Inviscid Fluid Dynamics MAT2050

Semester 2
Dr Matt Turner, Spring 2014

Inviscid Fluid Dynamics MAT2050

Semester 2
Dr Matt Turner, Spring 2014

LECTURER:

Dr Matt Turner,

Room 26aAA04
[email protected]

OFFICE HOURS:

Tuesday
Thursday

UNASSESSED COURSEWORKS:

Week 5 (13/03/14 4pm)

Week 9 (08/05/14 4pm)
Week 11 (22/05/14 4pm)

CLASS TEST:

Week 8 (03/04/14 10am)

3-4pm
3-4pm

PREFERED CONTACT METHOD: Please use my office hours above, and the discussion
forums on SurreyLearn when seeking help for problems. Emailed problems may be uploaded to
SurreyLearn if they are generic and would benefit the whole class. For personal problems either
come and see me or email me.
Books:
1. Acheson: Elementary Fluid Dynamics
2. Paterson: A first course in fluid dynamics
3. Batchelor: An introduction to Fluid Dynamics
4. Lighthill: An informal introduction to theoretical fluid dynamics
None of these books follow this course exactly. Each book includes most of the material covered in
this course, but also includes other topics from the field of fluid mechanics. The books by Acheson
and Paterson are probably the most useful, with Acheson the best one to purchase if you are going
to purchase one. On the reading list for this course there are also some e-books which you might
find useful.

ii

Aims of the Course

1. Understand what we mean by a fluid (includes both liquids and gases)
2. What equations govern the motion of fluid particles?
3. How is pressure related to flow field?
4. How quickly will a barrel empty from a hole in side?
5. What causes the Severn Bore a tidal wave moving up River Severn?
6. How can we model flows around obstacles?
7. What equations describe waves on the surface of water? (eg behind duck or ship)
Expectation of each student
1. You are expected to attend the lectures and prepare accordingly by reading the notes from
previous lectures.
2. You are expected to revise material covered in previous courses which is applied in this
course.
3. You are expected to work though the lecture notes filling in the missing in examples steps,
and completing all examples which are left as exercises.
4. You are expected to use the library resources (e-books and books) to research module content
which you are having trouble to understand.
5. You are expected to submit as complete a set of solutions as possible to all the unassessed
pieces of coursework.
6. You are expected to attempt all the coursework questions which are not handed in as
unassessed coursework, in the lead up to class tests and the exam.
7. The solutions given to the coursework questions will only be sketch solutions and you are
expected to fill in the missing algebraic steps for yourself.
8. If are having problems with any of the above items then you are expected to be proactive
and use my office hours and the discussion forums on SurreyLearn to help you resolve these
issues.

iii

Notation for module

Here is a list of some of the more common notation you will meet throughout this course. Some
letters have multiple definitions, but their context will always be clear. You can add more elements
to this list if you find it helpful.
Variable
Usage
(x, y, z)
Cartesian coordinates
u = (u, v, w)
Cartesian velocity components
x, y, z
Unit vectors in (x, y, z) directions
(r, , z)
Cylindrical polar coordinates
u = (ur , u , uz )
Cylindrical polar velocity components
r, , z
Unit vectors in (r, , z) directions
r
position vector
s
unit position vector
t
time
p
pressure

density
w(z) = w(x + iy)
complex potential
U
Velocity
g = 9.81ms2
gravity

vorticity

vorticity in z direction for 2D flows/frequency

velocity potential
log(x) = loge (x) = ln(x)
Natural logarithm

Circulation

Stream function
n, N
normal vector
F
Force
S
Surface area element

Free surface/height
T
Temperature, Kinetic energy
V
Volume, Potential energy
S
Surface

iv

The Greek Alphabet

Small

Capital
A

Name
alpha

beta

gamma

delta

epsilon

zeta

eta

theta

iota

kappa

mu

nu

xi

omicron

pi

rho

sigma

tau

upsilon

phi

chi

lambda

psi

omega

Contents
1 Vector Calculus
1.1 Vector Calculus . . . . . . . .
1.1.1 Definitions . . . . . . .
1.1.2 Directional Derivatives
1.1.3 Normals to Surfaces .
1.1.4 Identities . . . . . . .
1.2 Integral Theorems . . . . . . .
1.2.1 Divergence Theorem .
1.2.2 Important Corollary to
1.2.3 Stokes Theorem . . . .

. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
Divergence Theorem
. . . . . . . . . . . .

2 Introduction
2.1 Density and the Continuum Hypothesis
2.2 Pressure . . . . . . . . . . . . . . . . . .
2.3 Equations of State . . . . . . . . . . . .
2.4 Statics & Hydrostatics . . . . . . . . . .
2.5 Buoyancy and Archimedes Principle . . .
2.6 One-dimensional flow in tubes . . . . . .
3 Kinematics
3.1 Velocity . . . . . . . . . . . . . . . . . .
3.2 Describing flow fields . . . . . . . . . . .
3.2.1 Streamlines . . . . . . . . . . . .
3.2.2 Particle Pathlines . . . . . . . . .
3.2.3 Streaklines . . . . . . . . . . . . .
3.3 Convective Derivatives . . . . . . . . . .
3.4 Mass Conservation and Compressibility .
3.5 Vorticity . . . . . . . . . . . . . . . . . .
3.6 Velocity Potential for Irrotational Flows

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

4 Dynamics
4.1 Eulers Equation of Motion . . . . . . . . . . . .
4.2 Boundary Conditions . . . . . . . . . . . . . . .
4.3 Bernoullis Equation . . . . . . . . . . . . . . .
4.3.1 Irrotational Flow . . . . . . . . . . . . .
4.3.2 Steady Flow . . . . . . . . . . . . . . . .
4.4 Momentum Integral Theorem for Steady Flows .
4.5 Calculating Velocity Potential for general flows .
4.5.1 Uniqueness of velocity potential . . . . .

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

1
1
1
2
3
4
5
5
6
6

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

7
7
8
10
10
13
14

.
.
.
.
.
.
.
.
.

18
18
19
19
21
23
27
27
28
29

.
.
.
.
.
.
.
.

32
32
33
33
34
36
37
39
39

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

5 Channel Flows
42
5.1 Steady flow under a Sluice Gate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.2 Hydraulic Jumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.3 Flow over obstacle on river bed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

vi

6 Two-dimensional Flows
6.1 The stream function . . . . . . . . . . . . . . . . . . . . . . . .
6.1.1 Properties . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.2 Governing Equations and Boundary Conditions . . . . .
6.1.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Line Vortices . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.1 Velocity potential of line vortex . . . . . . . . . . . . . .
6.2.2 Circulation . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.3 Relation between vorticity and circulation . . . . . . . .
6.2.4 Kelvins Circulation Theorem Inviscid Flows . . . . . .
6.2.5 Significance of Irrotational Flows . . . . . . . . . . . . .
6.3 Combining simple examples of 2D velocity potentials . . . . . .
6.4 General solution of velocity potential in plane polar coordinates
6.5 Complex Potentials for incompressible, irrotational 2-D flows . .
6.5.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5.2 Boundary Conditions on w(z) . . . . . . . . . . . . . . .
6.5.3 Velocity in terms of Complex Potential . . . . . . . . . .
6.5.4 Simple Examples . . . . . . . . . . . . . . . . . . . . . .
6.5.5 Combining Simple Examples . . . . . . . . . . . . . . . .
6.6 Calculating Forces on Body using Pressure . . . . . . . . . . . .
6.7 Induced Velocity and Image Systems . . . . . . . . . . . . . . .
6.7.1 Induced Velocity . . . . . . . . . . . . . . . . . . . . . .
6.7.2 Image Systems . . . . . . . . . . . . . . . . . . . . . . .
7 Surface Waves
7.1 Introduction . . . . . . . . . . . . . . . . .
7.2 Basic wave theory . . . . . . . . . . . . . .
7.3 Potential Flow with a Free Surface . . . .
7.4 Small amplitude disturbances . . . . . . .
7.5 Small-amplitude travelling wave solutions .
7.6 Linear Superposition and Standing Waves
7.7 Phase Velocity and Group Velocity . . . .
7.8 Particle paths for water waves . . . . . . .

