge. -Kutta Condition

Flow #1 is physically correct!

Lets look at Flow #1 a little more closely:

Finite angle T.E. ) 0 ( >
te

Flow leaves t.e. smoothly.
Velocity is not infinite.
Flow turning around a sharp
corner has infinite velocity
at corner for potential flow.
Flow #1
Flow #2
t.e.
V
lower
V
upper
T.E.
Upper and lower surface
velocities must still be
tangent to their
respective surfaces.
This implies 2 different
velocities at TE.
Kutta Condition
16.100 2002 3
Only realistic option:

0 for finite angle T.E.
lower upper
V V = =

Note: from Bernoulli, this implies

2 2
. . . .
0
1 1
2 2
t e t e
p p V V

=
= +

2
. .
. .
1
2
is a stagnation point
with total pressure
t e
t e
p p V
TE
p

= +

Cusped TE ) 0 ( =
te

In order for the pressure at the TE to be unique:

upper lower
V V =

V
lower
V
upper
T.E.
In this case, velocities
from upper and lower
surface are aligned.
Solution

0 = n u

at control pt #1:

The velocity at control pt #1 is the sum of the freestream + 3 point vortices
velocities at that point:

1 2 3
1
2 2 2
2 2 2
u V i i i j

= + +
| | | | | |
| | |
\ . \ . \ .

The normal at control pt #1 is:

1
1 2
1 1
0
2 2
2 2
n i
u n V

=

= + =
| | | |
| |
\ . \ .

Rearranging:

1 2
V

=

(1)

0 = n u

at control pt #2:

Now, following the same procedure for control pt #2:

=10 miles
V = 100 mph
8
3
x
y
Solution
16.100 2002 2

2
2 3
2 2
1 3
2 2
0
2
n i j
V
u n

= +

= + =

2 3
2
V

=

(2)

0 = n u

at control pt #3:

3
1 3
3 3
1 3
2 2
0
2
n i j
V
u n

=

= + =

1 3
2
V

+ =

(3)

Final System of Equations

Combine the numbered equations:
1
2
3
vortex
strengths
(unknowns)
Influence matrix
1 1
0
1 1
0
2
1 1
0
2
i
V n
V
V
V

(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
( (

i

The problem with these equations is that they have infinitely many solutions.
One clue is that the determinant of the matrix is zero. In particular we can add a
constant strength to any solution because:

0
0
0
influence
0
matrix
(

(
(
(
(
=
(
(
(
(

Solution
16.100 2002 3
Given a solution
0
0
0
(
(
(
(

, then
1 0
2 0
3 0

( (
( (
+
( (
( (

is also a solution where
0
is arbitrary.

So, how do we resolve this?
Answer: the Kutta condition!

2 2
. . . .
1 1
2 2
t e upper t e lower
p V p V + = +
0
upper lower
V V =

Whats the Kutta condition for the windy city problem:

Kutta:
3
0 = no flow around node 3!

So, we can now solve our system of equations starting with 0
3
=

1
2
3
2
2
0
V
V
=
=
=

V
8
3
Thin Airfoil Theory Summary
Replace airfoil with camber line (assume small
c
)
Distribute vortices of strength ) (x along chord line for 0 x c .

Determine ) (x by satisfying flow tangency on camber line.

0
( )
0
2 ( )
c
dZ d
V
dx x

| |
=
|

\ .

The pressure coefficient can be simplified using Bernoulli & assuming small
perturbation:

x
z
c
(x) = thickness
z(x) = camber line
x
z
c
z(x) = camber line
x
z
c
(x)dx
16.100 2002 2
{ }
2
2 2 2
2 2
2
2
2 2 2
2
2 2
2
higher order
1
2
1 1
( )
2 2
( )
1
1
2
2
1
2
p
p p
c
V
p V u V p V
p p V u V
V
V
V V u u V
V
u u V
V V
=
+ + + = +
+ +
=
+ + +
=
+
=

2
p
u
C
V
=

It can also be shown that

( ) ( ) ( )
2
( )
lower upper
upper lower
p p p upper lower
x u x u x
C C C u u
V
=
= =

( )
( ) 2
p
x
C x
V
=

Symmetric Airfoil Solution

For a symmetric airfoil (i.e. 0
dz
dx
= ), the vortex strength is:

sin
cos 1
2 ) (
+
=

V

But, recall:

(1 cos )
2
c
x =
16.100 2002 3
2
2
cos 1 2
sin 1 cos
1 1 2
sin 2 (1 )
x
c
x
c
x x
c c
=
=
| |
=
|
\ .
=

1
( ) 2
(1 )
x
c
x V
x x
c c

1
( ) 2
x
c
x V
x
c

=

Thus,
1
4
p
x
c
C
x
c

= .

Some things to notice:
At trailing edge 0 =
p
C .
Kutta condition is enforced which requires
upper lower
p p =
At leading edge,
p
C ! Suction peak required to turn flow around
leading edge which is infinitely thin.

The instance of a suction peak exists on true airfoils (i.e. not infinitely thin)
though
p
C is finite (but large).

Suction peaks should be avoided as they can result in
1. leading edge separation
2. low (very low) pressure at leading edge which must rise towards trailing edge
adverse pressure gradients boundary layer separation.

x
c
C
p
16.100 2002 4
Cambered Airfoil Solutions

For a cambered airfoil, we can use a Fourier serieslike approach for the vortex
strength distribution:

1
flat plate
cambered
contributions
contributions
1 cos
( ) 2 sin
sin
o n
n
V A A n
=
(
(
+
(
= +
(
(
(

Plugging this into the flow tangency condition for the camber line gives (after
some work):

0 0
0
0 0
0
1
2
cos
n
dz
A d
dx
dz
A n d
dx
=
=

After finding the
n
A s, the following relationships can be used to find ,
ac
m
C C
A
, etc.

4
2 1
0 0 0 1
0
2 ( )
( )
4
1 1
(cos 1)
2
c ac
LO
m m
LO
C
C C A A
dz
d A A
dx
=
= =
= =
A

Note: in thin airfoil theory, the aerodynamic center is always at the quarter-chord
(
4
c
), regardless of the airfoil shape or angle of attack.
Important Concepts in Thin Airfoil Theory

1. This airfoil theory can be viewed as a panel method with vortex solutions
taking the limits of infinite number of panels & zero thickness & zero camber

{ }
0
1
0
0
1 ( )
1
2
2
lim lim vortex panel thin airfoil theory
c
N
j ij i
j
thickness N
camber
d dz
V
K V n x dx

=

=
=

2. 2 ( )
l LO
C =

0
1
(1 cos )
(1 cos )
2
LO o o
o
dz
d
dx
c
x
=
=

0 =
LC
for 0
dz
dx
= {i.e. symmetric airfoils}
thickness does not affect
l
C to 1
st
order

3. Moment at
4
c
is constant with respect to according to thin airfoil theory

4
c
= aerodynamic center

4
c
M only depends on camber!

4
0
c
M = for symmetric airfoil

4. Thin airfoil theory assumes:

2-dimensions
Inviscid*
Incompressible*
Irrotational*
Small
Small
max
c
Small
max
z c

* ' 0 D =
Prandtls Lifting Line Introduction

Assumptions:
3-D steady potential flow (inviscid & irrotational)
Incompressible
High aspect ratio wing
Low sweep
Small crossflow (along span)
flow looks like 2-D flow locally with adjusted

Outputs:
Total lift and induced drag estimates
Rolling moment
Lift distribution along span
Basic scaling of
i
D
C with
L
C & geometry

2
e
i
L
D
C
C
A
= dominated by A , where
2
b
A
S
=

2
2
2
2
2
1
i
i
L
q S D
b q S
e
S
L L
D
b q e q e b

| |
|
\ .
=
| |
= =
|
\ .

2
1
i
L
D
q e b
| |
=
|
\ .

In steady level flight where W L = , we have the following options for reducing
i
D :

Raise
q (i.e. raise cruise velocity) Friction increases

Decrease span loading,
W
b
b W & coupled due to structures
Improve wing efficiency, e Can be difficult

16.100 2002 2
Geometry & Basic Definitions for Lifting Line

Chord is variable, ) ( y c c =
Angle of attack is variable, ) ( y = and is a sum of two pieces

N N
N
( ) ( )
g
local freestream
geometric
twist
y y

= +

Effective angle of attack is modified by downwash from trailing vortices

N
( ) ( ) ( )
eff i
induced
y y y
=

Note: 0
i
> for downwash

Local lift coefficient linear with
eff

( )[ ( ) ( )]
l o eff LO
C a y y y =

Fundamental Lifting Line Equation

Basic model results by:

Assuming Kutta-Joukowsky locally gives L :

( ) ( )
( ) 2 ( )
( )
( ) ( )
l
L y V y
L y y
C y
q c y V c y

=

= =

y
c(y)
C
l
(
eff
,y)
x
y=-b/2 y=b/2
16.100 2002 3
Distributing infinitely many horseshoe vortices along wing
4
c
-line to model
induced flow:

| |
2 ( )
( ) ( ) ( )
( )
2 ( )
( ) ( ) ( ) ( )
( )
l o eff LO
o i LO
y
C a y y y
V c y
y
a y y y y
V c y

( = =

= =

2 ( )
( ) ( ) ( )
( ) ( )
LO i
o
y
y y y
a y V c y

= + +
where
2
2
( )
( ) 1
( )
4
b
o o
i
b o
d
y dy
w y dy
y
V V y y
= =

Note: only unknown is ) ( y !

We use a Fourier series to solve this.

=
N
n
n
n A bV
1
sin 2 where cos
2
b
y =

Thus, the governing equation is:

1 1
( )
4 sin
( ) sin ( )
( ) ( ) sin
i
N N
n LO n
n n o
b n
A n nA
a c

= =
= + +

y
x
(x)
Force Calculations for Lifting Line

Recall:

cos
2
sin 2 ) ( ) (
1
b
y
n A bV y
N
n
n
=
= =

=

The local two-dimensional lift distribution is given by Kutta-Joukowsky:

) ( ) ( y V y L =

=
N
n
n
n A V b L
1
2
sin 2 ) (

To calculate the total wing lift, we integrate L :

2
2
2
1
( ) sin
2
2 sin sin
2
b
b
N
n
n
o
b
L L y dy dy d
b
b V A n d
=
= =
( | |
=
|
(
\ .

But:
0,
sin sin
,
2
o
m k
m k d
m k

In this case, n m = and 1 = k . So, the only non-zero term is for 1 = n .

2
(2 )
2 2
n
b
L b V A

| || |
=
| |
\ .\ .

1
2 2
2
A V b L

=

2
1
1
2
1
2
L
L b A
C A
S
V S

= = =
16.100 2002 2
The induced drag is similar. In this case:

) ( ) ( y y V D
i i
=

From previous lecture,

=
=
N
n
n i
n
nA
1
sin
sin
) (

1 1
sin
2 sin
sin
N N
i n n
n m
n
D V nA bV A m

= =
| || |
=
| |
\ .\ .

