# - ,:; p 2.0_- .,' .........._ ...

.

13.024 - Numerical Methods in Incompressible Fluid Mechanics

Lecture Notes - Version 3 By Jerome H. Milgram Spring 2003

INCOMPRESSIBLE FLUID MECHANICS BACKGROUND

Conservation of Mass, Continuity Equation
divV

= V· V = 0
=0

ou+ov +ow ox oy oz

Newtonian Dynamics, Navier-Stokes Equations

-

DV
Dt

=-

oV
at

_, _, 1 + (V· V)V = --VP+
p

vV V

2 _,

<

The total pressure, PT, is the sum of the dynamic pressure and the hydrostatic pressure, -p 9 z, where z is positive upwards. PT = P - pg z.

P is the dynamic pressure.

1

PARTICLE IMAGE VELOCIMETRY

Laser Sheet
Light
~~-------+~~

Fluid Flow Region

Camera

Laser

2

PIV Example

u and v can be measured, so ~~ and ~~are known.

OW

oz

---ax

au

in:

oy

If PIV is done on multiple planes inside a fluid domain, then ~~ is known

over the whole domain. At a rigid boundary, w can be found anywhere by:
W=

= 0 and, in principle, w

1

z

boundary

-dz

ow
OZ

Example In a domain bounded by 0 < z < 4, u = (3eZ - zez - 3) sinx and v over a range of x and in 0 < z < 2. In this sub-domain,
=

0

au = (3e ax

Z

-

ze" - 3) cos(x)
_

aw = -(3e OZ
w
=

Z

ze" - 3) cos(x)

rz ow dz io OZ
w = (Z

= - [3eZ

-

3 - ze'

+e

Z

-

1 - 3z] cos(x)

io

oW dz OZ

= - [4eZ -

4 - ze' - 3z] cos(x)

3

Velocity Field at z 12

= 0.4

10

:;:::.. :::z;.

:::;::.-

e::::
EC

EC

:;:::..

..:::::
<:: <:: <::

-

:>-

-

>:;.. :>;::.. :;.. :>-

8

:;:::.. :;:::..

:;::.. e:::: ;:::.. ;:::.. :;:::-..
:;::..

-=

e:::: e::::

6

:;:::.. :;:::..

-

-

-= -

<
EC

..:::::
<:: <:: <:
EC

4

:> :>

;:::..
;:::..

e::::
<: <: <:

>;::.. :>-

2

::::0-

-= -

>-

o
-2

~

>-

<:

-

==-

-

o

5
x

10

4

dw/dz and w(z) at x

60 50 40 30 20 10 0
-10

=1

-20 -30

0

0.5

1

1.5

2

2.5

3

3.5

4

z

5

A More Interesting PIV Example
Consider the following flow for z > 0 in a range of x and y. Of course, in an experiment you would not know the mathematical formulation. Rather you would just measure u and v over a set of (x,y,z) points.

u(x, y, z) = (3eO.1Z

-

3 cos z) sin y cos x
-

v(x, y, z) = (3eO.1Z

3 cos z) cosy

The x and y derivatives of the velocities can be computed numerically from the measurements. If the experiment were done well, they would have values according to the following formulae:

~~ = - (3eO.1z
-

-

3 cos z) sin y sin x
-

8v
By

=-

(3eO.1Z

3 cos

z)

sin y

Then, the continuity equation is:

Ow

az

----

au
ax

Bv By

The experimentally determined values would have the values given by:

8w

oz

= (3eO.1Z - 3 cos

z)

sin y(sin

x + 1)

Integrating ~~ from 0 to z at a prescribed value of (x,y) would give w(z) there. The values obtained would obey:
w

= (30 eO.1z - 30 - 3 sin z) sin

y (sin x

+ 1)

6

Flow in Plane at z= t

7 6 >- 5

'\'\\

rrrf\\\\\\ [{{{H\\\\ r »»
t 1///11\\,\,\'\

f

tr

4
3 2 1

o

c______----L-.J_-'---'----'-~____j___J_J..____L_l__...L___l.___L_~~_____l

-2

0

2

4

x

6

8

10

12

7

dw/dz and w(z) at x=t, y=1

90~---,~---.~---.-----.-----.-----.-----.-----,-----.----,
80

70

60

50

40

30

20

10

0 0

1

2

3

4

5

z

6

7

8

9

10

8

AVERAGED NAVIER-STOKES

EQUATIONS

a (Vat Vi) +"[RJ + v' + v -)

1 . \7 ] (-=- + v' = - P \7 (p + p') + v\72 (-=- + v' V -) V -)

