436351 Thermofluids 2
Unit 1: Potential Flow
Jones M.B. & Ooi A.
March 10, 2003
Preliminaries
Main Text
Anderson J. D. Fundamentals of Aerodynamics, McGrawHill
Suggested Reading
Vallentine H. R. Applied hydrodynamics
Lamb Hydrodynamics
Streeter Fluid dynamics
Milne & Thomson Theoretical Aerodynamics
Contents
1 Introduction
. . . . . . . .
. . . . . . . .
or Lagrangian
. . . . . . . .
. . . . . . .
. . . . . . .
derivative)
. . . . . . .
.
.
.
.
7
. 7
. 8
. 9
. 10
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
17
17
18
19
19
23
26
27
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
31
31
31
32
34
35
38
41
43
44
47
47
48
49
53
56
59
60
61
64
69
70
.
.
.
.
.
.
.
4
6 The
6.1
6.2
6.3
6.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
73
73
75
76
77
7 Conformal Transformations
85
7.1 Conformal Transformation of velocities . . . . . . . . . . . . . . . . . 89
7.1.1 ExampleFlow over a Flat Plate . . . . . . . . . . . . . . . . . 90
7.2 Flow Over An Airfoil . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Chapter 1
Introduction
We wish to predict fluid motion, that is the flow patterns and associated forces they
create (eg. lift and drag). In many cases this is a difficult task and several different
approaches may be required.
Analytical
Simplify equations
Computational
Experimental
MB
OC
C
B
C
Here we will consider the analytical method known as classical hydrodynamics. This
involves the study of ideal fluids, by ideal we mean incompressible and frictionless
(inviscid), ie. =constant and = 0.
For example consider flow around a cylinder,
Classical model
Drag= 0
5
6
Actual flow
Vortex street
Better
agreement
Actual flow
thin wake
Chapter 2
Some Preliminary Concepts
2.1
(2.1)
(2.2)
If the velocity components are a function of space alone and are not a function of
time we have steady flow, ie V = V(x, y, z). Consider continuity
dA
Control volume V
8
The above is in integral form, we can also write it in differential form,
V =0 (steady flow)
(unsteady flow)
V = 0
steady or unsteady
u v w
+
+
= 0 .
x y
z
Note we are using the Cartesian coordinate system where
k
NB:
x
i
V = i u + j v + kw
2.2
Exercise 2.1: Show that for a threedimensional (sometimes written as [3]) flow
field, the mathematical equation for stream line can be written as
wdy vdz = 0
udz wdx = 0
vdx udy = 0
(2.3)
u is the velocity in the x direction and v is the velocity in the y direction and w
is the velocity in the z direction. For twodimensional (sometimes written as [2])
flows, only the third relatioship is important
vdx udy = 0
(2.4)
Exercise 2.2: Find the equations for streamlines and pathlines for the flow field
given by the following expressions
(a) V = x i y j
(b) V = x i + yt j
2.3
y
v+
A0
v
dy
A
v
v
v
dt +
dx +
dy
t
x
y
u+
u
u
u
dt +
dx +
dy
t
x
y
dx
10
Particle goes from A to A0 in time dt so acceleration in the xdirection is,
ax =
Note we use
D
Dt
Du Change in u velocity
=
Dt
dt
u u x
u y
=
+
+
t
x t
y t
u
u
u
+u
+v
=
t
x
y
=
+u
+v
Dt
t
x
y
similarly in ydirection
ay =
Dv
v
v
v
=
+u
+v
Dt
t
x
y
When we consider steady flow all derivatives of velocity with respect to time are
zero
Du
u
u
=u
+v
Dt
x
y
Dv
v
v
ay =
=u
+v
Dt
x
y
ax =
2.4
(2.5)
(2.6)
v+
v
v
dx
x
dx
x
its angular velocity is
1 =
v+
v
=
.
x
Similarly consider fluid element dy
v
dx
x
dx
11
y
u+
u
dy
y
dy
2
u
x
its angular velocity is
2 =
uu
u
dy
y
dy
u
=
.
y
Hence
v u
= 1 + 2
x y
and the above is called vorticity or rotation and is denoted by
=
v u
.
x y
(2.7)
12
Chapter 3
Momentum equations: Eulers
equations of motion
Forces on a particle
Consider [2] flow, frictionless fluid and ignore body forces (gravity).
p+
p
dy
y
dy
p
dx
p
p
Fx =pdy p +
dx dy
x
p
=
dxdy ,
x
similarly
Fy =
p
dxdy .
y
p+
p
dx
x
14
and the mass of the element is m = dxdy. Using Eqs. (2.5) and (2.5) the following
two equations are obtained
u
u
dxdy u
+v
=
x
y
v
v
dxdy u
+v
=
x
y
u
1 p
u
+v
=
x
y
x
v
1 p
v
u
+v
=
x
y
y
p
dxdy
x
p
dxdy
y
u
u
u
1 p
+u
+v
=
t
x
y
x
v
v
1 p
v
+u
+v
=
t
x
y
y
The above equations can be derived in other coordinate systems eg. streamline
curve linear coord.