Integrating along the wing:

( )
2
2
2 2
1 1
0
2 2
1 1
0
only 0 for
( )
sin
sin sin
sin
sin sin
b
i i
b
N N
n m
n m
N N
n m
n m
n m
D D y dy
n
b V nA A m d
b V nA n A m d
= =
= =
=
=
| || |
=
| |
\ .\ .
| || |
=
| |
\ .\ .

2 2 2
1
2 2 2
2
2
N
n
n
i n
b V nA
D b V nA
=
=

2
2
1
2
2
2 1
1
2
or (1 ),
where
i
i
N
i
D n
n
L
D
N
n
n
D
C nA
V S
C
C
A
n
A
=
= =
= +
| |
=
|
\ .

16.100 2002 3
Lift Distributions

The lift distributions due to each of the
n
A terms can be plotted as well:

root

Elliptic Lift Distribution

Recall that minimum induced drag is achieved when 0 =
n
A for 1 > n . In this
case:

sin 2 ) (
sin 2 ) (
1
2
2
A V b L
n A V b L
n
=
=

but:
2
2
sin 1 cos 1
2
y
b

| |
|
= =
|
\ .

2
2
1
( ) 2 1
2
y
L y b V A
b

| |
| =
|
\ .
Elliptic lift

) ( L
2
0
tip at
2
b
tip at
2
b

Trefftz Plane Analysis of Induced Drag

Consider an inviscid, incompressible potential flow around a body (say a wing).
We define a control volume surrounding the body as follows

Upstream flow is
V and is in x direction. Thus, drag is the force in

x direction. Apply integral momentum in x to find induced drag.

+ +
=
S S S S
body body
dS n p dS n u u
K K K K

First, on the body 0 = n u
K K
, so:

+
=
S S S
body
dS n p dS n u u
K K K K

Next, also on the body,

=
body
S
dS n p
K
force of body acting on fluid

We are interested in the exact opposite, i.e. the force acting on the body. In x ,
this is the drag, in z this is the lift, and in y this is a yaw or side force:

x, V
8
y
z
Trefftz plane
S
T
(part of S )
body
S
body
S
8
16.100 2002 2
k L j Y i D dS n p
body
S
K K K
K
=

S S
Di Yj Lk pndS u u ndS

+ + =

K K K
K K K K

Now, lets pull out the drag:

=
S S
dS n u u dS i n p D
K K K
K
K

The next piece is to apply Bernoulli to eliminate the pressure:

) (
2
1
2
1
2 2 2 2
w v u V p p + + + =

+ + + =

S S
dS n u u dS i n w v u V p D
K K
K
K
) (
2
1
2
1
2 2 2 2

But,
2 2
0 for a closed
surface
1 1
( ) ( )
2 2
S S
p V n idS p V n idS

=
+ = +

K K
K K

+ + =
S S
dS n u u dS i n w v u D
K K
K
K
) (
2
1
2 2 2

Next, we divide the velocity into a freestream and a perturbation:

w w
v v
u V u
=
=
+ =

where , , u v w are perturbation velocities (not necessarily small).

Substitution gives:

2 2 2 2
1
( 2 ) ( )
2
S
S
D V V u u v w n idS V u u ndS

= + + + + +

K
K K K

But, we note that

= =

S S
dS n u V dS n u V 0
K K K K
from conservation of mass
16.100 2002 3

+ + + =

S S S
dS n u u dS i n w v u dS i n u V D
K K
K
K
K
K
) (
2
1
2 2 2

If we take the control volume boundary far away from the wing, then the velocity
perturbations go to zero except downstream. Downstream the presence of
trailing vortices will create non-zero perturbations (more on this in a bit).

So, , , 0 u v w except on
T
S .

T
S
D V udS

=

2 2
2
1
( ) (
2
T
S
u v w dS u V

+ + +
)
T
S
u dS +

2 2 2
1
( )
2
T
S
D v w u dS = +

The final step is to note that far downstream the x velocity perturbation must die
away (in inviscid flow). The reason is that the trailing vortices, which far
downstream must be in the x direction, cannot induce an x component of
velocity.

So, this brings us to the final answer

+ =
T
S
dS w v D ) (
2
1
2 2

In other words, the induced drag is the kinetic energy which is transferred into the
crossflow (i.e. the trailing vortices)!

Problem #1

Assume:
Incompressible
2-D flow 0, 0
z
V
z
= =

Steady 0 =
t

Parallel 0 =
r
V

a) Conservation of mass for a 2-D flow is:

1
(
r
r V
r r
N
0
1
) ( ) 0
( ) 0 does not depend on
( )
V
r
V V
V V r
+ =
=

b) -mometum equation is:

V
t
N
( )
r
steady
VV
V V
r

+ +
K
N
2
2
0
1 2
(
r
r
V
p V
V
r r

=

= + +

N
2
0
)
r
V
V
r
=

In cylindrical coordinates:

( )
r
V V =
K
N
0
1
V
r r

=

+

Thus,

1
( )
V
V V V
r

K
N
0 from
continuity
0
=
=

Also,

2
2
2 2
1 1 V V
V r
r r r r

| |
= +
|

\ .
0 =

r
0
r
1
0
Problem #1
16.100 2002 2
Combining all of these results gives:

2
this side is independent of
1 1 p V V
r
r r r r r

( | |
=
|
(

\ .

Since the right-hand-side (RHS) is independent of , this requires that
constant
p
for fixed r . But as varies from 2 0 , it must be equal at

2 & 0 , that is ) 2 ( ) 0 ( = = = p p . If not, the solution would be discontinuous.
Thus, 0
p
constant must be zero!

The differential equation for
V is:

0 ) (
1
2
=
r
V
r
V
r
r r

A little rearranging gives:

0 ) (
1
=
(
rV
dr
d
r dr
d

Integrating once gives:

1
) (
1
C rV
dr
d
r
=

Integrating again gives:

2
2
1
2
1
C r C rV + =

2
1
1
2
C
V C r
r
= +
Problem #1
16.100 2002 3
Next, we must apply the no-slip boundary conditions to find
V . Specifically,

at
o o o
r V r r
= = ,

at
1 1 1
, r V r r
= =

because flow velocity equals wall velocity in a viscous flow.

So, apply
1
& r r r r
o
= = bcs:

1
1
1
2
1
1 1 0
0 1
2
1 1 1 1 1 1
2
1
1 0
0 1
1
1
2
2
1
( )
2
o
o
o
o o o
o
o o
r r
r r
C
C
r C r r r
r
r r
C
r C r r r
C
r
r r
r r

=
= +

`

= +
=
)

Or, rearranged a little gives:

1
1
1 1
1 1
1 1
o
o
o o
o o
o o
r r
r r
r r r r
V r r
r r r r
r r r r

= +

c) The radial momentum equation is:

2 2
2 2
1 1 2
( )
r r
r r
V p V V
V V V V
t r r r r

| |
+ = +
|

\ .
K

But 0 & 0 =
V
V
r
so this reduces to:

r
V
r
p
2

Since
2
0
V
r
always, then clearly 0
r
p
.

Thus, pressure increases with r .

Problem #1
16.100 2002 4
d) On the inner cylinder, the moment is a result of the skin friction due to the fluid
shear stress. For this flow in which only 0
V and is only a function of r , the

only non-zero shear stress is
r
and has the following form:

, the only
non-zero strain
r
r
V V V
r
r r r r

=
( | | | |
= =
| |
(

\ . \ .

Problem #1
16.100 2002 5
Rotating Cylinders

For the problem you studied in the homework:

1. What direction is the fluid element acceleration?

2. What direction are the net pressure forces on a fluid element?

3. What direction are the net viscous forces on a fluid element?
o

1
Viscous Flow: Stress Strain Relationship

Objective: Discuss assumptions which lead to the stress-strain relationship for
a Newtonian, linear viscous fluid:

j
i k
ij ij
j i k
k
k
u
u u
x x x
u u v w
V
x x y z

| |

= + +
|
|

\ .

= + + =

i

where = dynamic viscosity coefficient
= bulk viscosity coefficient

Note:
0,
1,
ij
i j
i j

=

=

1
shear strain rate in , plane
2
j
i
ij i j
j i
u
u
x x
x x
| |
= +
|
|

\ .

Thus, written in terms of the strain rates, the stress tensor is:

( )
viscous stress
using indicial notation
due to shearing
this is
of a fluid element
viscous stress due to an
overall compression or
expansion of the fluid
element's volume
2
kk
ij ij ij xx yy zz
= + + +

This stress-strain relationship can be derived by the following two assumptions:

1. The shear stress is independent of a rotation of the coordinate system
2. The shear stress is at most a linear function of the strain rate tensor.
So, for example,
xy
:

zz yz yy xz xy xx xy
a a a a a a
33 23 22 13 12 11
+ + + + + =

Clearly, assumption #2 gives 6 unknowns per shear,
12 11
, a a , etc. Note: why do
zy zx yx
& , not appear in this expression for
xy
? The total number of unknowns
for all stresses are: 36. But, this can be eventually reduced by applying #1 to the
most general linear form to the two unknowns & .

Viscous Flow Stress-Strain Relationship
16.100 2002 2
Stokes hypothesis

Stokes hypothesized that the total normal viscous stress on a fluid element
surface,

) )( 3 2 (
zz yy xx zz yy xx
+ + + = + +

should be zero, i.e. 0 = + +
zz yy xx
so that the normal force (stress) on a
surface is only that due to pressure. This requires that

3
2
0 3 2 = = +

Comments

Experimental studies have indicated that
3
2
and in general is not
negative!

For incompressible flow, 0 = = + + V
zz yy xx
so 0 = + +
zz yy xx

regardless of .

For most compressible flows V
is small compared to shearing strains (i.e.

yz xy
, , etc.) so again, Stokes hypothesis has little impact. So, as a result,
common practice is to assume that
3
2
= .
Integral Boundary Layer Equations

Displacement Thickness

The displacement thickness
*
is defined as:

*
0 0
1 1
e e e
compressible incompressible
flow flow
u u
dy dy
u u

= =

The displacement thickness has at least two useful interpretations:

Interpretation #1
0
0
(1)
e
u
dy
u
dy
=
=
A
A+ B

So, the difference is in area B.

*
represents the decrease in mass flow due to viscous effects, i.e. lost
*

e e visc
u m =

y
u
/
u
e
u
/
u
e
(y)
A
B
16.100 2002 2
Interpretation #2

Conservation of mass:

1 1
1 1
1
1
0 0
0 0
0
0
( )
1
y y y
e
y y
e e
y
e e
y
e
u dy udy
u dy udy yu
yu u u dy
u
y dy
u
+
=
= +
=

=

Taking the limit of
1
y gives

*
0
1
e
u
y dy
u

= =

So, the external streamline is displaced by a distance
*
away from the body
due to viscous effects.

Outer flow sees an effective body

s
tre
am
line
y
*(x)
y
y
1
u(y)
x
u
e
16.100 2002 3
Karmans Integral Momentum Equation

This approach due to Karman leads to a useful approximate solution technique
for boundary layer effects. It forms the basis of the boundary layer methods
utilized in Prof. Drelas XFOIL code.

Basic idea: integrate b.l. equations in y to reduce to an ODE in x .