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

47
47
47
49
49
51
51
52
52
53
53
53
55
56
56
57
57
57
61
64
69
69
71

.
.
.
.
.
.
.
.

75
75
75
76
77
78
79
80
83

A Cylindrical Polar coordinates

86
A.1 Vector operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
B Divergence theorem

87

C Stokes theorem

89

D Proof of Kelvins Circulation Theorem

91

E Flow around a cylinder

92

F Propagation of energy for deep water waves

93

vii

Vector Calculus
1.1 Vector Calculus
1.1.1 Definitions
In this course I will be using x , y , z for the unit vectors in the 3 Cartesian directions, rather
than i, j, k.
The position vector of a point P relative to a fixed origin O is written as
OP = r = xx + yy + zz .

We now consider scalars and vectors which are functions of position.

1. A single-valued function of the three variables x, y, z written as (x, y, z) can be considered as
a scalar function of position (r). This is known as a scalar field.
2. A single-valued vector function of x, y, z, u(x, y, z), can be written as
u (r) = u(x, y, z)x + v(x, y, z)y + w(x, y,
z)z , and this is known as a vector field.
Therefore given a scalar field (x, y, z) and a vector field
u(x, y, z) = u(x, y, z)x + v(x, y, z)y + w(x, y, z)z ,
we define
=

+
x
x
y

y +
z

2=

2
2
2
+
+
,
x2
y2
z2

u=

u v w
+
+
,
x y
z
y

z ,

(This is a
vector!)

(Pronounced divergence)

x
u= ,
x y z
u v w

(Pronounced curl)

2u 2ux + 2vy + 2wz .

=
The form of these definitions in polar coordinates can be found in Appendix A. You should know
them in Cartesian coordinates along with in polar coordinates, but the other polar expressions
will be given in an exam.

1.1.2 Directional Derivatives

Note that s is the component of a vector in the direction of the unit vector s .
We now show that
= Rate of change of in the s direction.
s =s
Proof
Consider a point P with position vector r. At this point is given by (r) = (x, y, z).
Now consider a second point Q close to P, with position vector r + r. The value of at Q is
(r + r) = (x + x, y + y, z + z).
This can be expanded using Taylors Theorem,
(x + x, y + y, z + z) = (x, y, z) + xx + yy + zz + smaller terms, so
the change in is given by
=(x + x, y + y, z + z) (x, y, z)
=xx + yy + zz
=(xx + yy + zz ) x
(x + yy +
zz ) =r
Now let r be in the direction of unit vector s , so
r = (s)s ,

where s = |r| = distance PQ

and hence
= s s , =
Taking the limit s 0 gives required result.

= s .
s

Example 1
Calculate the rate of change of = xyz at the point (1, 1, 1) in the direction parallel to x +2y +2z
. The unit vector parallel tox + 2y + 2z is
s =

1
1
(x + 2y + 2z ) =
(x + 2y + 2z ).
3
1 + 22 + 22

Then
=yzx + xzy + xyz
1(yz + 2xz + 2xy)
s =3
1
5
= (1 + 2 + 2) = ,
3
3

when x = y = z = 1.

1.1.3

Normals to Surfaces

We know that in two-dimensional space, f(x, y) =constant defines a line, and similarly the equation
(x, y, z) =constant defines a surface in 3-D space.
Examples:
Set of parallel planes for different values of k
x + 2y z =k
2
2
2
x + y + z =k
Set of concentric spheres
In 1.1.2 it was shown that the rate of change of in a direction s is
.
s =s
Now if s is chosen to be a tangent to the level surface, so s = t then in this direction

= 0, = t = 0.
s
Thus is orthogonal to any vector tangent to the surface =constant, and so is normal to the
surface. (Another proof will be seen later in the course).
The unit normal to the surface =constant is given by

,
=

n
||
the two possibilities corresponding to normals on either side of the surface.