R + R
Instantaneous streamlines
z } {
Vn
Vn
Vn
Vn +
dt +
ds +
dn
t
s
n
Vn
n
dn
Vs
ds
z } {
Vs
Vs
Vs
Vs +
dt +
ds +
dn
t
s
n
Vn = 0
Vs
1 p
Vs
+ Vs
=
t
s
s
2
Vn Vs
1 p
+
=
t
R
n
3.1
15
(3.1)
Proof:
Consider the steady x momentum equation in [2]
u
u
u
1 p
+v
=
x
y
x
u
u
1 p
dx + v dx =
dx
x
y
x
Using Eq. (2.4) on the second term on the LHS of the above equation gives
u
u
1 p
dx + u dy =
dx
x
y
x
u
u
1 p
u
dx +
dy =
dx
x
y
x
u
udu =
1 p
dx
x
1 2
1 p
du =
dx
2
x
(3.2)
(3.3)
p
p
dx +
dy
x
y
16
1
1
d V 2 = dp
2
dp = V dV
(3.4)
V1
1
1
p1 + V12 = p2 + V22
2
2
(3.5)
Exercise 3.1: The analysis above show that Bernoullis equation (Eq. (3.1)) is
valid only along a streamline. However, if the flow is inviscid and irrotational, it
can be shown that Eq. (3.1) is valid anywhere in the flow field. Prove that the
previous statement is true.
Chapter 4
Velocity Potential and Stream
function
4.1
The stream function is related to the rate at which fluid volume is streaming across
and elementary arc, ds.
For Incompressible [2] flow.
V
ds
I
or
V n
ds = 0
where V = V
ie.
This means
d = V cos()ds = V n
ds
I
d = 0
17
18
where a scalar point function and d is an exact differential ie.
d =
4.1.1
dx +
dy
x
y
d =
OBAO
d +
O
d +
B
d = 0
A
=(B O ) + (A B ) + (O A )
Along the streamline = 90o cos = 0
A
B
d = V cos ds = 0
Hence
along a streamline
A = B
4.1.2
19
4.1.3
d
= V cos = V n
ds
Say we move a small amount in the xdir
Have
y
V
Here ds = dx
20
then ds = dx and
Vn
= v
= v
x
Here ds = dy
x
then ds = dy and
Vn
=u
=u
y
Alternative derivation;
ds
dy
u
A
dx
v
Let d = flux crossing AB
d
{z}
flux in across AB
udy
{z}
vdx
{z}
flux in bottom
21
dx +
dy .
x
y
= v
x
=u
y
In polar coordinates
u
ur
Convention
ur = radial component
u = tangential component
Note ur and u correspond with u and v when = 0
dr
u
ds
dr
u
B
ur
A
ds
rd
r
d
ur
rd
x
d = flux across AB
d = ur rd u dr
but d =
dr +
d
r
22
Equate coefficients of dr and d
= u
r
1
= ur
r
If we had considered compressible flow
= v
= u
x
0
r
0
= u
= ur
y
0
r
0
where 0 is some reference density arbitraily chosen at some point in the flow. For
incompressible flow /0 = 1
Exercise 4.1: For the flow defined by the stream function = V y:
(a) Plot the streamlines.
(b) Find the x and y components of velocity at any point.
(c) Find the volume flow rate per unit width flowing between the streamlines
y = 1 and y = 2.
Exercise 4.2: An inviscid flow is bounded by a wavy wall at y = H and a plane
wall at y = 0. The stream function is
= A eky eky sin(kx) + By 2
(4.1)
4.2
23
It has been shown that along a streamline is constant. Therefore if we can determine the stream function we are then able to plot (or sketch) the streamlines for a
given flow.
Use Eulers equation of motion to determine a general equation for . Assuming steady flow we have;
u
u
1 p
+v
=
x
y
x
v
v
1 p
u
+v
=
.
x
y
y
(4.2)
(4.3)
yx
x x
x
x y
xy
(4.4)
(4.5)
0=u
x
v u
x y
+v
y
v u
x y
+
u v
+
x y
v u
x y
u v
+
x y
=0
hence
u
x
v u
+v
=0
y x y
D v u
OR
=0
Dt x y
v u
x y
(4.6)
n
o
v
The term x
u
is the vorticity which was defined in section 2.4
y
Note in streamline coordinates
=
Vs Vs
R
n
(4.7)
24
This means if we follow a fluid element its vorticity () does not change. Since
this is steady flow following a fluid element travelling on a streamline. Hence
streamlines are lines of constant as well as .
, v =
substituting into (2.7)
Since we have u =
y
x
=
= ( )
= 2
2 2
2
x2
y
2 = Laplacian operator
2
2
= 2+ 2
x
y
2 =
Equation (4.7) becomes
D 2
=0
Dt
Helmholtzs equation
This is effectively the Euler equation and continuity expressed in terms of the stream
function () for the case of incompressible [2] flow.
Exercise 4.4: Repeat the steps above and show that Helmholtzs equation is valid
for inviscid, incompressible AND unsteady flows.
u = ky
u
Here every streamline has the same vorticity and since remains constant
along a streamline then is constant everywhere.
2 = = const.
Poissons equation
(4.8)
25
U = U1
=0
Velocity profile as seen
by observer moving with wing
Now = constant along streamlines, therefore = 0 everywhere. With = 0
everywhere we have irrotational flow.
2 = 0
Laplace equation
(4.9)
This leads to what is know as potential flow and we can say if the flow is
irrotational ( = 0) the stream function () will satisfy the Laplace equation.
The beauty of the Laplace equation is that it is LINEAR. This means if we
have a series of simple flow solutions eg. 1 , 2 , 3 then the solution to more
complex flows can be obtained by superposition of the simple flows eg.