Derivation:

Add ) ( u x continuity + x momentum

2
2
( ) continuity momentum
e
e
u x
u v u u du u
u u v u
x y x y dx y

+ + + = +

2
( )
( )
e
e
u du u
uv u
x y dx y y

+ = +

Now, we integrate from 0 to
1
y :

1
1 1
2
1
0 0
0
( )
y
y y
e
e
u du
dy uv u y
x dx

+ = +

Note:

1 1
1
1
0
0 0
( )
y y
y
e e e
v u
uv u v y u dy u dy
y x

= = =

So, the equation becomes:

x
y
*(x)
u
e
(x)
16.100 2002 4

1 1
1
2
1
0
0 0
( )
y y
y
e
e e
u u du
dy u dy u y
x x dx

= +

After a little more manipulation this can be turned into (note we let
1
y also):

2 *
( )
e
w e e
d du
u u
dx dx
= + (1)

where momentum thickness =
0
1
e e e e
u u
dy
u u

incompressible form = 1
e e o
u u
dy
u u

Insight

Integrate (1) from stagnation point along airfoil & then down the wake

2 *
0
0 0
( )
e
w e e
du
dx u u dx
dx

= +

But: 0 =
e
u at stag. pt.
Bernoulli
( 0) &
e
e
dp du
x u
dx dx
= =

2 *
0 0
drag (see Anderson
Sec 2.6 for proof)
e w
x
dp
u dx dx
dx

= +

N
*
0 0
friction
form drag
drag
w
dp
D dx dx
dx

= +

x=0
at stag. point
x
y
x
along wake
8
16.100 2002 5

Another common form of the integral momentum equation is derived below:

dx
du
u u
dx
d
e
e e e e w
* 2
) ( + =

2
*
(2 )
where
known as "shape parameter"
w e
e e e
d du
H
u dx u dx
H

= + +
=

Correlation Methods for Integral Boundary Layers

We will look at one particularly well-known and easy method due to Thwaites in
1949.

First, start by slightly re-writing the integral b.l. equation. We had:

dx
du
u
H
dx
d
u
e
e e e
w

) 2 (
2
+ + =

Multiply by
v
u
e
:

) 2 (
2
H
dx
du
v dx
d
v
u
u
e e
e
w
+ + =

Then define
dx
du
v
e
2
= and this equation gives:

+ = ) 2 ( 2 ) ( H
u
dx
du
dx
d
u
e
w
e
e

Thwaites then assumes a correlation exists which only depends on .
Specifically:

) ( H H = and ) (

S
u
e
w
=

shape factor shear correlation
correlation

[ ] ) ( 2 ( ) ( 2 ) (
H S
dx
du
dx
d
u
e
e
+

now this is an approximation

Correlation Methods for Integral Boundary Layers

In a stroke of genius and/or luck, Thwaites looked at data from experiments and
known analytic solutions and found that

! ! 6 45 . 0 (
dx
du
x
dx
d
u
e
e

This can actually be integrated to find:

dx u
u
v
x
o
e
e

=
5
6
2
45 . 0

where we have assumed 0 ) 0 ( = = x for this.
16.100 2002 2
Method of Assumed Profiles

Here are the basic steps:

1. Assume some basic boundary velocity profile for ) , ( y x u . For example, this is a
crude approach but illustrates the ideas:

, 0 ( )
( , )
( )
( )
1, ( )
e
y
y x
u x y
x
u x
y x
<

where ) (x is the single unknown describing the velocity distribution.

2. Calculate ) ( ,
*
H or , and
f
C for the assumed profile:

2
*
0 0
0
1
1 1
2
e
u y y
dy dy y
u
| | | |
| |
= = =
| | |
\ .
\ . \ .

*
1
2
=

2 3
2
0 0 0
1 1
1 1
2 3
e e
u u y y y y
dy dy
u u
| | | |
| |
= = =
| | |
\ .
\ . \ .

1
6
=

Note:
*
1
2
3
1
6
H

= = =

Finally, to find
f
C we need
0
w
y
u
y

=

, for 0
e
e
u y u
u y
y y

| |
= = <
|

\ .

16.100 2002 2
2 2
2
1 1
2 2
e
w
f
e e
e e e e
u
C
u
u u

= = =

3. Plug results from step 2 into integral b.l. equation:

dx
du
H
u dx
d
C
e
e
f
) 2 (
2
+ + =

So, for our assumed linear profile:

1 5
6 6
e
e e e
d du
u dx u dx

= + (1)

where ) (x is the only unknown. We can solve this by specifying ) (x u
e
, setting
an initial value for at 0 = x (i.e. the leading edge) and then integrate in x .
Note: in many cases, this integration will need to be done numerically.

Example: Flat Plate

Since
= u x u
e
) ( is a constant, the governing equation (1) becomes:

dx
d
u

6
1
=

Re-arrange and integrate:

x
y
U
8
8
*(x)
e
=
8 U
e
(x)=U
8
16.100 2002 3

2
2
0 0
2 2
6
6 1 ( )
2
6 1 ( )
2
6 1
( ) (0)
2
x x
d
U dx
d
U dx
d
dx dx
U dx
x x
U

=
=
=
( =

But, our initial condition is 0 ) 0 ( = .

2
6 1
( )
2
x x
U

=

12 2 3 3.464
Re Re
x x
x U x

= = =

x
x x
Re
3 2
2
1
2
1
*
= =

x x
x
Re
732 . 1
Re
3
*
= =

and
f
C :

Re 2 2
2 3
1
3 Re
x
f
e e
f
x
C
u u x
C

= =
=

0.577
Re
f
x
C =
16.100 2002 4
Comparison with Blasius solution:

Blasius Int. Method with linear velocity

x
x
Re
*
1.720 1.732

x f
C Re 0.664 0.577
Falkner-Skan Flows

For the family of flows, we assume that the edge velocity, is of the
following form:
) (x u
e

arbitrary constant = = K Kx x u
m
e
) (

The pressure can be calculated from the Bernoulli in the outer, inviscid flow:

1 2 2
2
.
2
1
=
=
= +
m
m
e
e
e
e
e e
x K
dx
dp
dx
du
u
dx
dp
const u p

if 0 > m then < 0
dx
dp
e
favorable pressure gradient

if 0 < m then > 0
dx
dp
e
adverse pressure gradient

These edge velocities result from the following inviscid flows:

m
m
+
1
2

y

) (x u
e

y
x
2
) (x u
e

0 = x

Flow around a corner (diffusion) Wedge flow
0 2 2 0

Falkner-Skan Flows
Some important cases:

0 , 0 = = m : flat plat (Blasius)
1 , 1 = = m : plane stagnant point

The boundary layer independent variable from the Blasius solution generalizes
to:

vx
x u m
y
e
) (
2
1 +
and ) ( ) ( ) , (
1
m f x u y x u
e
=

An interesting case in 1 , 1 = = m , i.e. stagnation point flow:

) (x u
e

x

x e
K u =

inviscid flow velocity
increases away from
stag. pt. at 0 = x
y

x
x
v
K
y
2
1 1+
=

v
K
y = is independent of x
Boundary layer at a stagnation point does not
grow with x !

The skin friction can be found from:

) (
) (
11
0 0
2
0
o f
y d
f d
x u
y
u
y
e
y
w
= = =

16.100 2002 2
Falkner-Skan Flows
Since
vx
x u m
y vx
x u m
y
e e
) (
2
1 ) (
2
1 +
=
+
=

) (
) (
2
1
) (
11
o f
vx
x u m
x u
e
e w
+
= tabulated

The skin friction coefficient is normalized by ) (
2
1
2
x u
e
:

) (
) ( 2
1
2
) (
2
1
) (
11
2
o f
x x u
v m
x u
x C
e
e
w
f
+
=

v
x x u
o f
m
C
e
x
x
f
) (
Re
Re
) (
2
1
2
11
+
=

Note: separation occurs when 0 =
f
C which means . From the table,
this occurs for
0 ) (
11
= o f
19884 . 0 =

This is only an angle of 18
o
!
18
o
16.100 2002 3
Effect of Turbulent Fluctuations on Mean Flow: Reynolds-Averaging

In a turbulent flow, we can define the mean, steady flow as:

0
1
( , , ) lim ( , , , )
T
T
u x y z u x y z t dt
T
=

This allows us to split the flow properties into a mean and a fluctuating part:

mean turbulent
part fluctuating
part
( , , , ) ( , , ) ( , , , )
( , , , ) ( , , ) ( , , , )
( , , , ) ( , , ) ( , , , )
( , , , ) ( , , ) ( , , , )
u x y z t u x y z u x y z t
v x y z t v x y z v x y z t
w x y z t w x y z w x y z t
p x y z t p x y z p x y z t
= +
= +
= +
= +

Note: the mean of u is zero:

{ }
0 0
1 1
lim lim
T T
T T
u u u
u u dt u dt
T T

=
=

0
1
lim
T
T
u
udt u u
T

N
0
u u u
=
=
0 u = mean of fluctuations is zero.

Now, we will develop equations which govern the mean flow and try to develop
some insight into how the fluctuations alter the mean flow equations.

Lets start with incompressible flow and look at the x momentum:

x momentum:

2 2 2
2 2 2
1 u u u u p u u u
u v w
t x y z x x y z

+ + + = + + +

Lets look at the averaging of this equation in time to develop an equation for
the mean flow.


16.100 2002 2
[ ]
time-average 0
of x-momentum
1
lim or
T
T
x mom dt x mom
T

The first term is:

0 0
1 1 ( )
lim lim
T T
T T
u u u
dt dt
T t T t

+
=

But, 0 =
t
u
thus:

0 0
1 1
lim lim
T T
T T
u u u
dt dt
T t T t t

= =

Just as 0 u = , we will assume 0 =

t
u
.
End result:

0 =
t
u

Lets skip over to the pressure term and look at its average:

p p p p p p
x x x x x x
p p p
x x x
p p p
x x x

= + = +

= +

= +

But, 0 = p thus:

x
p
x
p

Similarly,

2
2
2
2
2
2
2
2
2
2
2
2
z
u
y
u
x
u
z
u
y
u
x
u

T

16.100 2002 3
Combining these into mom x , we now have:

2 2 2
2 2 2
must still work this out
1 u u u p u u u
u v w
x y z x x y z

+ + = + + +

Lets work out the last term:

z
u
w
y
u
v
x
u
u
z
u
w
y
u
v
x
u
u

Thats the easy part. Now, consider
x
u
u
:

Question: What does
x
u
u
equal in terms of & u u only?

a)
x
u
u

b)
u u
u u
x x

+

c)
x
u
u
x
u
u

d) none of the above

Poiseuille Flow Through a Duct in 2-D

Assumptions:
Velocity is independent of 0 , =
x
v
x
u
x
Incompressible flow
Constant viscosity,
Steady
Pressure gradient along length of pipe is non-zero, i.e. 0
x
p

Boundary conditions:

No slip:
( ) 0
walls are not moving
( ) 0
u y h
v y h
= =

= =

To be clear, we now will take the compressible, unsteady form of the N-S
equations and carefully derive the solution:

Conservation of mass:

0 ) ( = +
V
t

But 0 =
because flow is steady and incompressible. Also, since =

constant, then V V

= ) (

0 =
=
y
v
x
u
V

Finally, 0 =
x
u
because of assumption #1 long pipe.