Take Average of above equation:

ail + [(.... v' at v + -)

. \7 ] (V

-) + v'

= - 1\7 P

P

+ v \7 2-=V

Thus, the Reynolds-Averaged Equation is:

ail -=-=- 1 - + (V . \7)V + (v'· \7)v' = --"\7 - + v\72-=P V at
p

The" Reynolds Stress Term" is:

(v'· \7)v'

=i J

-( u'- + v'- + w'au' au' aU') + ax ay az -:( ,av' ,av' u-+v-+wax ay ~(aW' aw' u'-+v'-+w'ax ay ,aV') + az aW') az

k

9

THE PRESSURE EQUATION FOR AN INCOMPRESSIBLE FLUID Start with the Navier-Stokes Equation

8Y..... - + (V·

at

\7)V

.... = --\7P+ 1
p

v\7 V

2""

Take its divergence. Because \7 . Y

=

0, the only non-zero terms are: ..... .... 1 div (V· \7)V = --\7
p
2

P

Working out the details of the LHS and interchanging RHS results in:
--v

the LHS and the

In2p _
-

p

(8U)2 + (8V)2 + (8w)2 +28v8u 8w8u _ --+2--+2-ax ay az ax ay ax az

8wav 8y az

n2p _
y -

-p

{(8u)2 + (Bv.)2 + (Bw)2 +2avau +2-- +2aw av} Bu:8u -_ _-8y 8x ay Bz ax 8x 8z 8y 8z

The pressure, P, satisfies Poisson's Equation driven by products of the spatial derivatives of the velocity. This is different than the common Bernoulli Equations because here the flow can be unsteady and rotational.

10

The Vorticity equation

vorticity

=w -

curl iT

-

\7 x

V

-

av + (V. \7)V-. = --\7P -. 1
at
p

+ 1/\7 V

2 _,

Take the curl of this equation, term by term:

aw _, _, at + (V . \7)w + (w . \7)V = 1/\72w
-

Dsx:

at

= -(w.

_, 2 \7)V + 1/\7 w

The first term on the right hand side is the rotation and stretching of the vorticity by the non-uniform velocity field.

12

Inviscid Fluid Mechanics, Euler's Equation Set the viscosity, J1 and the kinematic viscosity, v to zero. Apply these "settings" to the Navier Stokes Equation.

-

DV
Dt

=

-+ (V· at
-+

av

V)V

-+

=

--VP

1

p

- + u- + v- + w- = --av av av -+u-+v-+w-at ax ay aw at aw ax aw ay av az

au at

au ax

au ay

au az

1aP

p ax
1aP

=---

p ay lap paz

- + u- + v- + w- = ---

aw az

13

Bernoulli Theorems for Inviscid Flow Theorem 1 - Irrotational Flow Vector Identity

Let For Irrotational

H

=

~(IVI)2

2

+P
p Vx

Flow,

V = V</>and

V=0

-V</>+ VH

a at

= V(- at + H) = 0

~~+ H = f(t)
The function f(t) V = V</>'. can be absorbed into </> letting </> </>' fL f(t)dt by = + 8</> . 8¢/ and

8t

=

at + f(t)

so,

8</>' H

at +
pgz.

=

0

Finally rename </>' ¢. as

a¢ + H=O at
8</> ~(V)2 +

Remember that the total pressure, PT

=Pp

at

2

+ PT + gz = 0

.

15

BV _, -{)t + (V·

\7)V

->

=

--\7P
p

1

{)t \7¢>

B

+ \7 H _,

V

_,

XW

=0

Vxw=\7H Streamlines and vortex lines are perpendicular to \7 H. Along either a streamline or a vortex line, H is a constant. So on anyone of these lines, H
=

-(V)
2

1 _,2

+ - = -(V) + - T + gz = constant
p 2 p .

PI_,

2

P

If the flow is both steady and irrotational, H is the same everywhere because \7 H = o.

16

Vorticity Dynamics and Kelvin's Circulation Theorem

Circulation =

r = f V . dr = is ta--\7P+v\7 P
1
2 ....

ds

DV Dt

....
=

V

Identity for an incompressible fluid: \72V= ---c\7 ....

XW

DV 1 -=--\7P-v\7xw Dt p

Following a closed material curve, ....

dr
dt

f ~~.dr = - f >P
-v f(\7 x tv) . dr

. dr - v j(V XIV)

. dr

Kelvin's Circulation Theorem

In an inviscid fluid, ~

=0

pathlines
Corollary: In an inviscid fluid with no circulation (such as starting from rest) the circulation remains zero.