{z}
complicated flow
= 1 + 2 + 3 +

{z
}
simple flows
1 2
2
+ 2
+
= 2 2+
r
r r r
z 2
 {z }
In the case of polar cylindrical
= i + j + k
= V
i
j
=
x y
u v
w
=i
z
w
v
u w
v u
+j
+k
z
z
x
x y
26
hence for [2] flow
=k
4.3
v u
x y
,
 =
If we have a scalar point function, ie. = (x, y, z) then surfaces of constant will
form plates.
eg. surfaces of constant temp
surfaces of constant voltage
=grad
where
+j
+k
.
x
z
y
There are many vector fields in nature that can be defined this way;
= i
Vector field
Current flux
Heat flux
Gravitational force
Scalar function
Voltage potential
Temperature
Potential energy
The scalar function is called the potential for the vector field and if a vector field
possesses a potential it is called a conservative field.
Often in fluid flow (but not always) the velocity field is a vector point function
V(x, y, z) which possesses potential scalar function (x, y, z),
V(x, y, z) = (x, y, z) .
When this happens such a flow is called potential flow and is referred to as the
velocity potential.
4.4
27
The velocity potential is analogous to the stream function. The stream function is
related to the rate of flow across an small arc, ds, but the velocity potential, is
related to the rate of flow along ds.
Let d = V ds sin
Z B
then A B =
V sin ds
A
d
= V sin
or
ds
B
s
ds
Here ds = dx
x
then ds = dx and
V sin = u
=u
x
28
V
Here ds = dy
x
then ds = dy and
V sin = v
=v
y
Now V = i u + j v
=i
+j
x
y
V = (= grad )
v
dy
y
u+
u
dx
x
dy
dx
v
From volume flux balance (ie. what goes in must come out)
u
v
udy 1 + vdx 1 = u +
dx dy + v +
dy dx
x
y
29
continuity equation
u=
u
2
=
x
x
x2
v=
v
2
= 2
y
y
y
and
2 2
+
=0
x2 y 2
2 = 0
Therefore the velocity potential function like the stream function follows the
Laplace equation (harmonic functions). Hence we can use superposition of solutions,
ie. to obtain a complex flow just add simple flows together.
30
Chapter 5
Some Simple Solutions
5.1
We will find the solution (ie. stream function ) for some simple flows.
5.1.1
Parallel flow
For the parallel flow with uniform velocity U shown in Figure 5.1, we have
u=
v=
and
In this case u = U , v = 0
= 0,
x
= U
y
This gives two partial differential equations which can be solved by integration
=0
x
= f1 (y)
and
= U
y
= U y + f2 (x)
where f1 (y) and f2 (x) are functions of integration. These equations are compatible
only if f2 (x) = k where k is an arbitrary constant. For convenience the value of
is normally set to zero when y = 0 K = 0. Hence
= U y
32
5.1.2
Source flow
In source flow we have Q m3 s1 emerging from a point and flowing in the radial
direction ie.
u
ur
C.V
Source
Strength= Q m3 s1
The volume flow rate through the control volume surface with unit depth is given
by
2r 1 ur = Q
(vol. flux)
hence
Q
ur =
,
u = 0 By definition of a source
2r
. We have
1
= u ,
= ur
r
r
33
=0,
r
1
Q
=
r
2r
=0
r
= f1 ()
and
Q
1
=
r
2r
Q
=
+ f2 (r) .
2
These two equations are compatible only if f2 (r) = k usually k = 0 when = 0.
In cartesian
=
y
Q
arctan
2
x
12
Usual to have
> >
16
=4
20
24
=0
4
20
Note the =
S/L has a discontinuity
16
8
12
34
5.1.3
Sink flow
Sink
Strength= Q m3 s1
Exercise 5.1: Follow the steps outlined above and see if you can derive the stream
function for typical flows shown in Figure 5.2.
y=Uy
y=Vx
a
x
5.2
35
Source, Q
1 = U y
y
Q
Q
2 =
=
arctan
2
2
x
Since the Laplace equation is linear we can add these solutions to get the solution
for the new flow
= 1 + 2
y
Q
= U y +
arctan
2
x
We want to sketch this flow, ie. plot lines of constant . To do this we first find the
stagnation points, which are points where u = v = 0. The velocity components
in the new flow are;
u=
= U +
Q
1
1
2 1 + y2 x
x2
= U +
Q
x
2
2 (x + y 2 )
(A)
36
and
x
Q
1
y
=
2
2 1 + y
x2
v =
x2
Q
y
2 (x2 + y 2 )
(B)
y0 = 0
and from (A)
Q
x
2 (x2 + y 2 )
Q
x0 =
2U
U =
Q
So there exists one stagnation point at ( 2U
, 0).
Streamlines that pass through the stagnation points are called sepratrix streamlines. The value of on the sepratrix must be constant = x0 ,y0 and in this case
x0 ,y0 = 0. Lets plot this streamline
y
Q
arctan
= 0 = U y +
2
x
x = y cot
2U y
Q
Q
2U
Q
4U
Separtrix streamline
stagnation point
Q
x=
2U
37
Locate x intercepts
x =lim
y0
y
tan
2yU
Q
Q
2U
Locate y intercepts (x = 0)
Q
=0
2 2
Q
y=
4U
U y +
To aid sketching;
Find stagnation points and note at a stagnation point 2 streamlines come in
and two streamlines come out
sketch sepratrix streamline
consider flow close to origin (ie. source dominates) and in far field (ie. uniform
flow dominates)
streamlines cannot cross each other
adjacent streamlines must flow in the same direction.
38
Now any streamline can be replaced by a solid boundary, eg. we can replace the
= 0 streamline with a solid boundary. Hence we have solved the flow field about
a body whose shape is
2U y
x = y cot
Q
and such a shape is called a HalfRankine body (or semiinfinite body) ie.