Example 2
Calculate the normal to the surface x2y + 2xz = 4 at the point A = (2, 2, 1).
Calculate the equation of the tangent plane to the surface at A and find the perpendicular distance
of this plane from the origin.
=x2y + 2xz,
=(2xy + 2z)x + x2y + 2xz , =
at A

6x + 4y + 4z = 2(3x + 2y + 2z

). is
Hence the unit normal to the plane
1
(3x + 2y + 2z ).
=

n
17
The equation of a plane with unit normal n is rn = p, where |p| is the perpendicular distance
of the plane from the origin.
1
1
rn = (xx + yy + zz )(3x + 2y + 2z= ) (3x + 2y + 2z) = p.
17
17
Since the plane passes through (2, 2, 1),
1
(6 4 + 2) = p.
17
3

Hence the perpendicular distance of the plane from the origin is

8
|p| = ,
17
and the equation of the tangent plane is
3x + 2y + 2z = 8.

1.1.4

Identities
() = 0,

( u) = 0.

A further identity we will use, but which you have probably not seen before is
u ( u) =

1 u2
2

(u)u.

It is important to understand how to calculate each term in this identity. Note that u2 = u u.
If u = ux + vy + wz , then

1 u2
2

+1
u2 + v2 + w2 y +2 1
u2 + v2 + w2 ,
x 2 y
z

u2 + v2 + w2

1
x
2

+v
+w
(ux + vy + wz ),
x
y
z
=(uux + vuy + wuz)x + (uvx + vvy + wvz)y + (uwx + vwy + wwz)z .

(u)u= u

(Proof not included here show by components)

Example 3
Show that the identity is valid when u = yx 2zy +
xz .
u=2x y z ,
u ( u)=(2z x)x + (y + 2x)y + (4z y)z
, 1 u2 = 1 +
y2 + 4z2 + x2 ,
2

1 u2
2

= xx + yy + 4zz ,

(u)u= y2z + x
(yx 2zy + xz ),
x
y
z
=2zx 2xy + yz ,
(u)u=(x + 2z)x + (y + 2x)y + (4z
y)z ,
=u ( u).

Another useful identity that we shall use is

(u) = u + u .
This identity is analogous to the product rule for differentiation.
4

1.2
1.2.1

Integral Theorems
Divergence Theorem

Let w be a vector field, V be a volume of space with surface S and outward unit normal

Then
n.

w dV,

wn dS =
S

where dS is an element of surface area. For the proof of this see appendix B.

Example of verifying Divergence Theorem (Example 4)

Verify the divergence theorem for w = 4xx 2y2y + z2z taken over the region bounded by x2 +
y2 = 4, z = 0 and z = 3.
First the volume integral:
(4x) +

w dV =
V x

2
(z )
(2y2) +
y
z

dV,

(4 4y + 2z) dV,

=
V

4x2

x=2

y= 4x2

(4 4y + 2z) dxdydz,
z=0

=84.
The surface S of the cylinder consists of a base S1 (z = 0), the top S2 (z = 3) and the convex
portion S3 (x2 + y2 = 4). Thus the surface integral is
wn dS =
S

wn dS1 +
S1

wn dS2 +

wn dS3.

S2

S3

On S1 (z = 0)
2
n= z , w = 4xx 2y y and w n = 0,

wn dS1 = 0.
S1

On S2 (z = 3)
2
n= z , w = 4xx 2y y + 9z and w n = 9,

therefore
dS2 = 36.

wn dS2 = 9
S2

S2

On S3 (x2 + y2 = 4), the unit normal to this surface is

u=

1
(xx + yy ) ,
2

thus
w n = 4xx 2y2y + z2z

1
(xx + yy ) = 2x2 y3.
2

On the surface of the cylinder

x = 2 cos and y = 2 sin ,
thus
dS3 = 2ddz.
Therefore
2

=0

z=0

wn dS3 =
S3

2 (2 cos )2 (2 sin )3 2dzd,

48 cos2 48 sin3 d,

0
2

cos2 d,

=48
0

=48.
Then the surface integral = 0 + 36 + 48 = 84, agreeing with the volume integral and thus
verifying the divergence theorem.