0 =
y
v

h y + =
h y =
y
x

16.100 2002 2
Now, integrate this:

C v = = constant

Apply boundary conditions: ( ) 0 ( ) 0 v h v y = =

We expect this but it is good to see the math confirm it.

Now, lets look at momentum y .

Conservation of momentum y :

xy yy
xy yy
Dv p
Dt y x y
v p
V v
t y x y
v
t

= + +

+ = + +

0
steady
v
u
x
=
0
v
=
+
0
xy yy
v p
y y x y

=

= + +

Now, what about
yy xy
&

xy
u v
y x

= +

0
2
xy
yy
u
y
v
y

=
| |
|
=
|

|
\ .
0 v
u v
x y
=
| |
+ +
|

\ .
0
0
yy
V
=
=

So momentum y becomes:

0
p u
y x y
| |
= +
|

\ .

But 0
u
x y
| |
=
|

\ .
because constant = =
& 0
x
u


16.100 2002 3

0 (steady)
0 ( , , ) ( )
p
t
p
p x y t p x
y

=
= =

Conservation of momentum x :

( )
only
yx
xx
p p x
Du dp
Dt dx x y
= + +

u
t
0
steady
u
u
x
=
0
v
=
+
0
2
u dp u
y dx x x
=

= +

0
0
u
y y
dp u
dx y y
=
| |
| | |
+
| |

\ .
|
\ .
| |
= +
|

\ .

Now, we just need to solve this

u dp
y y dx
| |
=
|

\ .

so, & const ) ( y u u = =

2
2
only ( ) only ( )
1
must be constant
f y f x
d u dp
dy dx
=

= const.
dx
dp
pressure can only be a linear function of x !

Now, integrating twice in y gives:

2
1 0
1
( )
2
dp
u y y C y C
dx
= + +

Finally, apply BCs:


16.100 2002 4

2
1 0
2
1 0
( ) 0
1
( ) 0
2
1
( ) 0
2
u h
dp
u h h C h C
dx
dp
u h h C h C
dx
=
+ = + + =
= + =

Solve for
0 1
& C C gives:

2
0
1
1
2
0
dp
C h
dx
C

=
=

2
2
1
( ) 1
2
dp y
u y
h dx h
(
| |
=
(
|
\ .
(

Assume steady
0 =
t

Assume 1 >>
h
L

0
V
x
K

Assume 2-D
0 , 0 =
=
z
w

Incompressible N-S equations:

1. 0 =
y
v
x
u

2.
2 2
2 2
1 u u u p u u
u v
t x y x x y
| |
+ + = + +
|

\ .

3.
2 2
2 2
1 v v v p v v
u v
t x y y x y
| |
+ + = + +
|

\ .

BCs

w
u h x u
h x u
h x v
= +
=
=
) , (
0 ) , (
0 ) , (

Turning the crank:

N
0
0 0 ( )
but 0 const
Apply bc's 0
x
u v v
v v x
x y y
v
v
x
v

+ = = =

= =
=

Now, momentum y : Since 0 = v , we have:

0 ( , ) ( )
p
p x y p x
y
= =

y
2h
x
y=-h
y= h
p
L
p
R
u
w
Lecture 28
16.100 2002 2
Note: since the pressure does change from to
L R
p p over the length , ( ) L p p x = .

Finally momentum x :

N N
N
N
2 2
2 2
0
0
0
0
2
2
1
1
, where
steady
x
t
u u u dp u u
u v
t x y dx x y
u dp
y dx

=
=
=
| |
|

|
+ + = + +
|

|
\ .

Observe that ) ( ) ( x g RHS y f LHS = = and

L
p p
dx
dp
x g y f
L R

= =
= =
const
const. ) ( ) (

For this problem, Ill just use the gradient
dx
dp
but realize this is specified by the
end pressures.

Next, integrate in y :

2
2
1
2
1
1
1
1
2
o
d u dp
dy
dy dx
du dp
y C dy
dy dx
dp
u y C y C
dx
(
=
(

(
= +
(

= + +

Now, apply bcs:

w o
o
u C h C h
dx
dp
h y u
C h C h
dx
dp
h y u
= + + = + =
= + = =
1
2
1
2
2
1
) (
0
2
1
) (

Solving for : &
1
C C
o

Lecture 28
16.100 2002 3

2
1
1 1
2
2
o w
w
dp
C u h
dx
u
C
h
| |
= +
|
\ .
=

2
2
( ) 1 1
2 2
w
h dp y u y
u y
dx h h
(
| | | |
= + +
(
| |
\ . \ .
(

Suddenly started flat plate (Stokes 1
st
Problem)

0
IC: 0,
0
( , 0)
BC: 0,
( , 0) 0
w
u
t
v
u x u
t
v x
=
=

=
=
>

=

Assume infinite length, 0 =
x

Continuity:

N
0
0 ( )
but 0 so 0
u v
v v x
x y
v
v
x
=

+ = =

= =

momentum y :

N N
N
2 2
2 2
0 0
0
0
1
0 ( )
v v v p v v
u v
t x y y x y
p
p p x p
y
= =
=
=
| |
+ + = + +
|

\ .
= = =

y

p
w
u
Lecture 28
16.100 2002 4
momentum x :

N
N
N
N
2 2
2 2
0
0 0
1
p p
x x
u u u p u u
u v
t x y y x y
=

=
= =

| |
|

|
+ + = + +
|

|
\ .

2
2
u u
t y

=

This is the diffusion equation (also known as heat equation).

There are many ways to solve this equation
Well use a similarity solution approach used in boundary layer theory.

Similarity Solution

Assume that ). , ( ) ( ) , ( y t u y t u = = where Reduce PDE to ODE.
Usually, the assumption is made that:

y
b
y bCt
y
t
a
y aCt
t
y Ct
b a
b a
b a

= =
= =
1
1

d
du
t
a
t d
du
t
u
=

Lecture 28
16.100 2002 5

2
2
2
2
1
2
2
2
2
2
( )
( 1)
a b
a b
u u du d
y y y y d dy
du b
y d y
b du du b
y y d d y y
b u du
bCt y
y d y
b u du
b b Ct y
y d
| | (
= =
| (

\ .
(
=
(

| | | |
= +
| |

\ . \ .
| |
= +
|

\ .
| |
= +
|

\ .

d
du
y
b b
d
u d
y
b
y
u
2 2
2
2
2
2
) 1 ( ) ( + =

Thus:

2
2
2
2
2 2
becomes
( 1)
u u
v
t y
a du b d u du
b b
t d y d y d

=

| |
= +
|
\ .

Re-arranging:

N
2 2
2 2
needs to be
( ) only
1
f
b
d u a y b du
d b t b d
y
C
t
(
(
(
=
(
(
(

| |
=
|
\ .

For simplicity,
1
1 and
2
b C
= =
1
2
a =

y
t
=
Lecture 28
16.100 2002 6

d
du
d
u d
2
2
2
=

Note: bc is
w
u u = ) 0 (
= 0 ) ( u Also is correct initial condition

2

= Ce
d
du

Integrate again

2
0
( )
o
u C e d C
= +

Laminar Boundary Layer Order of Magnitude Analysis

u

Assumptions (in addition to incompressible, steady & 2-D)

Changes in direction x occur over a distance c
)
1
( write we or,
1
~
C
O
x C x
=

Changes in direction y occur over a distance
)
1
( or,
1
~

O
y y
=

C << (boundary layer is thin)
1 Re >> =
c u
(Reynolds number is large)
Changes in velocity x are proportional to
u
) (

= u O u
Changes in pressure are proportional to
2

u
) (
2

= u O p

Now, lets do the order of magnitude analysis:

Conservation of Mass

0 =
y
v
x
u

Now, we introduce our assumed orders. For example:

) (
c
u
O
x
u

=

c
y
x
) (x

16.100 2002 2
Since we have not assumed an order for v variations, just leave this as simply
v and say,

) (
v
y
v
=

Since 0 =
y
v
x
u
, they obviously have equal (though opposite) magnitudes and
therefore orders. Thus, the changes in v scale as:

) (
C
u v

=

Since ) ( wall, a at 0 v v v = = itself and we can say

) (
c u
v
=

momentum x

) ( ) ( ) ( ) ( ) (
2 2
2 2 2
2
2
2
2
u
c
u
c
u
c
u
c
u
y
u
x
u
x
p
y
u
v
x
u
u

If we compare the last two terms, clearly

2 2

<<
u
c
u

Since we have assumed C << . Thus, we can assume that
2
2
y
u
is small.


16.100 2002 3
In order for any viscous terms to remain, this requires that
2
2
x
u
be equal in
order to the remaining inviscid terms. That is, e.g

) ( ) (
) ( ) (
) ( ) (
2
2
2
2
2
c u c
u
c
u
y
u
x
u
u

=
=

In other words,

)
Re
1
( =
C
Major result!

momentum y

)
1
( ) ( ) ( ) ( ) (
3
2
2
2
2
2
2
2
2
2
c
u
c
u
u
c
u
c
u
y
v
x
v
y
p
y
v
v
x
v
u

Normalizing all of these orders by
C u /
1
2

and substituting in for
C u
Re
and )
Re
1
( =
C
gives:

)
Re
1
( )
Re
1
( ) Re ( )
Re
1
( )
Re
1
(
2
3

Clearly, except for
y
p
, all other terms are small as Reincreases. Thus, we must

conclude that:

layer boundary a for equation momentum y the is
small is
y
p
y
p
0

Solutions of the Laminar Boundary Layer Equations

The boundary layer equations for incompressible steady flow, i.e.,

2
2
0
y
u
dx
dp
y
u
v
x
u
u
y
v
x
u
e
+ =

have been solved for a handful of important cases. We will look at the results for
a flat plate and a family of solutions called Falkner-Skan Solutions.

Flat Plate (Laminar): Blasius Solution

For a flat plate, constant =

p p
e

0 =
dx
dp
e

Blasius was able to show that the boundary later equations could be rewritten to
only depend on a parameter,

vx
V
y
2

and its derivatives

The resulting solution has been tabulated and compared to experiments on the
following page. Note:

=
= =
V vx
f V
y
u
d
df
f f V y x u
y
w
/ 2
) 0 (
where ) ( ) , (
0

These values from the solution of ) ( f can be used to find:

= V y x u y 99 . 0 ) , (
% 99
which at location

Note: since ) ( set we , 0 x p p
y
p
e
= =
,
i.e. the boundary layer edge pressure.
Solutions of the Laminar Boundary Layer Equations

16.100 2002 2
From the table, 5 . 3 at 99 . 0 ) ( = f :

x
V
vx
vx
V
vx
V
y

2
5 . 3
2
5 . 3
2
% 99
% 99
as grows layer boundary =
=
=

Typically, this result is written non dimensionally as:

x
v
x V
x
x
x
on based number s Reynold'
where

= Re
Re
0 . 5
% 99

We can also find:

x
w
f
x
x
V
C
x
x
Re
664 . 0
2
1
Re
664 . 0
Re
7208 . 1
2
*
= =
=
=

Comment:
At leading edge of a flat plate
f
C x gives this and 0 !
In reality, the leading edge of an infinitely thin plate would have very large,
but not infinite
f
C .
The problem is that near the leading edge of a thin plate, the boundary
layer equations are not correct and the Navier-Stokes equations are
needed. Question: Why did the boundary layer approximation fail at
0 x ?
Similarity in Wind Tunnel Testing

In terms of non-dimensional force and moment coefficients, these depend on
numerous non-dimensional input parameters.