Exercise 5.2: Repeat the derivation outlined in Section 5.2 with the free stream
velocity going from left to right. Assume that Q/(2U ) = 1. In addition, plot
the pressure coefficient, Cp , along the centerline of the body. The solution to this
exercise is shown in Figure 5.3.
5.3
Singularities
Regular singularity
5
4
3
2
1
0
1
2
3
4
5
5
39
/2
10
5
4
3
2
1
0
5
10
10
1.0
0.5
Cp
0.0
0.5
1.0
5
40
Note: Sources and sinks are called irregular singularities since they cannot occur
in practice (ie. u = v = ). However they can be used to approximate certain
practical situations.
Example
Fan
approximated by sink Q2
Fume bed
approximated by source Q1
5.4
41
Sink Q
Source Q
y
P
B
s
Have
Q
A
2
Q
B =
B
2
A =
hence
=A + B
Q
= (A B )
2
Q
=
2
Plot the streamlines, ie. lines of = constant = constant. It can be show that
the locus of = constant corresponds to circles all intersecting the xaxis at s
and +s.
42
eg.
2s
Exercise 5.3: Show that lines of = const. (ie. = const.) gives the family of
equations
1
s 2
2
2
x + y
=s 1+ 2
a
a
where a = tan = tan (2/Q)
y
1+
1
a
0, as
Source
Sink
s
5.4.1
43
Now let us add a flow from right to left, then the new flow stream function is
= U y +
Q
(A B ) .
2
Know,
tan A tan B
1 + tan A tan B
)
( y
y
A B = arctan xs yx+s
2
1 + x2 s2
tan(A B ) =
Q
= U y +
arctan xs yx+s
.
2
2
1+ 2 2
x s
It can be shown that the stagnation points lie on the = 0 streamline and this
is called the separatrix streamline, lets sketch it. The solution for = 0 has two
branches;
y=0
and
ie.
x + y s = 2ys cot
x2 y 2
2y
U s y
+ 2 1=
cot 2
s2
s
s
Q
s
2U y
Q
= 0 Oval shape
(5.1)
44
x2
s2
y2
s2
1=
2y
s
n
o
y
s
cot 2 UQ
s
Like before the = 0 streamline can be replaced by a solid body. Hence we have
infact solved the problem of flow past a body whose shape is given by equation (5.1).
This is called a full Rankine body. The shape of the body depends on the nons
dimensional parameter UQ
while the scale (size) depends on the length scale s.
Exercise p
5.4: Show that the stagnation points for the full Rankine body occur
at x = s Q/(U s) + 1.
5.4.2
The flow pattern changes dramatically if the free stream flow is in the opposite
direction. Consider the case when the flow is from left to right
Q
= U y +
arctan
2
y
xs
1+
y
x+s
y2
x2 s2
)
.
45
=0
and
and
v=0
=0
x
It turns out (exercise show this) that the location of the stagnation points depends
s
on the strength of the parameter UQ
, there are three cases;
1.
U s
Q
1 Q/(U s).
Q
U
Q
2
=0
46
2.
U s
Q
= 1/, then the stagnation points are at the origin (repeated root) this
is a degenerate case (unstable saddle)
Structurally unstable saddle
(note 3 S/L in, 3 S/L out)
3.
U s
Q
Q/(U s) 1.
s
Note as the parameter UQ
is varied the saddles (stagnation points) move, merge
and split. When this happens it is called a bifurcation.
5.5
47
This is a special case of the Rankine body where the spacing between the source
and sink goes to zero.
5.5.1
The doublet
When a source and sink of equal strength are superimposed upon one and other we
get a doublet. First consider source/sink pair spaced 2s apart;
y
P
M
r
Sink Q
B
B
Source Q
then let the source and the sink move together (ie. s 0), such that the product
Qs remains constant (K). Then as s 0
=
AM
2s sin
=
r
r
2
Q 2s sin
=
2 r
=
Hence
=
Qs sin
K sin
=
r
r
Doublet .
y
1
(x2 + y 2 ) 2
1
r = (x2 + y 2 ) 2
48
=
Ky
(x2 + y 2 )
Streamlines are a family of circles, whose centres lie on the yaxis and which pass
through the origin.
5.5.2
Now we are going to add parallel flow from right to left to the doublet. Hence the
stream function for this flow is
= U y +
Ky
(x2 + y 2 )
K
,
U
y=0.
Consider the sepratrix streamline, it passes through the stagnation point and corresponds to the = 0 streamline. The solution for = 0 has two branches
y=0
K
and x2 + y 2 =
= a2
U
r
equation to a circle with radius a =
K
U
49
x 2 + y 2 = a2
q
K
U
K
U
Again we can replace the = 0 streamline with a solid body and hence we have
solved for the flow past a circular cylinder.
Exercise 5.5: Find an expression for the velocity on the surface of the cylinder
( = 0 streamline). Use this expression to find the pressure distribution and hence
the lift and drag forces on the cylinder.
Exercise 5.6:
Two half cylinders of outer radius a are joined together in a uniform potential
flow, as shown in figure 5.4. A hole is to be drilled at an angle such that there
will be no nett force between the half cylinders at the joints. Determine the angle
assuming the internal pressure Pint to be equal to the static pressure on the
external surface of the cylinder at the point where the hold is drilled.
Hint: Remember from potential flow theory that the predicted pressure on the
surface of a cylinder is given by
1 2
2
P = P + U
2U
sin2
2
(5.1)
Exercise 5.7: Integrate Eq. (5.1) and show that the lift and drag on a circular
cylinder as predicted by potential flow theory is zero.