17

Practical Implication In a high Reynolds number streaming flow, fluid which has not passed close to a boundary or a free surface has negligible vorticity. Therefore, in high Reynolds number streaming flows, vorticity is limited to boundary layers, separated zones, and wakes. Concept of Vortex Lines

vortex lines

This line represents the vorticity in this area which is the circulation around its boundary Each line represents circulation around an area which is the same as the vorticity inside the area.
= curl

ti:

V

-+

=

\7 x V

-+

Therefore: div zo = \7 . to

=0

The vortex field is solenoidal. Vortex lines are continuous. They can have curves and turns,but they cannot have ends in the fluid.

13

v
W/A./G-

r

r

19

z a: w Ien

a: w
:J

0 0

a:

20

r

r:

9

f

Potential Flows and Mostly Potential Flows

For an irrotational fluid \7 x

V=0

This means that there exists a velocity potential, ¢, such that,

For an incompressible fluid \7 . V = 0

Thus, \7 . (\7¢)

=

0

For a completely potential flow, the velocity potential satisfies Laplace's equation. For an incompressible flow that is "nearly" irrotational except in boundary layers and wakes, the flow outside these boundary layers and wakes is approximately described by a velocity potential that satisfies Laplace's equation.

21

Green Functions, Green's Theorem and Boundary Integral Equations The following development is for three-dimensional flows. The development is similar for two-dimensional flows except that two dimensional source functions are involved and the dimensionality of some integrals and associated constants are different. Green's Theorem If ¢ and 'l/J both satisfy Laplace's equation (\J2¢
=

0, \J2'l/J

=

0), then:

Green Functions A three-dimensional (~, 1], () space is considered with a "sink" at location (x, y, z). The "sink" has the velocity potential 'l/Js,

~=

1 J(x - ~)2 + (y - 1])2 + (z - ()2

=r

1

where: r = J(x - ~)2 + (y - 'r/)2

+ (z -

()2

Importantly, \J2'l/Js = 0 both for the differentiations done in (~, 'r/, () space as well as for the differentiations done in (x, y, z) space. The following development can be formulated either way and we will choose to differentiate over (~, 'r/, (). A Green Function, Gis: G = 'l/Js(x,

u, z,~, 'r/, () + 'l/Jr(x, y, z,~, n.()

where, \J2¢r = 0 in the fluid domain, and (x, y, z) is called the point P.

,:::: UI 0 L

22

If P is outside the fluid domain, the bracketed terms on the right hand side in

Green's Theorem are zero. However, if P is inside the fluid domain, '\J21jJ i- 0 at P. Then, if P is enclosed by a small sphere of radius e, which is excluded from the fluid domain, Green's Theorem applies in the modified fluid domain. However, now the integrals include an integral about the small sphere.

FLUID

11

sphere

1>-8 ds n

ec

1

is+sphere

r

[¢Be - eBan ¢] an
-1
f

ds

~,"7,( -

-0

=

-¢[P(x, y, z)]-2

47rf =

47r1>[P(x,y, x)]

j Is [4>~~ ~

G~~] ds~,",( = ~47r4>[P(x, y, z)]

If P(X, y, z) is on the boundary, the integral is not defined. However, if we replace the real boundary by one which has an infinitesmal hemisphere syrrounding P, the Green Function integral is zero because the functions have no singularities in the revised fluid domain.

1f
fA
hemisphere

S+hemisphere

oc a¢] [1>- - Gan an
y, z)]-2
f

ds ~,"7,( = 0
27rf

.

¢-a ds = -¢[P(x,
n

oc

-1

= 27r¢[P(x, y, x)]

j

Is [4>~~

~

G~~] ds~,",( = ~47r4>[P(x,y,z)1
23

Putting the preceding parts together, if a closed fluid domain of surface S is considered with ii being the outward normal vector (out of the fluid) and 'IjJ is taken as G with proper exclusion of the singular point of G when (x, y, z) is inside the domain or on its boundary,

J is [¢a

8G

-G

n

a n

8¢]

dS

=

{0 -2Jr¢(x, -47r¢(x,

(x, y, z) outside S y, x) (x, y, z) on S y, x) (x, y, z) inside S

The integral is over the closed area in (~, 'TI, (). When the singular point is on the surface, an infinitesimally small circle surrounding the singular point is excluded from the integral.

24

Example of method of solution
Generate integral equation on surface of an object in a uniform flow.