,......) Re, , (
,......) Re, , (
=
=
M C C
M C C
D D
L L

In many aerodynamic applications,

) Re, , (
) Re, , (
=
=
M C C
M C C
D D
L L

So, to match flight condition
D L
C C & in an experiment, we should match flight
. Re, , and
M

Matching is relatively easy
How about simultaneous Re &
M matching?

= =
L V
a
V
M Re

Consider the Wright Brothers Wind Tunnel:

a is the atmospheric speed of sound at ground level to good approximation

=
=
a
a
V V
M M
T
T
T

For altitudes from 0-30k ft.,
a varies by only about ~15%.

!
V V
T

Since Wright Brothers Wind Tunnel is limited to about mph 200 <
T
V , we cant
match
M unless it is low.

For
M low, say 3 . 0
~
<
M , the effects of
M are small. So, in this case, we

only need match Re:

=
=
L V L V
T
T T T
T
Re Re

ft/sec
ft/sec
950
1100
30

k
ok
a
a

16.100 2002 2
Consider VAT @ T/O and the use of WBWT to simulate it:

=
=
V
V
L
L
L V L V
T T
T T
T T
. conditions ground same the y essentiall are & , ,

mph Max
mph T/O, For
200 ,
200
T T
V V
V
! 1
L
L
T

Transonic & Supersonic Tests

In the case where
M is larger than about 0.3, we need to consider matching of
M & Re in the tunnel.

What can be done?

= =
= =
L V L V
a
V
a
V
M M
T
T T T
T
T
T
T
Re Re

Note:

for ideal gas

So,
T
V is largely set by
M and the values of

T
a , which are achievable.
Typically ) (

= V O V
T

Then, how can we match Re?

Note: we would like to reduce
L L
T
from to reduce model & tunnel size.

T T T
T
V
V
L
L

=

/
/

) (
) (
RT
p
T
RT a
=
=
=

some largely set by
M
flexibility
16.100 2002 3

) (
/
T T T
T
T
T T
T
T
T RT
p RT p

= =

One possible approach: increase pressure in tunnel.

Engineering Solutions to
M & Re Matching

Traditionally:

Test at
Re but neglect
M match if
M is low

Test at
M , trip boundary layers, ignore Re match at high Re (Re >10

6
)

Estimate effects of not matching comp. & flight tests
Single Horseshoe Vortex Wing Model

Lift due to a horseshoe vortex

Kutta-J oukowsky Theorem

b V dy V L
b
b
= =

2
2

S
b
b V
C
S V
b V
S V
L
C
L
L
2
2 2
2
2
1
2
1
= =

A
b V
C
L
=
2

b
S
b
trailing
vortices
bound
vortex

16.100 2002 2
Induced Drag

To estimate the induced drag using this simple model, we will assume that the
3-D lift is tilted by the downwash occurring at the wing root ( 0 = y ).

i i eff i eff i
eff i eff
i eff
L L L D
L L L

= =
=
=

sin
cos

For small
i
&
V
w
i
i

To calculate downwash, we apply Biot-Savart:

=
filament
r
r d
z y x w
3
4
) , , (
v
l
v
v

L
eff
L
i

i
D
V
i
w
i

eff

x
b
y
x

r
v

+ x
+ x
) 0 ,
2
, (
b
x
l d
i
D L
V L


16.100 2002 3
At 0 = = = z y x (i.e. wire root), the bound vortex does not produce a downwash,
so, we only have the 2 trailing vortices at
2
b
y = .

=

+ = =
2 2
4
) 0 , 0 , 0 (
b
y
b
y
w
v

Lets do the
2
b
y = integral first:

+
=
+ = + =
+

=
+ +
=
0
2
3
2 2
0 0
2
3
2 2
2 2
3
)
2
(
2
)
2
(
)
2
( ) ( ) (
b
x
k x d
b
b
x
j
b
i x i x d
r
r d
x
x x
v
v v v
v
l
v

b
w
= ) 0 , 0 , 0 (

Then, combining this we can find:

=

=
=
bV
L
D
bV
L D
V
w
L L D
i
i
i
i i
) (

But
b V
L
b V L

= =

2 2
2
b V
L
D
i

=


16.100 2002 4

2
2
2 2 2
)
2
1
(
2
1
b
S
S V
L
S V
D
C
i
D
i

= =

ok?? this is Hmmm
2
2
A
C
C
L
D
i
=

Recall: Lifting line:

1 ,
2
= e
e A
C
C
L
D
i

Ground Effect Using Single Vortex Model

What is the boundary condition at ground ( 0 = z ) and does a single horseshoe
vortex satisfy it??

B.C.: solid wall

Consider from far downstream:

So, to satisfy bc:

Image vortices

b
h
! 0 0
0
= =
=
z w
n u
at
v v

h b
z

y
b
b

h
h
Ground Effect Using Single Vortex Model

16.100 2002 2

Result: downwash is decreased due to upwash caused by ground plane (which
is modified by image vortex).

For our simple model:

=
2
) ( 16 1
1
b
h
b b
w
i

original ground
isolated effect
wing effect

+
=
+
=
i
i i
w
b
h
b
h
w
b
h
b
h
b
w
2
2
2
2
) ( 16 1
) ( 16
) ( 16 1
) ( 16
=

i
i
i
D
b
h
b
h
bV
L
V
w
L D
2
2
) ( 16 1
) ( 16

2
2
) ( 16 1
) ( 16
b
h
b
h
D D
i i
+
=

2
2
) ( 16 1
) ( 16
b
h
b
h
C C
i i
D D
+
=

Image vortex
(opposite
strength)
i
i
) (
) (
are
true for
fixed
) (i.e. L
Three-Dimensional Wall Effects

In a freestream, recall that a lifting body can e modeled by a horseshoe vortex:

V

Consider a rectangular cross-section tunnel:

Flow is into page

The image system for this looks like:

wing
y y y
y y y
y y y
y y y
y y y
y y y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
y
images
images
images
images
actual tunnel
images
images
images
images
y y
y y
y y
y y
y y
y y
y y
y y
y y

16.100 2002 2
The effect of these images is:

For fixed lift, such that is constant,

an upwash exists due to images is effectively larger

i tunnel
+

effective AOA of correction due to
freestream model in upwash induced by
AOA tunnel images

Similarly, this creates decrease in induced drag relative to freestream flight:

Recall,

i
tunnel
i i
i
i
D D D
i L D
i L D
C C C
C C
C C
+ =
=

Or, since we are interested in the total drag:

i tunnel
D D D
C C C + =

Specific formulas derived in detailed analysis give that:

L i
C
C
S
) ( =

where area reference = S

geometry model & tunnel on depends which factor
area sectional - cross tunnel
=
=
C

Wright Brothers is an elliptic cross-section with dimensions 10 ft wide by 7 ft high.

b
B
h
16.100 2002 3
Define:

e
b

actual trailing vortex

b b
B
b
k
B
h
e
e
9 . 0 span effective

horseshoe
model
b
caused because wing tip trailing
vortices are inboard of tips a
short distance downstream
Values of for a wing with uniform loading in a closed elliptical jet
K =Effective span/J et width
B
o
u
n
d
a
r
y

C
o
r
r
e
c
t
i
o
n

F
a
c
t
o
r
,

Drag Tare Due to Mount

Force balance will measure drag which is due to exposed portion of mount:

fairing
Forces
measured
here
extra drag
due to
exposed
mount

mechanism
to adjust
Two techniques to estimate drag tare:
Remove model and run tests to find drag of mount
- will include extra stuff hidden inside model
- not same as presence of model
Approximate it from known drag coefficients since mount is nearly cylindrical.
- approximate
Should try both approaches!
Subsonic Small Disturbance Potential Flow

1. j v i u V V
v v v
) ( + + =

where 1
| |
2
2 2
<<
+
V
v u

2. = +

j v i u
v v
perturbation potential

y
v
x
u

3. small-disturbance (?) and bcs=
0

) 1 (
2
2
2
2
2
=

y x
M

BC: ) ( ) 0 , ( x
dx
dy
V x v
c
= where =
c
y camber line
Note: this eqn and bc are valid for subsonic and supersonic flow.
Also note,

=
=
V
u
q
p p
C
u
p
2

4. For subsonic flow, we utilize the following mathematical transformation:
Define:

2
1

= M

=
=
=
y
x

) ( ) 0 , (
0
2
2
2
2

dx
dy
V
c

The implications are that the subsonic compressible flow around an airfoil can be
related to the incompressible ) 0 ( =
M flow about the airfoil.

small
disturbances
are assumed
Same eqn
Same bc
0 =
M
0

2
2
2
2
=
y x

) ( ) 0 , ( x
dx
dy
V x
y
c
)

1 0 < <

M
0
2
2
2
2
=

) (
dx
dy
V
c

Subsonic Small Disturbance Potential Flow

16.100 2002 2

Implications:

) , (
1 1
) , , (
0

u
x
M y x u =
= =
V
u
V
u
M y x C
p
0
2 1 2
) , , (

0 =
M

What about the forces?

0
M
flow
0 =
M
flow
) , (
1
) , , (
0

p p
C M y x C =

Critical Mach Number

We can estimate the freestream Mach number at which the flow first accelerates
above 1 > M (locally) using the Prandtl-Glauert scaling and isentropic
relationships.

Recall from P-G:

If we have ) 0 ( =
M C
p
say from an incompressible panel solution, we could then
find
p
C anywhere on the airfoil for higher
M under the assumptions of P-G

(linearized flow, ) 1 <
M .