5.6
Circulation
We wish to solve flow about bodies that produce lift. This can be achieved by
introducing circulation around the body. Circulation is the line integral of velocity
around a closed loop. Suppose we are in a flow field where the velocity at one of the
points is V .
50
Joint
Pint
Joint
B
s
ds
Line integral of velocity from A to B is equal to the component of velocity along the
line from A to B, and we will denote this integral by LAB ie
LAB =
V sin ds .
A
work done =
F sin ds .
A
51
line of integration
ds
I
=
V sin ds
The above expression can also be expressed in terms of the velocity components u
and v ie.
I
I
= V sin ds = (udx + vdy)
Proof:
B
y
s
ds
V
x
dx
dy
52
y
u+
u
dy
y
dy
v+
v
dx
x
u
x
dx
v
u
d =udx + v +
dx dy + u +
dy (dx) + v(dy)
x
y
v u
dxdy
d =
x y
d = dxdy
Hence we can say vorticity = circulation around an element per unit area.
What is the area integral of vorticity over a finite area ?
y
yu
v1
v2
Finite area
dA
yl
x
x1 (y)
x2 (y)
v u
dxdy
x y
A
ZZ
ZZ
v
u
=
dxdy
dxdy .
x
y
A
A
53
v
x
yu
Z
x2
dxdy =
Zylyu
=
Zylyu
=
x1
(v2 v1 )dy
Z yl
v1 dy
v2 dy +
Iyl
=
Similarly it can be shown that,
ZZ
u
y
v u
x y
v
dx dy
x
yu
vdy .
I
dxdy =
udx
hence
ZZ
A
I
dxdy =
(udx + vdy)
This expression implies the area integral of vorticity = line integral of velocity on a
closed circuit around the area.
5.7
We wish to introduce circulation into potential flow problems but this requires we
introduce vorticity since
Z
=
dA .
A
However we want a irrotational flow field so put all the vorticity at a single point
called a point vortex (this will be a singularity in the flow field). For a point vortex
A 0 while such that remains finite, ie. vorticity is concentrated at a
point.
H
V sin ds =
Point vortex ()
V sin ds = 0
54
Derivation of the stream function for a point vortex with circulation (or strength)
.
u
Point vortex
ur = 0
=2ru
In polar coordinates we have
= u
r
and
1
= ur
r
1
=0
r
= f1 (r)
and
=
r
2r
= ln(r) + f2 ()
2
r
ln
2
b
Check whether the point vortex satisfies the Laplace equation ie.
2 = 0
55
In polar coordinates
1 2 1 2
= 2 2 +
+ 2
r
r r
r
For the point vortex we have
2
=
,
r
2r
2
=
2
r
2r2
and
Hence,
1
=
r
=0
2
2r
+
2r2
2
=0
2
56
5.8
It has been shown that flow around a cylinder can be generated from a doublet in
a uniform flow
=
= U y
y
K
2
x + y2
K
y
2
x + y2
K 1
= U sin r
U r
1 = U y +
r
ln
2
b
57
1
1
a2
ur =
= U cos r
r
r
r
2
a
= U cos 1 2
r
1
a2
u =
= U sin 1 + 2 +
r
r
2 r
(5.1)
(5.2)
solutions are r = a or =
2
Check if u = 0 has solutions for r = a, from (5.2);
a2
U sin 1 + 2 +
=0
a
2a
sin =
solutions exist for
<1
4U a
4U a
This means when the nondimensional parameter 4U a < 1 there exists two stagnation points located on the surface of the cylinder (r = a) and at 0 = sin1 ( 4U
),
a
ie.
y
<1
For
4U a
x
0
58
For
4U a
= sin
4U a
4U a
= ,
2
r=a
For
=1
4U a
a2
1+ 2
r
r + a2 = 0
2U
+
=0
2r
>1
4U a
This means when the nondimensional parameter 4U a > 1 there exists two stagnation points located on the yaxis (one at r < a and one at r > a) ie.
For
>1
4U a
r = a, = 0
59
Note for all of the above cases the shape of the = 0 streamline is preserved as
a circle of radius= a. If we looked at the streamlines inside the circle we would see;
=1
4U a
<1
4U a
5.8.1
>1
4U a
Pressure distribution
In order to determine the lift generated we need to know the pressure distribution
around the cylinder. Let pp
be the static pressure at some point P1 (a, ) and q be
the resultant velocity (q = u2r + u2 ).
y
q = v0
p = patm
U
P1
a
60
Now we know the velocity distribution on the surface is;
u = 2U sin +
2a
2
1
p =pt 2U sin +
2
2a
2U sin
2
1
2
2
=pt 4U sin +
+ 2 2
2
a
4 a
Lift = normal force perpendicular to the free stream direction
p
L=
pad 1 sin
(lift per unit length)
0
Z 2
1
2U sin2
2
2
3
=
pt a sin + a 4U sin +
+ 2 2 sin d
2
a
4 a
0
Z 2
1
U
2
2
=
pt a sin + a U (3 sin sin 3) +
(1 cos 2) + 2 2 sin d
2
a
4 a
0
1
U
= a
2
2
a
L =U
If the cylinder had a length of l
total lift = U l
5.8.2
Magnus effect
61
Wake
L U
D 6= 0
Exercise 5.8:
A cyinder of diameter 2.5 cm rotate as indicated at 3600 rpm in standard air which
is flowing over the cylinder at 30 ms1 . Estimate the lift per unit length of the
cylinder.