<u
Suppose uniform flow onto an object is known ~ is known. q,=-Ux+¢ Boundary condition: a~

an

=

0

'

- U~ - aA. l·n+~= an

0

'

J fs [¢~~

-

ctn n] = -27r¢ sas

d~

J fs¢~~ as + 27r¢ = J fsCui.
Solve for values of ¢ on boundary (panel methods). Then ¢ and ~ are known on boundary. Green's Theorem then gives ¢ in all space.

Right hand side is known in integral equation for ¢ on boundary.

25

.r rn h
I •-

00 f'" (J J ",

.
j

Interpretation of Boundary Integral Equation in terms of source and Dipole Layers

J isr [2_ 8¢ G _ 2_¢ 8G] 47f8n 47f 8n

as = { 0

¢(x,y,x)

(x, u, z) outside S (x,y,z) inside 3

"Inside S" means inside the fluid and outside "3" means outside the fluid. G is the potential of a "unit sink" and -8g / im is the potential of a unit dipole. The Green's Theorem integrals are integrals of sink distributions per unit area of (1/47f)8¢/8n over the object and of dipole distributions of strength per unit area of ¢/47f over the object. The sinks

Consider the effect of a unit sink. 1 Vn=2"' r influx

=

1 2 -47fT
r2

= 47f

Now, look at the small patch of area A on the surface:
pqtc~

B is the effect of the integrals on the remainder of the object. Call the sink strength per unit area (5. Total sink Strength on the patch is (5A. Net influx, based on the velocities is 2AVs.

2A Vs = 47f(5A

2Vs = 47f(5

26

is the jump in normal velocity. This must equal the normal velocity, 8¢/8n in the fluid at the boundary since V = 0 inside the object.

2Vs

-

an

= 47r(J

1 a¢ (J = --47ron

N ow consider an infinitesmal dipole patch of strength fL

Inside the infinistemally
V=--

thin dipole layer of thickness f, ¢ in fluid - ¢inside object
=

47rfL

f

47rfL

since:

¢inside object

= 0,
fL

= 47r¢in fluid

1

27

Kelvin-Neumann

Problem

z

u

y
x

28

The Kelvin-Neumann Problem

¢ is the perturbation

potential (does not include -U x).

The integral over 81, which is the part of the ship hull below the waterline is of the same form as a Green's theorem or "panel method" integral for any finite size body, except here the top is open.

¢ and C decay with distance from the ship fast enough for the integral
over 83 to vanish. The integral over 82, which is the free surface external to the ship special. We consider it here and call it ¢2. ¢2= -1
IS

4~

11 [O¢ - ¢-on dx dy OC] C~ On
If ~ [O¢ Coz
U; ~:W
¢-

Since n is a unit vector in the z direction on the mean free surface, ¢2 = -1

4~

OC] dx dy oz

On the mean free surface, ~ = and we choose G (the KelvinNeumann Green function) such that it satisfies the same boundary condition , 8C = _ u2 82C 8z g 8x2 Then, applying these boundary conditions, ¢2 becomes,
1U ¢2 = - 4~

2· [02¢ 9 I is2 G ox
2

2

-

¢ ox2

02C]

dx dy

¢2=--- 1

4~ g

U

I!

S2

oX

0 [O¢ C--¢ox

OC] dxdy
ox

29

Now, the integral over x can be done. The contributions at x = ±oo vanish so the result is:

(P2 = _!_ U {
471"g
Jfore

2

2 [ca¢ _ ¢aC] dy _ _!_ U { [Ca¢ _ ¢aG] dy ax ax 471" g Jaft . ax ax

The curve of the waterline is called C, with the part forward of the maximum beam called C] and the part aft of this is called Ca. Consider the integrals taken along C in the counterclockwise direction. Then, dy is positive on the forebody (Cj) and negative on the afterbody (Ca). ¢2 =

_!_ U ( [Ca¢ _ ¢ aC] dy
471"g Jc ax ax

2

If v is the vector in the horizontal plane that is perpendicular to the wa-

terline and pointed out of the fluid into the ship, and dl! is the differential of arc length along C, along the waterline dy = -vxd.€, so, ¢ = -1

J f [..G-o¢ 471" s, an

2 ac] d5 - --1 U [a¢ - ¢ac] vxdl! - ¢Can 471" g c ax ax

£

Now, consider a potential function, ¢' defined in the region bounded by 51 and 54. In other words, ¢' is some function of space such that \12¢'= 0 in the region where it is defined outside the actual fluid. The boundary condition we impose on ¢' on 54 is the same as the one we impose on ¢ on 52. On Sl we impose ¢' = cp. = - ~2 ~x{ 0 Sf