We can also use isentropic relationships:

+
+
=
=

1
) 1 (
2
1
1
) 1 (
2
1
1
2
) 1 (
2
1
2
2
2
2
M
M
M
C
p
p
M
C
p
p

The
p
C for 1 = M at a given
M is:

This critical freestream
M occurs when ). ( ) (
cr p cr p
M C M C
crit G P
=

This critical
M can be found graphically or can be solved for with a root-finding

method. Lets look at what happens graphically:

2
1
) 0 (
) (
=
=
M
M C
M C
p
p

On the airfoil
surface:
+
+
= = =

1
) 1 (
2
1
1
) 1 (
2
1
1
2
) , 1 (
1
2
2
M
M
M M C C
p p
crit

Critical Mach Number
16.100 2002 2

1. Find minimum
p
C at 0 =
M
2. Plot
G P
p
C
min
vs.
M
3. Plot
crit
p
C from isentropic relationships

p
C
min
p
C
crit
p
C
cr
M
M
Behavior of Isentropic Flow in Quasi-1D

Recall cons. of mass:
. const uA =

Consider a perturbation in the area

0
. . .
) )( )( (
= + +
+ + + + =
+ + + =
udA Adu uAd
T O H udA Adu uA d uA
dA A du u d uA

Or

(1)

Similar manipulations to x-momentum gives:

(2)

Also, since the flow is assumed isentropic, we can use the definition of the speed
of sound to relate d and dp :

=
= .
2
const s
p
a

A A
u u
p p
L
L
L
L
=
=
=
=

L
x
R
x
dA A A
du u u
dp p p
d
L R
L R
L R
L R
+ =
+ =
+ =
+ =

0 = + udu dp
d a dp
2
=
0 = + +
A
dA
u
du d
Since we have assumed . const s = in the flow (3)


16.10 2002 2
We are interested in how the flow properties change when the area changes. So,
we use (2) and (3) to eliminate terms from (1). For example, lets determine how
u changes with A:

0 ) 1 (
0
0
0
2
2
2
= +
= + +
= + +
= + +
A
dA
u
du
M
A
dA
u
du
a
udu
A
dA
u
du
a
dp
A
dA
u
du d

When 1 < M : 0 >
u
du
when 0 <
A
dA

u increases when A decreases
0 <
u
du
when 0 >
A
dA

u decreases when A increases
This is what we expect from our understanding of incompressible flow.

When 1 > M : 0 >
u
du
when 0 >
A
dA

u increases when A increases!
0 <
u
du
when 0 <
A
dA

u decreases when A decreases!
This is very different from incompressible flow.

Whats happening for 1 > M ? Clearly, u must behave the opposite of A
regardless of the Mach number. So, what must be happening is that changes
more rapidly than A for 1 > M and, thus u behaves opposite of what we expect
from subsonic flow behavior. Lets check this:
A
dA
M u
dp
A
dA
M u
du
2 2
2
1
1
1
1
=

=

A
dA
M u
du
2
1
1
=
A
dA
M u
dp
2 2
1
1


16.10 2002 3
A
dA
M u
d a
2 2
2
1
1

These results show that, regardless of whether 1 > M or 1 < M , dp and d have
the same sign as dA:

Recall conservation of mass:

0 = + +
A
dA
u
du d

0
1
1
1
2 2
2
= +
A
dA
A
dA
M
dA
M
M

A
dA
M u
d a
2 2
2
1
1

Thus we find that:
< , : 1 p M when A
, p when A
> , : 1 p M when A
, p when A
Which are the opposite of how u behaves. It is also useful to consider the
magnitudes of the different changes:

A
dA
M
M d
2
2
1
=

, p when A
, p when A
A
dA
M u
dp
2 2
1
1

A
dA
M
M d
2
2
1
=


16.10 2002 4
0 = + +
A
dA
u
du d

Flow low , M changes are small compared to u changes. But for , 1 > M
changes more rapidly.
Also, of interest is that , ,dp d and du become extremely large as 1 M .
In fact, this requires that 1 = M must occur at a minimum in the area where
0 = dA .
Lets consider a converging-diverging duct:

When
bach
p is only a little less than
0
p , the flow will be subsonic:

As
bach
p is lowered from
o
p , we will eventually hit the
bach
p at which 1 = M at
the throat.
What happens if
bach
p is lowered further?
There is one other isentropic flow through this geometry which occurs when
bach
p is very low. In this situation, the flow will become supersonic in the divergent
section:

0
p
bach
p
M
1
0
decreasing
bach
p
0
1
1
1
2 2
2
= +
A
dA
A
dA
M A
dA
M
M

16.10 2002 5

M
1
0
Supersonic
isentropic flow
(very low
bach
p )
Subsonic
isentropic flows
Derivation of Sound Wave Properties

Assume:
Sound wave creates small disturbances in an isentropic manner.

0 ) (
0
0
= +
u u dx x
dt
d

l
l

0 ) ( ) (
0
= + +
+ +

du d u dx dx d
dt
d s
s
x
x

l
l

0 ) (
0 0
= +
+

du d dx dx d
dt
d s
x

l

[ ]
0 ) (
0 ) (
= +
= +
du d
dt
dx
d
du d x d
dt
d
s
s

0 = du d du a d

Sound wave
propagating to
right
a
s
x
l
p
u 0 =

dp p
du
d
+
+
= ) (x
s
s
x x
x x d
>
< +
,
,

Similar for ) ( ), ( x p x u
Mass
Constant
time
Higher
order
du a d =
0

16.100 2002 2
Isentropic disturbances

Constant entropy disturbances for perfect, ideal gases satisfy:
. const
p
=

Conservation of Momentum
= + + +
l
l
0
0
2 2
0 ) ( ) ( p u p u udx
dt
d

0 ) ( = + + dp p p dua

Summarizing:
Mass: du ad = (1)
Isentropic:
d
p
dp = (2)
Momentum: dp adu = (3)

Combining ) 3 ( ) 1 ( * a gives:
dp d a =
2

Then, using (2) gives:

0
2
2
=
d
p
a
d
p
d a

Since 0 d , then . Weve just derived the speed of sound for an

d
p
dp
d
p
dp
d
p
dp p
p
d
dp p
+ = +
+ = +
+
=
+
=
+
+
1 1
1 1
) (
d
p
dp =
Higher order
terms eliminated
dp adu =
p
a =
2


16.100 2002 3
Ideal, perfect gas.

Note: without assuming ideal, perfect gas, the general result is
.
2
const s
p
a
=

One other thing of interest: Suppose the sound wave caused a change in
pressure, dp . Then, the change in velocity is:
du dp
a
du =
1
has same sign as dp
Waves in 1-D Compressible Flow

Imagine we have a steady 1-D compressible flow. Then suppose a small
disturbance occurs at a location
o
x x = . This disturbance will cause waves to
propagate away from the source. Suppose that the flow velocity were u and the
speed of sound a . Then 3 waves exist:

(1) Downstream propagating acoustic wave:
Speed: a u +
This is an isentropic disturbance (and what is commonly called a sound
wave).

(2)Upstream propagating acoustic wave:
Speed: a u
Again, this is an isentropic disturbance (and is commonly called a sound
wave).

(3)Entropy wave:
Speed: u This wave is just a change in entropy.

When 1 > =
a
u
M , all the waves propagate in the downstream direction:

For example, if 0 > > a u , then

When 1 < =
a
u
M , the slow acoustic wave propagates against the stream:
For example, if 0 > > u a then

In supersonic flow, this means that the presence of a disturbance cannot be felt
upstream while in a subsonic flow it can be:
1 >
M
0 , , > + a u a u u
Supersonic flow
Waves travel only
downstream
0
0 ,
<
> +
a u
a u u

Subsonic flow
Waves travel up and
downstream
Shock
wave
Disturbance not
felt upstream
Normal Shock Waves

In our quasi-1D flows, shocks can occur from a supersonic-to-subsonic state.
These shocks are discontinuous in our inviscid flow model (recall that shocks are
very thin and their thickness scales with
e
R
1
):

The shock jump relationships come from the conservation equations we have
seen before:

Since the jump is discontinuous (i.e. it has zero thickness)
0 = = = dA A A A
R L

Here are some important things to know about shock waves from these
relationships:
Mathematically, shocks exist which jump from subsonic-to-supersonic flow.
However, these shocks can be shown to violate the 2
nd
Law ( 0 < s ).
Only shocks which jump from supersonic-to-subsonic states satisfy the 2
nd

Law. The Mach number downstream of shocks is given by :

And it can be shown that 0 > s .
1
M

Throat
L
M
R
M
shock Upstream Mach: 1 >
L
M
Downstream Mach: 1 <
R
M
R o R R L o L L
A
A
R R R R L L L L
R R R L L L
A h u A h u
pdA A p u A p u
A U A U
R L
R
L

=
+ = +
=
) ( ) (
2 2
R L
o R R L L
R R R L L L
R R L L
h u h u
p u p u
u u

=
+ = +
=
0
2 2

Ranhine-Hugoniot
Shock J ump
Relationships
) 1 (
2
1
) 1 (
2
1
1
2
2
2

+
=

L
L
R
M
M
M where 1 >
L
M
Normal Shock Waves
16.100 2002 2
The stagnation enthalpy (and therefore the total temperature) is constant
through a shock (shocks are adiabatic).

Total pressure decreases through a shock (this is a direct result of the entropy
Increasing while .
0
const T = ):

R L
h h
0 0
= or, equivalently,

R L
T T
0 0
=
R s s
L R
L
R
e
p
p
) (
0
0
=

Propagation of Disturbances By a Moving Object

Consider an object moving at speed V :

Suppose that the atmospheric speed of sound is a . The body emits sound
waves as it travels through the atmosphere and these wave propagate away
from the body at speed a in an isotropic manner. For example, consider a
stationary sound source (like a lecturer or a stereo speaker):

1 1
at R = &
2 2
at R =
1
Vt
3
t
2
t
1
t 0 = t V
Sound waves at
time t
2
R
1
R
a
a
Sound wave emitted
at
2
t t =
Sound wave emitted
at
1
t t =
Linearized Compressible Potential Flow Governing Equation

Recall the 2-D full potential eqn is:

0
2
) (
1
1 ) (
1
1
2
2
2
2
2
=

xy y x yy y xx x
a a a

Where [ ]
2 2 2 2
) ( ) (
1
y x o
a a
=
As you saw, for small perturbations to a uniform flow, the linearized form of the
equation was:

Where
y
x J y j x
i
v
u u
u u u
=
= + =
+ =

v v
v
v
v
v

This equation is valid for both 1 <
M and 1 >
M . Note, though, it is not correct

for 1
M or for
M large, say greater than about 2 or so.

So what happens to the linearized potential equation for : 1 >
M

It also turns out that 0

) 1 (
2
= +
yy xx
M is much easier to solve when 1 >
M .
Define
0

1
2
2
=
=

yy xx
M

Then,
) (
) , (
= y x where y x =
is a solution to the linearized potential. To see this:
0

) 1 (
2
= +
yy xx
M
=
perturbation
potential
perturbation
velocity
Subsonic Flow
0 1
1
2
>
<
M
M

Elliptic PDE
(Laplaces eqn)
Supersonic
0 1
1
2
<
>
M
M

Hyperbolic PDE
(wave eqn)

16.100 2002 2

= =
= =

y y
x x
d
d

Similarly,

=
=

2
yy
xx

This means that
is constant for lines described by . const y x = = For

example, consider an airfoil:

So, the values of
will be determined by the boundary conditions on the

surface. Recall:
0 = n u
v v
on boundary

For linearized flow, this b.c. reduces to:

Also, note that:

u v v
u
y
x

= = =
= =

on boundary
=

2 2 2
yy xx
0 =
) (
is solution to linearized
potential!
1 >
V
1
.
2
=
= =
=
M
const y x
const

1
1
2
M
dx
dy

n
v
dy
dx

= V v
=
V
u
)


16.100 2002 3

This is very useful because the linearized pressure coefficient is:

=
V
u
V
p p
C
p
2
2
1
2

on boundary!