L
FreeStream
AngularVelocity
5.9
Method of images
Often we want to study flow patterns in the vicinity of a solid plane boundary. To
get the correct flow requires that the boundary corresponds to a streamline. This
can be achieved by treating the boundary as a mirror and placing images of the flow
structures behind the mirror.
For example say we have a sink located near a plane wall;
62
Wall, which must correspond to a S/L
This would happen if it was a line of symmetry
acts as a mirror
Sink Q
Image sink Q
A0
63
64
5.10
Vortex pair
Imagine we have a vortex pair held fixed in space with a uniform flow superimposed.
We can analysis the flow pattern by finding the stagnation points and sketching the
flow.
2l
ln rA +
ln rB U y
2 2
rA
=
ln
U y
2
rB
!
1
(x2 + (y l)2 ) 2
U y
=
ln
1
2
(x2 + (y + l)2 ) 2
2
x + (y l)2
=
ln
U y
4
x2 + (y + l)2
u=
= 0
v=
= 0
x
It turns out that we get different flow patterns depending on the strength of the
nondimensional parameter lU
65
<1
lU
2l
s
s
=1
lU
Bifurcation
lU
66
Kelvin Oval
>1
lU
To find the shape of the Kelvin Oval consider = 0 streamline, show this gives
"
#1
(y l)2 exp 4U y (y + l)2 2
x=
1 exp 4U y
In real physical situations we cannot have vortex pairs fixed in space hence the
pattern is unsteady. This is because the vortices induce each other along with a
velocity.
u =
2(2l)
u =
2(2l)
2(2l)
67
In order to achieve steady flow we must analyse the pattern in a frame of reference
moving with the vortex pair, this implies we see a uniform flow of
U=
4l
right to left
=4
U l
and hence the shape of the Kelvin Oval is fixed and the streamline pattern looks
like
2l
l 3
l 3
4.174l
68
y
V =
D
4D 1
ln
4
Vortex core
V
5.11
69
Point vortex
Have,
ur = 0 =
u =
1
=
2r
r
integrating gives
+c
2
Sketching lines of constant and gives
=
2
3
8
7
The stream function and the velocity potential are orthogonal to each other.
Conjugate harmonic functions.
Source
As an exercise show that the velocity potential for a source is given by =
3
Q
ln r
2
2
3
5
2
1
8
7
70
The table below shows the velocity potential and stream function of some simple
cases.
Flow
Uniform flow
Velocity potential,
U y
Q
2
Source
Potential vortex
(anticlockwise circulation)
Doublet
(anticlocwise top, clockwise bottom)
5.12
ln
arctan(y/x) =
x2 + y 2 =
K
x
x2 +y 2
Q
2
Streamfunction,
U x
Q
2
ln(r)
2
ln
K cos
r
arctan(y/x) =
p
x2 + y 2 = 2
ln(r)
y
K
x2 +y 2
Electrical analogy
The flow of electrical current in a two dimensional conductor is analogous to irrotational flow and follows the Laplace equation
2 V =
2V
2V
+
=0
x2
y 2
where V is the electrical potential and is the counterpart of the velocity potential
. Therefore we can use the electrical analogy to obtain the flow pattern through a
conduit.
Flow
Method
Q
2
K sin
r
71
Line of constant potential
= const.
Insulator
Conducting
strip
Conducting
strip
Conducting material
Probe
V
Voltage source
Voltage divider
2. Establish a voltage drop along the conductor between the flow entrance and
exit boundaries.
3. Use a potentiometer or voltmeter probe to locate line of constant potential.
To locate the lines of constant we swap the conducting strips with the insulators
and repeat the above.
conducting strip
Insulator
V
Line of constant
72
Chapter 6
The complex potential function
6.1
Introduction
In order to extend the range of patterns we can analyse it is useful to define the
complex potential function
w = + i
Potential
function
OC
C
C
Stream
function
Applies only to flows which have both a stream function 2 dimensional and
a velocity potential function irrotational.
z is the complex variable
z =x + iy
=rei ,
it can be interpreted as a position vector. The complex potential function can then
be expressed as a function of the complex variable z, ie put
w = f (z)
where f is an analytic function ( finite number of singularities).
We need to prove an analytic function given by w = f (z) = + i gives the
solution to 2 dimensional irrotational flow (for example is w = cz 2 a valid solution
?)
Consider
w = A + iB = f (z)
where z = x + iy. Differentiate with respect to x
w
dw z
dw
=
=
x
dz x
dz
differentiate with respect to y
w
dw z
dw
=
=i
.
y
dz y
dz
73
74
Hence
dw
w
1 w
=
=
dz
x
i y
Also
w A
B
=
+i
x x
x
B
w A
=
+i
y y
y
and
+i
=
i
x
x y
y
Equating real and imaginary parts
A
B
x
y
CauchyRiemann equations
B
A
=
x
y
Hence,
2B
2A
=
x2
xy
therefore
2 A =
and
2A
2 B
=
y 2
xy
2A 2A
+
=0.
x2
y 2
Similarly show
2 B = 0 .
Hence we can choose A = and B = and
w = + i
Example: complex potential function for a point vortex
From earlier lectures we have derived that
ln(r)
2
=
2
i ln(r)
2
2
= ( i ln(r))
2
w=
(6.1)
75
Flow pattern
Uniform Flow
Q
2
Source
w=
ln(z)
i
w = 2
ln(z)
w=
K
z
w = U z +
w = U z +
a2
z
a2
z
i
2
i
ln(z)
2
Exercise 6.1:
Follow the steps outlined in the above example and show that some complex potential functions of some of the flows you have seen before are as given in the Table
6.1.