¥z

f)

In the fluid region,

pointed into the fluid on 51 and n' is a unit vector in the z direction on 54.
n' -n

=

Call the contribution to ¢' from the integral on 54 by ¢~.
¢~ =

_!_

J{ 471" JS

4

2 [Ga¢, - ¢,OG] dx dy = -_!_ U az az 471" g

J JS {

4

2 a2 [C ¢' - ¢,a C] dx dy ax2 ax2

30

Carrying out the integral over x gives,

¢'
4

=_

_!_ U
47r

2

9

1fore

r

[GO¢I _ ¢IOG] dy + _!_ U ox ax 41f 9

2

laft

r

[GO¢I _ ¢IOG] dy ax ax

1 0=-/ 41f

r lSI

[a¢1 G-+¢an'

OG] dS+-- 1 u2 r [O¢I G--¢an 41f 9 1c ax

aG] vxdf ax

To this, we add the equation for ¢ derived before:

The sum is:

¢ = -1
41f

/

r lSI

G [a¢ -

a¢,] 1 + - dS - -- U r G [o¢ on on' 41f 9 1c ox
2

a¢/] -ox

vxdf

4~ [~

+~]

is the source (actually it is a sink) strength

CT.

The normal derivative of 4~ ¢ jumps at the interface by the source strength, CT. The tangential derivative of ¢ is continuous across the interface because ¢ is continuous. ~ jumps across the interface by the jump in the normal derivative times nx. Therefore,

31

The Kelvin-Neumann Green Function The Kelvin Neumann Green Function, Gk(x, y, z) is the velocity potential for a source located at (a, b, c) and moving at speed U and which satisfies the linearized free surface boundary condition: G~x(x, This function is:

u, 0) + vG~ = 0,

v

= U2

9

--+-+ r rl
4v (7r/2 de 100
7r

1

1

ek(z+c)

cos[k(x

io

fo
eV(z+c) sec B
2

- a) cos e] cos[k(y k cos? v

e-

- b) sin eJ dk

+

4v fo7r/2

sin[v(x - a) sec e] cos[v(y - b) sin e sec2 eJ sec2 ede

where:

32

Source Only and Dipole Only Distributions

• (XI ~I~)
HYPOTHETICAL FLUID DOMAIN REAL FLUID DOMAIN

t/J'
nl
n= -n'

a an

a an'

¢ is the velocity potential in the fluid.
¢' is a function that satisfies \l2¢'

= 0 in the region inside the object.

For a field point in the fluid domain, the following equations apply:

¢ = _!_ 41r

J J G (a¢ + an' dB - _!_ J J (¢ - ¢') aGds a¢,) an 41r an

33

Suppose ¢/ is chosen as the harmonic function whose values on S are the same as ¢. The hypothetical interior flow would have the same tangential velocity on the object as the real outer flow. Then:
¢>

= _2_
41r

IIG

(D¢ + a¢,) as
an an'

This is a representation for ¢> in terms of surface sources only. This representation does not apply to lifting flows since they have wakes across which the potential jumps. Now consider the case for which ¢' is chosen such that on B, = _~. The normal velocity is continuous across the surface for this case. Then:

1n

¢> = _ _2_
41r

I I (¢> _

¢')

aG ds

an

This is a distribution of dipoles on the object surface S.

34

Green's Theorem in Two Dimensions
For two dimensional flow, the source potential is In r and the Green function becomes: C(x, y,

e, 'f]) = In r + 'lfJr(x, y, e, 'T])
FLUtO

The analysis proceeds exactly the same as in the 3D case. When a point P = (x, y), which is the local origin for In r, is outside the fluid domain,

[¢ac _ Ca¢] ds is an an
f

=0

When the point P is inside the fluid domain, Green's Theorem is valid in a domain in which the point P is excluded by a small circle, circle f surrounding it since '\72 =I- 0 at P.

Then: -.

oc a¢] a¢] ¢1 [¢- - C- ds + ts [8C - C- ds = 0 ar ar an an
circle

Here: -.

oc a¢] [¢-ar - G-ads = -27r¢(P) 1 r
circle

Therefore:

Is [¢a

oc
n

-G

a n

a¢]

ds

=

{0

(x, outside S 7r¢(x,y) (x,y) on S 27r¢(x, y) (x, y) inside S

y)

35

Sometimes, the two-dimensional Green function is taken as: G(x, Then,

u.t, 17) = -In

r

+ 'l/Jr(x, y,~, 17)
0 (x, y) outside S -7r¢(x, y) (x, y) on S { -27r¢(x, y) (x, y) inside S

t [¢
S

oG - G ds on an

O

¢]

=

36

FM6VJ

Force on a Vortex

/ /

-:

\

\

\

/

l I

\

n = -icosB - ksinB

u;

=

-~sinB

2~T

Wv

= ~cosB

P=-- P {( U--smB
2

2~T

r.)2
J.