0 >
p
C for 0 > (i.e. a compression)
0 <
p
C for 0 < (i.e. an expansion)
Using linear potential theory, lets calculate the lift and drag coefficients for a flat
plate at
M , ,

Find
d
C C &
l
:

=
=
=
1
2
1
2
2
2
M
C
M
C
p
u
p
u
l
l

Then, the result is a force in the y - direction:
( )
( )
u y
u
u
p p
y
f
p p y
p p
c
o
y
C C
c V
F
C
C C C V F
dx C C V F
=
l
l
l
2
2
2
2
1
2
1
2
1

Finally, we need to rotate this into lift and drag directions:

sin cos
cos sin
y x
y x
f f d
f f
C C C
C C C
= =
+ =
l

But, in our case, 1 & 0 << =
x
f
C

1
2
2
M
C
p
y
x
pu
P
C
C
l
p
C

1 >
y
y
f d
f
C C
C C
l


16.100 2002 4

M
0
d
C
(linear
potential
theory)
0
Implications of Linearized Supersonic Flow on Airfoil Lift & Drag

To begin, we will divide the airfoil geometry into camber and thickness
distributions:

) (
2
1
) ( ) (
) (
2
1
) ( ) (
sin cos
x x y x y
x x y x y
j U i u U U
c
c u

=
+ =
+ =

l
v v v

Note:
The x -axis, which is the axis defined through the leading edge and trailing edge
points, is used to define the freestream angle of attack. The chord line is the line
connecting . .e i to . .e t and, the chord length is the distance between these points.

To calculate the lift and drag, we need to integrate the pressure forces around
the airfoil.
=
airfoil
ds n p F
v
v
when n
v
points out of the airfoil surface.

Now, we begin specializing this formula for the assumptions of linearized flow,
i.e.

(small thickness)

(small camber)

(small )

The normal on the upper surface is
y
i
x

1
) (x y
c

) (x y
u

) (x
) (x y
l

1
1
1
<<
<<
<<
c
y
c
c

16.100 2002 2

j
ds
dx
i
ds
dy
n
j
ds
dy
i
ds
dx
t
u
u
u
u
v v
v
v v v
+ =
+ =

But, since we have thin airfoils dx ds , thus j i
dx
dy
n
u
v v
v
l
1 + =
Similarly, on the lower surface j i
dx
dy
n
v v
v
l
l
1 =
Thus, the force may be written:
( )

+
+ =
=
c
u
c
u
u
c c
u
u
c c
u u
dx p p j dx
dx
dy
p
dx
dy
p i F
dy j i
dx
dy
p dx j i
dx
dy
p F
dx n p dx n p F
0 0
0 0
0 0
l
l
l
l
l
l
l
v v v
v v v v v
v v
v

The drag component is in the freestream direction:
( )
+ =
+ =

j F i F D
j i F
u
u
F D
v v v v
v v v
v
v
sin cos

Similarly, the lift is:
( )
j F i F
j i F L
v v v v
v v v
+ =
+ =
cos sin

y
x
u n
v
u t
v
) (x y
u

1

( )

+
=

c
u
c
u
dx p p dx
dx
dy
p
dx
dy
p D
0 0
0
l
l
l

1
( )

+
=

c c
u
u
u
dx p p dx
dx
dy
p
dx
dy
p L
0 0
l
l
l


16.100 2002 3

Or, manipulating the D L & slightly:

Then substituting in for the
2
2
1
u
p p
c
p
, we can write this as:

dx
dx
dy
c
dx
dy
c
c
c
dx
dx
dy
c
dx
dy
c
c
c
c
u
p p
p
u
c
p d
u
u
+ =
=

0
0
1 1
1
1

l
l
l
l
l

As shown in Anderson

1
2
2
M
c
p

1
2
1
2
2
2
+
=
M
dx
dy
c
M
dx
dy
c
p
u
p
u
l
l

=
dx
dy
u
u

Lower surface

+ =
dx
dy
l
l

Next, lets substitute
l
p p
c and c
u
into
l
c :
+ =
=

c
u
u
c
u
u
dx
dx
dy
p
dx
dy
p L
dx
dx
dy
p
dx
dy
p D
0
0
1 1

l
l
l
l

In supersonic linearized
flow where is the flow
direction relative to the
freestream (and
assuming an isolated
airfoil).

Upper surface

u
dy
dx
u

dx

l
dy
l


16.100 2002 4
c
u
c
c
u
u
c
u
c dx
dx
dy
dx
dx
dy
M c
dx
dx
dy
dx
dy
M c
c
dx
dx
dy
dx
dy
M
dx
dy
dx
dy
M
c
c
0 0 2
0 2
0 2 2
2
1
2
2
1
2
1
1
2
1
1
2 1

l
l
l
l l
l

But 0 0 0 ) 0 ( ) (
0
= = =
l l
l
y c y dx
dx
dy
c

And, similarly,
=
c
u
dx
dx
dy
0
0

Important result!!

l
c is linear with
but note the slope is different than subsonic case
2
1
2
M
d
dc
l
.
l
c does not depend on camber! All the y -dependence has disappeared in this
result. Thus,
l
c also does not depend on thickness.

Now, lets look at
d
c :
dx
dx
dy
dx
dy
M c
c
dx
dx
dy
dx
dy
dx
dy
dx
dy
M c
dx
dx
dy
dx
dy
M c
c
c
u
d
c
u u
c
u
d
0
2
2 2
2
0
2
2 2
2
0
2 2
2
2
1
2
2 2
1
2
1
2

l
l l
l

small small
1
4
2
M
c

l

This term will
integrate to 0

16.100 2002 5

Note:
d
c is >0 unless 0 =
and airfoil is a plate ) 0 ( = = = =

l l
y y const y y
u u
.
A little manipulation gives another form dependent on the camber and thickness:

Thus, for a given
l
c , the lowest
d
c occurs when the airfoil is a flat plate
) 0 ( = =
c
y !!
c
u
d
dx
dx
dy
dx
dy
c
M M
c
0
2 2
2 2
2
0
1
1
2
1
4
l

( )

c c
c d
dx
dx
d
c
M
dx dy
c
M M
c
0
2
2
2
0 2 2
2
1
1
1 1
1
4
1
4

Oblique Shock Waves

Heres a quick refresher on oblique shock waves. We start with the oblique shock
as shown below:
(1) (2)

Also, the specific flow quantities above are:
v : flowspeed M : Mach number =
a
v

u : normal velocity to shock
n
M : Normal Mach
a
u
= #
w: tangential velocity to shock
t
M : Tangential Mach
a
w
= #
The next step is to apply the 2-D Euler equations to derive jump conditions.

Lets consider the following (well-chosen) control volume across the shock:

Where: d a & are parallel to shock
e c f b , , , are parallel to local flow
Apply conservation of mass:

=
s
ds n V 0
v
v

But 0 = n V
v
v
on , & , , e c f b thus:

= +
= +
a d
s s
a d
ds n V ds n V
ds n V ds n V
0
0
2 2 1 1
v
v
v
v
v
v
v
v

1
u
2
u
y
x
1 1
,
t
M w
1 1
,M v
1 1
,
n
M u

2 2
,M v
2 2
,
n
M u
2 2
,
t
M w
( )
1
: upstream flow
condition
( )
2
: downstream flow
condition
: angle of shock wave
. . . t r w upstream flow
: deflection angle of flow
b c
a
f e
d
g a
n n
v v
=
1
v
2
v
s d
n n
v v
+ =
s
n
v

Control volume s
s
t
v

Shock wave
Oblique Shock Waves
16.100 2002 2
0
2 2 1 1
= +

a d
ds d u ds u

2 1
A A

But,
2 1
A A = since all lines of control volume edges f e c b , , , are parallel.

The next equation well look at is tangential momentum.

= +
=
e
s
c
s
b f
s s
s s
s
ds t n p ds t n p
ds t n p ds t n p A u w A u w
ds t n p ds n V w
v
v
v
v
v
v
v
v
v
v v
v
2 2 2 2 1 1 1 1

Plugging into the pressure terms:

= +
b f
s f s b
ds t n p ds t n p A u w A u w
v
v
v
v
1 1 2 2 2 2 1 1 1 1

But
e c f b
n n n n
v v v v
= = &

c e
s e s c
ds t n p ds t n p
v
v
v
v
2 2

2 2 2 2 1 1 1 1
A w u A w u =
Using const uA = .

Tangential velocity is unchanged across a shock wave!
The last two equations (see Anderson for more) give:

These equations can be solved and results are displayed in graph and tables in
Anderson.
2 2 1 1
u u =
2 1
w w =
Normal momentum :
2
2 2 2
2
1 1 1
u p u p + = +
Energy:
2
2 2
2
1 1
2
1
2
1
u h u h + = +
Example:
Find
2
&M ?
5
1
= M
2
M

= 25
ramp
Prandtl-Meyer Expansion Waves

When a supersonic flow is turned around a corner, an expansion fan occurs
producing a higher speed, lower pressure, etc. in an isentropic process.

J ust as we saw with shock waves, if we apply conservation of mass and
momentum across a single wave, the tangential velocity is unchanged.
Unlike a shock wave, an expansion wave is isentropic.

So lets pick out a single wave:

From law of sines:

+
=
+

d
V
dV V
2
sin
2
sin

Using
M
d
1
sin & 0 = , we can find :

1
1
> M
Forward
Mach line
fan
Rearward
Mach line 1

2

waves
V
u
du
V
w
dv V +
d
Mach wave
V
dV
M d 1
2
=
Prandtl-Meyer Expansion Waves

16.100 2002 2
Next,
a
da
M
dM
v
dv
M v
a
+ = =
Using adiabatic relationships, we can re-write:

+
=

+ = = =
M
dM
M
v
dv
dM M M
a
da
M
T
T
a
a
RT
RT
c o
o
2
1
2
2
2
1
1
1
1
1
2
1
1
1

Finally, integrating d we find:
) ( ) (
1 2
M v M v =
Where
1 tan ) 1 (
1
1
tan
1
1
) (
2 1 2 1

+
=

M M M v

Prandtl-Meyer function

Problem: Estimate the rates of the pressure inside the pitot probe to the
freestream static pressure,
p p
a
.

M
dM
M
M
d
2
2
2
1
1
1
+

=

3 =
M
15 Pressure
in pitot
probe
a
p =
Computational Methods for the Euler Equations

Before discussing the Euler Equations and computational methods for them, lets
look at what weve learned so far:

Method Assumptions/Flow type
2-D panel 2-D, Incompressible, Irrotational Inviscid

Vortex lattice 3-D, Incompressible, Irrotational Inviscid, Small
disturbance

Potential method 3-D, Subsonic compressible, Irrotational, Inviscid,
Prandtl-Glauert Small disturbance

Euler CFD 3-D, Compressible (no
M limit), Rotational,
Shocks, Inviscid

The only major effect missing after this week will be viscous-related effects.