6.2
=
=u
x
y
=
=v
y
x
76
Now
dw w x
=
dz
x z
=
+i
1
x
x
=
+i
x
x
dw
=u iv
dz
To find stagnation points we then solve
dw
=0
dz
6.3
for z
Velocity
dw
=u iv
dz
=2cz note stagnation point at z = 0
=2c(x + iy)
equate real and imaginary parts
u =2cx
v =2cy
77
y
=2cxy
=c(x y )
Exercise 6.2:
A very long processing vat in a factor is giving off poisonous fumes at a rate of
Qf cubic units per unit length of vat. This vat is located at x = 0 and y = 0. At
a height h directly above the vat, a long exhaust duct with uniform distributed
openings along its length exist. This duct is sucking Q cubic units/unit length. The
source of fumes from the vat can be regarded as a point source in twodimensional
flow and the exhaust duct can be regarded as a point sink.
(a) Write down the complex potential function for this problem. Remember that
you HAVE TO use a sink image and a source image (the method of images)
in order to correctly model the problem.
(b) From your answer in part (a), derive expressions for the u and v components
of the velocity field.
6.4
From previous lectures, it was shown that the combination of a doublet with uniform
flow gives a flow pattern that is similar to that of a uniform flow past a circular
cylinder (see Fig. (6.1)). In this example, the flow over a circular cylinder will be
analyse using the complex potential function w introduced in the previous lecture.
For this flow, the complex potential function is given by the sum of the complex
78
potential of uniform flow plus the complex potential of a doublet
w=
wuniform flow +
=
U z +
wdoublet
U z+
Hence
w=U
a2
z+
z
(6.2)
where
r
a=
a2
+iU y 1 2
x + y2
+i
Equating the real and imaginary parts to the the velocity potential and stream
function for a flow past a cylinder going from left to right.
a2
= U x 1 + 2
x + y2
a2
= U y 1 2
x + y2
79
5
5
0
x
Figure 6.1: Flow over a circular cylinder obtained from the complex potential function w = z + a2 /z
To obtain the velocity field, calculate dw/dz. From Eq. (6.2),
a2
dw
= U 1 2
dz
z
a2
= U 1
(x + iy)2
a2
= U 1 2
(x y 2 + i2xy)
Exercise 6.3:
Show that the above expression simplifies to
dw
a2 (x2 y 2 )
=U 1
dz
(x2 + y 2 )2
=u
2U a2 xy
+i
(x2 + y 2 )2
+i(v)
Hence
a2 (x2 y 2 )
u = U 1
(x2 + y 2 )2
and
2U a2 xy
v=
(x2 + y 2 )2
80
y
r
x
Figure 6.2: Cartesian and polar coordinate system
Sometimes, it is more convenient to work in polar coordnates (see Fig. 6.2). Let
z = rei . Substitute this into Eq. (6.2) to obtain
a2 i
re + e
r
a2
r(cos( + i sin ) + (cos i sin )
r
w =U
=U
a2
r+
r
a2
r
r
=
cos
sin
81
a2
z2
a2 i2
e
(use z = rei )
r2
a2
(cos(2) i sin(2))
r2
= u iv
Equating real and imaginary parts will give
a2
u = U 1 2 cos(2)
(6.3)
r
2
a
v = U
sin(2)
(6.4)
r2
Note that u and v are the Cartesian velocity components and NOT the radial and
tangential velocity components.
Exercise 6.4:
Prove that
u = U
a2
a2 (x2 y 2 )
1 2 cos(2) = U 1
r
(x2 + y 2 )2
and
v = U
a2
2U a2 xy
sin(2) =
r2
(x2 + y 2 )2
From Eqs. (6.3) and (6.4) the speed, V , of the fluid at any point is given by
V 2 = u2 + v 2
2
2
2
a
a2
2
2
sin(2)
= U 1 2 cos(2) + U
r
r2
a2
a4
2
2
2
2
= U 1 2 2 cos (2) + 4 (cos (2) + sin (2))
r
r
2
a
a4
2
= U 1 2 2 cos(2) + 4
r
r
2
a
a4
2
= U 1 2 2 cos(2) + 4
r
r
On the surface of the cylinder, r = a, so
2
V 2 = U
(2 2 cos(2))
2
= 2U
(1 cos(2))
2
= 2U
(1 cos(2))
2
= 4U
sin2 ()
82
4
3.5
2.5
1.5
0.5
150
100
50
0
(degrees)
50
100
150
pcylinder p
= 1 4 sin2
1
2
U
Exercise 6.5:
Determine the points on the cylinder where pcylinder = p
83
0.5
Cp
0.5
1.5
2.5
150
100
50
0
(degrees)
50
100
150
Exercise 6.6:
Show, from first principles, that the radial and tangential velocity components of the flow is related to the complex potential function, w by
ei
dw
= ur iu
dz
(6.5)
The complex potential function, w, of the flow over a circular cylinder can
be expressed a combination of free stream velocity U and doublet with
strength, .
w = U z +
(6.6)
Differentiate Eq. (6.6) and use Eq. (6.5) to find the expression for ur
and u on the surface of the cylinder expressed in cylindrical coordinates.
Find the pressure coefficient, Cp , on the surface of the cylinder.
84
Exercise 6.7:
A mathematical model of the flow in a factory with an exhaust duct (of strength
Q) and a fume bed (of strength Qf ) with cross flow (U ) is given by the complex
potential function
Qf
Q
ln z
[ln(z ih) + ln(z + ih)]
2
where h is the distance between the exahust duct and the fume bed.
w = U z +
(6.7)
(a) Use the Root Locus analysis (c.f. refer to your Control Theory lecture notes)
to locate the stagnation points in the flow field.