1.2.2

Important Corollary to Divergence Theorem

By writing w = pa where p(r) is a scalar field of the position vector r and a is an arbitrary constant,
w = (pa) = (p) a,
the divergence theorem
gives
a p dV, = a

pan dS =
S

p dV

pn dS = a

taking the constant vector a out of the integrals. Since a is

arbitrary,

1.2.3

p dV.

pn dS =
S

Stokes Theorem

Let C be a simple closed curve in space, spanned by surface S with unit normal
( w)n dS =
S

wdr.
C

For the proof of this see appendix C.

then
n,

Example of verifying Stokes Theorem (Example 5)

Verify Stokes Theorem for the vector field
w = z2x + y2y + xz
where the closed curve C is the triangle with vertices A = (1, 0, 0), B = (0, 1, 0) and C = (0, 0, 1).
We first calculate the line integral going from A to B to C and back to A.
AB

z =0

BC

x=0

CA

y=0

dr=dy(x + y ) wdr= 2(yy + xz )(x + y )dy = y2 dy

1
IAB = 0 y2dy = 13
dr=dz(y + z ) wdr=(z2 + (1 z)2y )(y + z )dz
1
2
IBC = x
0 (11 z) dz = 3
dr=dx(x z )
wdr= x)2x + xz )(x z )dx
3
2
ICA= ((1
+ 13 = 16
0 11 3x + x dx = 1 2

x=1 y
0<y < 1
y=1 z
0<z < 1
z =1 x
0<x < 1

Hence
wdr =
C

1
3

13 16 = 16

To calculate the surface integral over the plane x + y + z = 1,

w=

y
y2

x
z2

z
x

= 0x + (2z 1)y + 0z ,

+y + z
= x
.
n
3

Hence

1
( w)n dS =
(2z 1)dS.
3
To evaluate this integral we project the surface onto the x, y plane
so
1
dx dy
1 1x
( w)n dS =
(2(1 x y) 1) z =
1 2x 2y dy dx,
3
0
0
n
2
1 (1 2x)(1 x) (1 x) dx =
1
=
x + x2dx = 16 .
0

Introduction

2.1 Density and the Continuum Hypothesis

Consider a cube of fluid, centred at P, with side a. If the mass of the fluid contained in the cube is mP
(a), then the average density of the fluid in the cube is
=

Mass of Fluid in Cube

.
Volume of Cube

The variation of with a will then look something like

7

For larger a, the average density will change due to general variations of the fluid (eg temperature of
gas, salinity of seawater etc). As a gets very small and approaches the molecular separation distance
there will only be a small number of molecules in the cube and the average density will change
significantly for small changes of a.
For a small, but much larger than the molecular separation distance, the average desnsity is
approximately independent of a and we define this as the density at point P. This is known as the
Continuum Hypothesis. Other quantities such as fluid velocity are defined in a similar way. Typical
values
For most liquids the density depends on temperature (but is almost independent of pressure see
later). Typical values at room temperature and atmospheric pressure:
Water
Mercury
Air

2.2

= 1000kg m3
= 13, 500kg m3
= 1.29kg m3 sensitive to temperature and pressure

Pressure

Consider a small element of area S with normal by

a fluid.

lying on the boundary of a region occupied

N

The fluid exerts a force, F on the area element perpendicular to the element,
F = |F |.

F = F N ,
The pressure, p on the area element is defined by
p = lim
S0

F
,
S

where again the continuum hypothesis is applied (S is not taken so small that motion of individual
molecules becomes important).
For a general area S, the total force on S due to the fluid is
F =

is normal out of fluid.

N

pN dS,
S

(Note that if the surface is curved,

is in different directions at different points on S.)

N

Example 6
A water tank consists of a cube occupying the region
0 x 1,

0 y 1,

0 z 1,

and the pressure of the water inside the tank is given by p = 1 z (we will look later at how p can be
calculated).
Calculate the force on each face of the cube.
Let n the normal out of the fluid.