2-D Euler Equations in Integral Form

Consider an arbitrary area (i.e. a fixed control volume) through which flows a
compressible inviscid flow:

n
v
outward pointing normal (unit length)
dS elemental (differential) surface length

j dx i dy dS n
v v
v
=

Note: Path around surface is taken so that interior of control volume is on left.
y
x
c
dS
C
n
v
dS n
v

dy
dx
dS
16.100 2002 2
Conservation of Mass
C in mass of
dA
dt
d
change of rate
dA C in Mass
C of out
flow mass of rate
C in mass of
change of rate
C
C
=
=
=
0
where fluid of destiny

Now, the rate of mass flowing out of C :
Mass flow out of
= =
C
u dS n u C

v v v
velocity vector

Conservation of x-momentum

Recall that: total rate of change momentum = forces

For x-momentum this gives:
=
C of out
momflow x of rate
C in momentum x
of change of rate
Forces in x-direction

= +
C C
dS n u u udA
dt
d

v v
Forces in x-direction
Now, looking closer at x-forces, for an inviscid compressible flow we only have
pressure (ignoring gravity).

Recall pressure acts normal to the surface

=
C
dS i n p x in s Force
v
v

= +
C C
dS n u dA
dt
d
0
v v

dS
dS n p
v

Into surface
Normal to surface
Gives x-direction
16.100 2002 3

Conservation of y-momentum

This follows exactly the same as the x-momentum:

Conservation of Energy

Recalling your thermodynamics:
C to
added heat
C in fluid
on done work
C in energy of
change of rate total

For the Euler equations, we ignore the possibility of heat addition.
C of out flow
energy of rate
C in energy
of change of rate
C in energy of

The total energy of the fluid is:
) (
2
1
2 2
v u e E + + =

Note: T c e
v
= where
v
c specific heat at constant volume

Static temperature

So,

+ =
C C
dS n u E EdA
dt
d
C in energy of

v v

The work done on the fluid is through pressure forces and is equal to the
pressure forces multiplied by (i.e. acting in) the velocity direction:
( ) ( )
=
C
dS u n p work
v v

= +
C C C
dS i n p dS n u u udA
dt
d

v
v v v

= +
C C C
dS j n p dS n u v vdA
dt
d

v
v v v

Total
energy
Internal
energy
Kinetic
energy
Pressure force
16.100 2002 4

Summary of 2-D Euler Equations

= +
= +
= +
= +
C C C
C C C
C C C
C C
dS n n p dS n u E EdA
dt
d
dS j n p dS n u v vdA
dt
d
dS i n p dS n u u udA
dt
d
dS n u dA
dt
d

v v v v
v
v v v
v
v v v
v v
0

These are often written very compactly as:
( )

= + +
c sc
ds n j G i F UdA
dt
d
0
v
v v

E
v
u
U
uH
uv
p u
u
F
vH
p v
vu
v
G
p
E enthalpy total H +
Ideal gas:
+ = = ) (
2
1
) 1 (
2 2
v u E RT p
A Finite Volume Scheme for the 2-D Euler Eqns.

Heres the basic idea:
(1) Divide up (i.e. discretize) the domain into simple geometric shapes (triangles
and quads)

= +
C C C
dS u n p dS n u E EdA
dt
d

v v v v

Conservative
state vector
Flux vector
for x-direction

Flux vector for
y-direction
16.100 2002 5

Looking at this small region:

Cell 0 is surrounded by cells 1, 2, & 3.
i.e. cell 0 has 3 neighbors: cell 1, 2, & 3.

Nearest neighbors

9
0
10
8
2 1
7
6
5
4
3
11
12
13
16.100 2002 6
(2) Decide how to place the unknowns in the grid.
(a) Cell-centered: cell-average values of the conservative state vector are
stored for each cell.
(b) Node-based: point values of the conservative state vector are stored at
each node.
The debate still rages about which of these options is best. We will look at cell-
centered schemes because these are easiest (although not necessarily the
best). Also, they are very widely used in the aerospace industry.

(3) Approximate the 2-D integral Euler equation on the grid to determine the
chosen unknowns.

16.100 2002 7

16.100 2002 8

16.100 2002 9

16.100 2002 10
Lets look in detail at step (3):

Cell-average unknowns:
( )
( )
( )

=
0
0
0
0
0
E
v
u
U

( )
( )
( )

=
1
1
1
1
1
E
v
u
U
... ..........
2
= U . ..........
3
= U

Specifically, we define
0
U as:
0
0
0
1
C
UdA
A
U where
0
0
0
0
cell of area A
cell C

Now, we apply conservation eqns:
( )

= + +
0 0
0
C C
dS n j G i F UdA
dt
d
v
v v

The time-derivative term can be simplified a little:
=
0
0
0
C
dt
dU
A UdA
dt
d

The surface flux integral can also be simplified a little:
( ) ( )
( )
( )

+ +
+ +
+ = +
a
c
c
b
b
a C
dS n j G i F
dS n j G i F
dS n j G i F ds n j G i F
v
v v
v
v v
v
v v
v
v v
0

f b e
a c
d
1 2
0
3
Cells: 0,1, 2, 3
Nodes: a, b, c, d, e, f
16.100 2002 11
Combining these expressions:

Now, we make some approximations. Lets look at the surface integral from
b a :
( )
+
b
a
ds n j G i F
v
v v

The normal can easily be calculated since the face is a straight line between
nodes a & b. Recall, the unknowns are stored at all centers. So, what would be a
logical approximation for :
( )
= +
b
a
ab
dS n j G i F ???
v
v v

Option #1=
Option #2=
Note: Option #1 option #2 in general.

There is very little difference in practice between these options. Lets stick with:
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
ca ca
a
c
ca ca
bc bc
c
b
bc bc
ab ab ab
b
a
ab
s n j G G i F F dS n j G i F
S n j G G i F F dS n j G i F
S n j G G i F F dS n j G i F

+ + + = +

+ + + = +

+ + + = +
v
v v v v
v
v v
v
v v
v
v v
v
v v
3 0 3 0
2 0 2 0
1 0 1 0
2
1
2
1
2
1
2
1
2
1
2
1

Where

( )
( )
( )
( )
3 3
2 2
1 1
0 0
U F F
U F F
U F F
U F F

( )
( )
( )
( )
3 3
2 2
1 1
0 0
U G G
U G G
U G G
U G G

( ) ( )
( )

= + +
+ + + +
a
c
b
a
c
b
dS n j G i F
dS n j G i F dS n j G i F
dt
dU
A
0
0
0
v
v v
v
v v
v
v v

b
a c
2 1
0

3
ab
n
v

No approximations so far!
16.100 2002 12
Finally, we have to approximate
dt
dU
A
0
0
somehow. The simplest approach is
forward Euler:
0
0
0
= + + +
ca bc ab
dt
dU
A

And
n
ab
etc. are defined as:

( ) ( )
( )
( )
n n
n n
ab ab
n n n n n
ab
U F F
etc U F F
S n j G G i F F
1 1
0 0
1 0 1 0
.
2
1
2
1
+ + +
v
v v

For steady solution, basic procedure is to make a guess of U at 0 = t and then
iterate until the solution no longer changes. This is called time marching.

Question
What assumptions have we made in developing our 2-D Euler Equation Finite
Volume Method?
0
0
1
0
0
= + + +
+
n
ca
n
bc
n
ab
n n
t
U U
A
Where ( )
n
o
n
o
t U U and iteration n t n t
n
,
Structured vs. Unstructured Grids

The choice of whether to use a structured or an unstructured mesh is very
problem specific (as well as company/lab specific). The answer is one of
engineering judgement. Here are some of the issues:

(1) Complex geometry: unstructured grid generation is usually much faster
than structured grid generation. However, if the geometry is only slightly
modified from a previously existing geometry with a structured grid, then
structured grid generation can occur very rapidly.

For a problem which is different from previous applications:
Structured grid: man weeks to 1 man month
Unstructured: man hours to a few days
(2) Accuracy: For simpler problem such as airfoil (single element) or an
isolated wing, structured grids are generally more accurate per unknown
than unstructured. However, for more complex flows, the adaptivity
facilitated by an unstructured grid may allow more accurate solutions.

(3) Convergence CPU time: structured grid calculations usually take less
time than an unstructured grid calculation because, to date, the existing
algorithms are more efficient.

1 1 1 1
1 1
1 1 1 , 1 1 1
+
+
+ + + + +
ij i ij j i
j i ij j i
j i j i j i

( ) j i U , data stored in a 2-D array
( ) i U data stored in a 1-D
array

So, in order to calculate a residual for all, the neighboring states must be known.
Structured: neighbors found by adding/subtracting 1 from cell indices.

Unstructured: requires storage of cell-to-cell pointers.

More storage, slower execution of code.

12 14
11 13 15
10 0 2
9 1 3
8 6 4
7 5
Solution Convergence

Recall for our triangular grid finite volume scheme, the basic iterative scheme
looked like:

Suppose we are interested in the steady answer to our problem, i.e.
= =
=

E
v
u
U U
t
0 0
0

Initial guess of uniform flow answer as t
when 0 =
dt
dU

(steady-state)
0
1
= + + +
+
n
ca
n
bc
n
ab
n
i
n
i
i
i i i
t
U U
A
n
i
R residual of cell
Approximation of
( ) ds n j G i F
i i
c
v
v v
+

n
i
i
n
i
n
i
R
A
t
U U

=
+1

Update formula for cell i
to iteration 1 + n from
iteration n
We want

Full Elementary Aerodynamics Course by MIT

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Full Elementary Aerodynamics Course by MIT

Hochgeladen von

Copyright:

Verfügbare Formate

Sensitivity Analysis

Often in engineering analysis, we are not only interested in predicting the

Thus, under these nominal conditions:

These can often be more information by looking at percent or fractional changes:

in terms of v u, and their derivatives:

cos sin ( cos sin )

cos cos sin cos sin sin

sin cos ( sin cos )

sin sin cos sin cos

sin cos ( ) (cos sin )

Coordination Transformations for Strain & Stress Rates

terms (and enforcing

is freestream velocity (flight speed)

Similarly, M M . The non-dimensional coefficients for 2-D are defined:

Equations of Aircraft Motion

for small range of angle of attack by numerical

, cos sin tan

q (i.e. raise cruise velocity) Friction increases

V and is in x direction. Thus, drag is the force in

for fixed r . But as varies from 2 0 , it must be equal at

constant must be zero!

always, then clearly 0

V and is only a function of r , the

Viscous Flow: Stress Strain Relationship

is small compared to shearing strains (i.e.

= and this equation gives:

equal in terms of & u u only?

because flow is steady and incompressible. Also, since =

and substituting in for

, all other terms are small as Reincreases. Thus, we must

a is the atmospheric speed of sound at ground level to good approximation

a varies by only about ~15%.

M are small. So, in this case, we

M is larger than about 0.3, we need to consider matching of

M & Re in the tunnel.

M and the values of

M , trip boundary layers, ignore Re match at high Re (Re >10

Single Horseshoe Vortex Wing Model

Three-Dimensional Wall Effects

M flow about the airfoil.

M under the assumptions of P-G

M can be found graphically or can be solved for with a root-finding

Since we have assumed . const s = in the flow (3)

M . Note, though, it is not correct

M large, say greater than about 2 or so.

is constant for lines described by . const y x = = For

will be determined by the boundary conditions on the

, we can write this as:

but note the slope is different than subsonic case

and airfoil is a plate ) 0 ( = = = =

Das könnte Ihnen auch gefallen