(b) Indicate how the location of the stagnation point changes for different values
of Qf /Q and Q/U .
(c) Sketch the flow pattern for various values of Qf /Q and Q/U .
Chapter 7
Conformal Transformations
A large amount of airfoil theory has been developed by distorting flow around a
cylinder to flow around an airfoil. The essential feature of the distortion is that the
potential flow being distorted ends up also as potential flow.
The most common Conformal transformation is the Jowkowski transformation
which is given by
c2
f (z) = z +
z
To see how this transformation changes flow pattern in the z (or x y) plane,
substitute z = x + iy into the expression above to get
= + i = z +
c2
z
c2
= x + iy +
x + iy
c2 (x iy)
(x + iy)(x iy)
+
= (x + iy)
(x + iy)(x iy) (x + iy)(x iy)
(x + iy)(x2 + y 2 ) + c2 (x iy)
=
(x2 + y 2 )
c2
c2
=x 1+ 2
+ iy 1 2
x + y2
x + y2
This means that
c2
=x 1+ 2
x + y2
c2
=y 1 2
x + y2
and
86
iy
plane
zplane
r
Jowkowski
Transformation
Figure 7.1: Jowkowski Transformation, f (z) = z + c2 /z, applied to a circle on the
zplane of radius r. In this figure, a = (r + c2 /r) and b = (r c2 /r).
c2
=x 1+ 2
r
c2
=y 1 2
r
and
So in the z plane,
x 2
r
y 2
= 1,
+
c2 2
r
2
r
c2 2
r
=1
The circle of radius r in the zplane is seen to transform into an ellipse with semiaxes a = (r + c2 /r) and b = (r c2 /r) in the plane (see Fig. (7.1)), provided
c < r. In the special case where r = c, a = 2c and b = 0. This means that if the
circle in the zplane that we wish to transform has a radius c, it will be transformed
to an infinitely thin plate of length 4r in the plane.
87
y2
5
5
0
x
0
x
0
x2
y2
5
5
y2
5
5
Figure 7.2: Flow over an ellipse obtained by applying the Jowkowski transformation
on flow over a circular cylinder. The top figure was calculated with c=0.8, middle
figure with c = 0.9 and the bottom figure with c = 1.0.
88
z1
plane
5
5
0
x
z2 =iz 1
z2
plane
5
5
2
a
z3 =z 2 + z
2
z3
plane
5
5
Figure 7.3: Figure showing the various conformal transformation used to obtain the
flow over a flat plate.
89
(7.1)
a2
z1
z2 = iz1
z3 = z2 +
7.1
a2
z2
In the zplane, the components of velocity, u and v, are given by the expression
dw
= u iv.
dz
In the plane, the components of velocity, u and v, are given by
dw
= u i
v.
d
(7.1)
(7.2)
u i
v=
(7.3)
(7.4)
= (u iv)
dz
d
(7.5)
(7.6)
From Eq. (7.5), we can determine the velocity in the plane knowing the velocity
in the zplane. To determine that, we substitute Eq. (7.6) into Eq. (7.5) to obtain
u i
v = (u iv) (A + iB) .
(7.7)
90
Take the complex conjugate of Eq. (7.7) we get
u + i
v = (u + iv) (A iB) .
(7.8)
A2 + B 2 .
(7.9)
Hence
dz
q = q
d
(7.10)
where
u2 + v2 ,
q = u2 + v 2
q =
(7.11)
(7.12)
and
dz
= A2 + B 2
(7.13)
d
Equation (7.10) shows that the velocity in the plane can be obtained from the fluid
velocity in the zplane by multiplying the fluid velocity in the zplane by dz/d.
7.1.1
From the previous lecture, it has been found that the flow over a flat plate can be
obtained from the following sequence of transformation
z2 = iz1 = iz
(7.14)
a2
z2
(7.15)
= z3 = z2 +
From Eqs. (7.14) and (7.15), we obtain
d
a2
= i i 2
dz
z
a2
= i(1 + 2 )
z
2
d
a
= 1 +
dz
z2
2
a
= 1 + 2 i2
r e
2
a
i2
= 1 + 2 e
r
a2
= 1 + 2 (cos(2) i sin(2))
r
91
1
2 cos()
= U tan()
We now need to express tan() in terms of the variables in the plane, and .
From Eqs. (7.14) and (7.15), we obtain
a2
= iz +
iz
ia2
= iz +
z
ia2
= irei + i
re
ia2 i
= irei +
e
r
On the surface of the cylinder r = a, so
= iaei + iaei
= ia(ei ei )
= ia(cos() i sin() cos() i sin())
= ia(2i sin())
+ i = 2a sin()
92
Equating real and imaginary parts gives
= 2a sin()
(7.16)
(7.17)
4a2 2
= U p
2
4a 2
To obtain the pressure distribution on the plate, use Bernoullis theorem
1
1 2
p + U 2 = pplate +
q
2
2
1 2
2
= pplate + U
2
4a2 2
Hence,
pplate p
Cp =
=
1
U 2
2
2
1 2
4a 2
Exercise 7.2:
7.2
93
94
1
1
2
2
2
1
2
2
1
1
1
B
1
Jowkowski
transformation
1
2
2
1
1
2
2
B
A
1
2
2
1
1
2
2
Figure 7.4: Figure showing the effects of moving a circle in the zplane closer and
closer to the Jowkowski transformation circle.