+

ru P=-- p { --sInB+
2 7fT

(r-cosB )2 -U 2} - r )2}
27fT
(

2~T

27fT

FP

-- ;:r (-USinO + 4~r) (-icoso - kSinO) rdO t ~ pur . t pur
k
27r

c
o

Pn ds

---SIn
27fT

2

BrdB = -k--

~

2

Momentum influx = Min

Uin = iu . n = -U cos B

Ftotal =

Fe + FM

=

-kpUr

~

37

Lift on a Vortex in a Cylinder
When a vortex is in a uniform stream, to determine the lift force both the pressure force and the momentum influx into a circular cylinder must be considered. If the vortex is in a flow whose streamlines form a cylinder around it, there is no momentum influx so the pressure force is the complete force. A closed circle in a stream can be represented by a dipole.

o

>

The velocity potential of a 2D dipole is <Pd = A X 2+x z2. For the flow to make a circle of radius equal to 1 in a stream of speed U, A = U. the x- and z-directed speeds on the circle of radius 1 due to the dipole are:

Wd

= -2Uzx = -2UsinOcosO =

-Usin20

The speeds obn the circle due to the vortex are:
Uv

=

--sinO
27r

.r

Wv

= -cosO
27r

r

The pressure on the circle is: p = -~

[(u -Ucos2{}- ~Sin{}r + (-USin2{}+ ~COS{})' -U'] - f!_ [U + (i_) 2U cos 20 _ ur sin 0 + ur sin 0 cos 20 _ ur cos 0 sin 20j
2

2

27r

2_

2

7r

7r

7r

38

The vertical force, Fw is: Fw

= 10 Pn·

{27r

kdi)

A

= 10 -

(27r

P sin ()d()

Fw

pur _-

p~:
2IT 2IT

fo27r (1027r (
0

sirr' () + sin2

()

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

ao

1 1. - sin 2 () - - cos 2 2() - - sm 2 2()) d() 2 2
2 2

pur (-IT _ IT _ IT) = -pur

39

Example: Design of 2D Airfoil Mean Line using Dipoles and Vortices

t~",.
. ~= 0
~

.... 2 p_=

121

PU

- 2P(U

+ u)

2
~ -pUu

~ -.............. J] -~·~x-------~------~\~\-----~= 1

e = x/c
Design Condition: Ptop

x=

ce
2

'f]=y/c

= -1.0pU

e(1

- e)

Ut

=

1.0U~(1 - ~)

[Dipole StrengthJioil = I-l = 2.0U c

2 ~3) (~2- 3

J.Lwake

=

TUc

1.0

C

= In

r

= In [(x - xo)2 + (y - Yo)2J1/2 = ~ In [(x - xo)2 + (y - Yo)2]
8C) (8n 8C -t

8yo

8C) ( 8n

b=

8C 8yo

40

Consider the upper surface:

oc (ac) t = - ayo = (x an
¢(x, y) = 1
7r

y - Yo

- xo)2

+ (y

- Yo)2

roo Y - Yo 10 p,(xo) (x - xo)2 + (y _ Yo)2dxo

ay

The above analysis has an incorrect non-integrable singularity at x = cause a careful limiting analysis requiring \l2¢ = 0 was not done. However, another, and simpler, approach exists.

Xo

be-

A dipole represents a jump in the potential. Another way to achieve a potential jump is a vortex distribution.

In length dx.; vortex strength =l'(xo)dxo. Ut(x) = U + -2l'(x)

GCCCCGCC

l'(xo) is vorticity/unit-length.