Face 1

z=0

N= z

p=1
p=0

F2 =0,

p=1z

F3 =

Face 2

z=1

=z
N

Face 3

x=0

N= x

F1 =
0

(z )dx dy = z ,

1
0

(1 z)(x )dy dz,

=x

1
0 (1

Face 4

x=1

=
N x

p=1z

F4 = 12

Face 5

y=0

N= y

p=1z

F5 =21 y ,

Face 6

y=1

=y
N

p=1z

F6 = 12 y .

,x

z)dz = x z 12 z2

1
0

= 12 ,x

2.3

Equations of State

The equation of state relates the pressure p and the density .

For a liquid the density is constant, independent of pressure.
Ideal Gas Laws:
For an ideal gas, the pressure p, density and temperature T are related by
p
= constant.
T
If the temperature is constant then p . Other special cases also apply.

2.4

Statics & Hydrostatics

Statics: Consider a body of mass m suspended on a string.

g
mg

There is a an upward force due to tension in the string, T = T z and a force mg = mg z due to
gravity.
If the body is in static equilibrium the total force on the body is zero,
T + mg = 0,

= T = mg z.

Hydrostatics: Now consider a large volume of stationary fluid and a smaller region R of fluid
within this volume. The surface of this region is taken to be S with n the unit normal out of R.

10

Since the fluid is stationary, the weight of the fluid in the volume must be balanced by a force on R
due to the pressure p of the surrounding fluid.
From 2.2 the total force on R is
F =

p N dS,
S

where N is the unit vector normal to S and out of the surrounding fluid. Since N = n , applying
the corollary to the divergence theorem (see 1.2) gives
F=

p dV.

p n dS =
S

If the density of the fluid is , the mass of fluid in region R is

m=

dV.
R

Since the fluid is stationary, the total force on the fluid inside R is zero and so
p dV = 0.

g dV
R

This result must be true for all volumes and so

p = g.
Since g = gz , this vector equation is equivalent to the three scalar equations
p
= 0,
x

p
= 0,
y

p
= g,
z

and so p is a function of z only.

If is known as a function of p from the equation of state (see 2.3) then p can be found as a function
of z.
Example 7
A diver is distance h below the surface of water with constant density, = 1000kg m3.
At the surface the air pressure is pa = 105N m2. Calculate the pressure experienced by a diver 10m
below the surface. At what depth is the pressure three times atmospheric pressure?Defining the zaxis as being vertically up, with the surface of the water at z = 0, the pressure is given by
dp
p(0) = pa.
= g,
dz
Since the density is constant, we can integrate to get p(z),
p = gz + constant = pa gz.
Thus for a diver at z = h, the pressure is pa + gh.
When h = 10m,
p = (105 + 1000 9.8 10)N m2 = 198000N m2 = 1.98 pa.
When p = 3pa,
3pa = pa + gh, = h =

11

2pa
20.4m.
g

Example 8
In an isothermal atmosphere,
p=

p0
,
0

p0, 0 pressure and density at sea level.

If the 0 = 1.3kg m3 and p0 = 1000N m2, calculate the pressure at 10000m above sea-level.
dp
0g
1.27 104m1.
= g = kp, k =
dz
p0
Solving by separation of variables,
dp
= k
p

dz, = log p = kz + constant, = p = p0ekz.

So when z = 10000m, (approximately summit of Mt Everest),

p = p0e1.27 0.28p0,
the pressure is approximately a quarter of that at sea level.

Measuring Pressures
1. Simple Barometer

Pressure depends only on z so

pA = pB = pAtm.
Vacuum at top of tube pC = 0.
Hence choosing z = 0 at C, p
= gz, pAtm = gh.

If liquid is mercury, = 13, 500kg m3, pAtm = 105N m2 and so

h=

105
0.756m,
9.8 1.35 104

or 756 mmHg (millimetres of mercury).

12

2. Pressure differences using a manometer

Consider a U-shaped tube attached to two vessels containing gases at different pressures. It is
assumed that the density of the gases in the two vessels is much lower than the density of the
fluid in the tube.

pA = pC ,
gas)

pB = pD

(ignoring density of

pE = pF , (same vertical level & no flow)

pE = pC + g(EC), pF = pD + g(FD),
pA = pE g(EC) = (pD + g(FD)) g(EC),
= pB + gh.