41

Now, we can solve for the mean line shape of the airfoil For an arbitrarily defined pressure distribution, the integral for the slope can be done numerically. Here, for the particular pressure distribution given, we will solve for the slope analytically. Then the shape is found by integrating the slope, s(~). numerically. This will be done

42

C:\MATLAB6pl\work\foiltd.m
July 11, 2002

Page 1 9:24:51 AM

format compact x=0 : 0.01 : 1. 0; fac = -1.0/pi; s = fac.*(O.S - (LO-x) .*( x.*log«1.0-x+eps) ./(x+eps))+1.0)); h(l) = 0.0; for i = 2:101 h(i) = h(i-1) + (x(i)-x(i-1))*0.5*(s(i)+s(i-1)); end; fid = fopen('ht.dat', 'w'); for m = 1:101 fprintf(fid, '%6.2f %7.4f %7.4f\n' ,x(m) s(m), h(m)); end; fclose (fid); plot(x,s) ylabel (,Slope' ) xlabel (,\xi ') pause
I

plot(x,h) ylabel ('Camber' ) xlabel (,\xi ') title('Camber vs \xi') axis([O 1 0 0.066]) grid

43

j..
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22 0.23 0.24 0.25 0.26 0.27 0.28 0.29 0.30 0.31 0.32 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.40 0.41 0.42 0.43 0.44 0.45 0.46 0.47 0.48 0.49 0.50 0.51 0.52 0.53 0.54 0.55 0.56

f;J, oPt0.1592 0.1705 0.1771 0.1818 0.1853 0.1878 0.1895 0.1905 0.1909 0.1908 0.1903 0.1893 0.1879 0.1862 0.1842 0.1818 0.1792 0.1763 0.1731 0.1697 0.1661 0.1623 0.1583 0.1541 0.1497 0.1451 0.1405 0.1356 0.1306 0.1255 0.1203 0.1150 0.1095 0.1040 0.0983 0.0926 0.0868 0.0809 0.0749 0.0689 0.0628 0.0567 0.0505 0.0443 0.0380 0.0317 0.0254 0.0191 0.0127 0.0064 0.0000 -0.0064 -0.0127 -0.0191 -0.0254 -0.0317 -0.0380

IfE:~I6--ftT
0.0000 0.0016 0.0034 0.0052 0.0070 0.0089 0.0108 0.0127 0.0146 0.0165 0.0184 0.0203 0.0222 0.0240 0.0259 0.0277 0.0295 0.0313 0.0331 0.0348 0.0364 0.0381 0.0397 0.0413 0.0428 0.0442 0.0457 0.0471 0.0484 0.0497 0.0509 0.0521 0.0532 0.0543 0.0553 0.0562 0.0571 0.0580 0.0587 0.0595 0.0601 0.0607 0.0612 0.0617 0.0621 0.0625 0.0628 0.0630 0.0632 0.0632 0.0633 0.0632 0.0632 0.0630 0.0628 0.0625 0.0621

44

0.57 0.58 0.59 0.60 0.61 0.62 0.63 0.64 0.65 0.66 0.67 0.68 0.69 0.70 0.71 0.72 0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88 0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1. 00

-0.0443 -0.0505 -0.0567 -0.0628 -0.0689 -0.0749 -0.0809 -0.0868 -0.0926 -0.0983 -0.1040 -0.1095 -0.1150 -0.1203 -0.1255 -0.1306 -0.1356 -0.1405 -0.1451 -0.1497 -0.1541 -0.1583 -0.1623 -0.1661 -0.1697 -0.1731 -0.1763 -0.1792 -0.1818 -0.1842 -0.1862 -0.1879 -0.1893 -0.1903 -0.1908 -0.1909 -0.1905 -0.1895 -0.1878 -0.1853 -0.1818 -0.1771 -0.1705 -0.1592

0.0617 0.0612 0.0607 0.0601 0.0595 0.0587 0.0580 0.0571 0.0562 0.0553 0.0543 0.0532 0.0521 0.0509 0.0497 0.0484 0.0471 0.0457 0.0442 0.0428 0.0413 0.0397 0.0381 0.0364 0.0348 0.0331 0.0313 0.0295 0.0277 0.0259 0.0240 0.0222 0.0203 0.0184 0.0165 0.0146 0.0127 0.0108 0.0089 0.0070 0.0052 0.0034 0.0016 -0.0000

45

0.15

0.1

0.05

U5

0.. 0

<ll

0

-0.05

-0.1

-0.15

0.1

0.2

0.3

0.4

0.5
~

0.6

0.7

0.8

0.9

1

46

Cambervs

S

0.06

.

0.05

0.04
..c

.... Q)
E 0.03

o

en

0.02

0.01

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

S

47

0.2 w c, 0
.....J

0.1 0 -0.1 -0.2 0 0.2 0.4 0.6 0.8 1.0

en

..... s: ._
Q)

0.08 0.06 0.04 0.02 0.2 0.4 0.6 0.8 1.0

0>

.£:.

~

48

foiltd format compact x = 0 : 0.01 : 1.0; fac = -1. O/pi ; s = fac.*(0.5 - (1.0-x).*( x.*log((1.0-x+eps)./(x+eps))+1.0)); h(l) = 0.0; for i = 2:101 h(i) = h(i-1) + (x(i)-x(i-1))*0.5*(s(i)+s(i-1)); end; fid = fopen('ht.dat', 'w'); for m = 1:101 fprintf(fid, '%6.2f %7.4f %7.4f\n' ,x(m), sCm), hem)); end; fclose (fid); plot(x,s) ylabel C'Slope') xlabel C'\xi ') pause plot(x,h) ylabel('camber') xlabel C'\xi ') title('camber vs \xi') axis([O 1 0 0.066J) grid

49

foi 1tda . % version of foiltd with one less loop for computing speed improvement x = 0 : 0.01 : 1.0; fac = -1.0/pi; s = fac.*(O.5 - (1.0-x).*( x.*log((1.0-x+eps)./(x+eps))+1.0)); h(l) = 0.0; for i = 2:101 h(i) = h(i-1) + (x(i)-x(i-1))*0.5*(s(i)+s(i-1)); end; fid = fopen('hta.dat', IW q = [x;s;h]; fprintf(fid,'%6.2f %7.4f %7.4f\n',q); fclose (fid); plot(x,s) yl abel ('slope I) xlabel (,\xi I) pause
l );

plot(x,h) ylabel ('camber I) xlabel ('\xi I) title('camber vs \xi') axis([O 1 0 0.066]) grid

50

foiltdb % This verSlon uses even more vectorization and no "for" loops at all. % version of foiltd with one less loop for computing speed improvement x = 0 : 0.01 : 1.0; fac = -1.0/pi; 5 = fac.*(O.S - (1.0-x).*( x.*log((1.0-x+eps)./(x+eps))+1.0)); h(l) = 0.0; xd = [0 diff(x)]; % This is [0 x(2)-x(1) x(3)-x(2) ...] 55 = [0 s(1:end-1) + s(2:end) ] % This is [0 5(2)+5(1) 5(3)+5(2) h = O.S*xd .* 55; h = cumsum(h); % Each element is the sum of the ones bvefore it. fid = fopen('htb.dat', 'Wi); q = [x;s;h]; fprintf(fid, '%6.2f %7.4f %7.4f\n',q); fclose (fid); plot(x,s) ylabel ('Slope') xlabel (,\xi ') pause plot(x,h) ylabel('camber') xlabel (,\xi ') title('camber vs \xi') axis([O 1 0 0.066]) grid

51

C:\MATLAB6pl\work\foiltda.m July 11, 2002

Page 1 9:23:45 AM

% Version

of foiltd with one less loop for computing

speed improvement

x=0 : 0.01 : 1.0; fac = -1.0/pi; s = fac.*(0.5 - (1.0-x) .*( x.*log((1.0-x+eps) ./(x+eps))+1.0)); h(l) = 0.0; for i = 2:101 h(i) = h(i-1) + (x(i)-x(i-1))*0.5*(s(i)+s(i-1)); end; fid = fopen('hta.dat', 'w'); q = [x;s;h]; fprintf(fid, '%6.2f %7.4f %7.4f\n',q); fclose (fid); plot(x,s) ylabel (,Slope' ) xlabel (,\xi ') pause plot(x,h) ylabel ('Camber' ) xlabel (,\xi ') title('Camber vs \xi') axis([O 100.066]) grid

52

A:\foiltdb.m July 11, 2002

Page 1 9:35:36 AM

% This version uses even more vectorization and no "for" loops at all. % version of foiltd with one less loop for computing speed improvement x = 0 : 0.01 : 1.0; fac = -1.0/pi; s = fac.*(0.5 - (1.0-x) .*( x.*log«1.0-x+eps) ./(x+eps))+1.0)); h(l) = 0.0; xd = [0 diff(x)]; % This is [0 x(2)-x(1) x(3)-x(2) ...] ss = [0 s(1:end-1) + s(2:end) ] % This is [0 s(2)+s(1) s(3)+s(2) h = 0.5*xd .* ss; h = cumsum(h); % Each element is the sum of the ones bvefore it. fid = fopen('htb.dat', 'w'); q = [x;s;h]; fprintf(fid, '%6.2f %7.4f %7.4f\n',q); fclose (fid); plot(x,s) ylabel (,Slope' ) xlabel (,\xi ') pause plot(x,h) ylabel ('Camber' ) xlabel (,\xi ') title('Camber vs \xi') axis([O 1 0 0.066]) grid

53