2.5

Buoyancy and Archimedes Principle

Definition
For a body in a static fluid the buoyancy is the force due to the pressure of the surrounding fluid.
Consider a body, volume V, completely submerged in a fluid of density . The buoyancy force is
given by
B=

pn dS,
S

where S is the surface of the body, n is the unit normal out of the body and into the fluid, and p is
the fluid pressure at the surface of the body.

13

The fluid pressure is a function of z only, with the pressure at A the same as at B. Thus the pressure
on S is the same if the body is replaced by fluid (figure (b)). So we can then apply the divergence
theorem
p dV = z g

pn dS =

B=
S

dV = V gz ,
V

=mW gz ,
where mW is the mass of fluid displaced by the body.
This is Archimedes Principle the buoyancy force is equal to the weight of displaced fluid.
The same result holds for a partially submerged body.

V1 = Volume of water displaced W

= Water density
mB

= Mass of body

The buoyancy force is then B = W V1gz . Since it is floating, the buoyancy force is equal and
opposite to the weight of the body, and so
mB = W V1.
So if the volume of displaced water can be measured and the density of water is known, the mass
of the body can be found.

2.6 One-dimensional flow in tubes

We now consider flow through a tube (which may have varying cross-sectional area, and may not be
level).
We define s to be the distance along the tube from some fixed point, A(s) to be the cross-sectional
area, and z(s) the height of the mid-point of the tube above some fixed level surface.

14

We assume the density is constant, that the flow is parallel to the walls of the tube and that the
average velocity across the tube is given by u(s). (NB As long as A(s) changes slowly, the
assumption of parallel one-dimensional flow is valid.)
We now consider a section of tube.

In time t, the mass flow of fluid through A1 is tu1A1, while the flow out of A2 is tu2A2. Since
the density of the fluid is constant, the total mass in = total mass out, so
u1A1 = u2A2.
A more general form of this equation will be derived in 3.4 and is known as the Mass Conservation
Equation or Continuity Equation.
Energy Equation
For a particle of mass m in a gravitational field, with velocity u and vertical coordinate z, we define:
Kinetic Energy (due to motion) T = 12m|u|
2
Potential Energy (due to position) V =mgz
15

In MAT1036 (Classical Dynamics) last year it was shown that the total energy is conserved:
T + V = 12m|u|2 + mgz = constant.
A similar relation can be derived for fluid motion along a pipe
1
2
2 |u| +gz

KE

PE

=constant.

Pressure
Effect

This is a form of Bernoullis Equation.

We do not derive this here a general derivation is given in 4.
NB If gravity is ignored, then Bernoullis Equation says that at higher flow speeds the pressure is
lower.
Example 9: Canvas sided truck inside the truck the fluid velocity is zero, while outside the fluid
velocity is high, so pressure is lower and sides bulge out.

Returning to our example of a pipe, Bernoullis equation becomes

1
2
2 u 1 +

gz1 + p1 = 12 u22 + gz2 + p2,

or
p2 p1 = 1 2 (u21 u22) + g(z1 z2).
This is supplemented by the conservation of mass equation
u1A1 = u2A2.

Example 10 Venturi Tube

For a tube of cross-sectional area A1, a constriction of cross-section area A2 is inserted. Vertical
tubes are attached to the pipe in the uniform section and at the constriction. If the difference in
height between the fluid levels in the pipe is H we can find the fluid velocity in the tube using
Bernoullis equation and the equation of mass conservation.

16

Applying Bernoullis equation at the points P and Q, where the pressure is p1 and p2 respectively
2
p1 + gr1 + 12u21 = p2 + gr2 + 1 u
2.
2

Using the hydrostatic balance in the two vertical tubes and noting that the pressure at R and S is
atmospheric pressure, pa,
p1 = pa + gh1,
p2 = pa + gh2,
Eliminating p1 and p2 from Bernoullis Equation,
1
2
2 (u 2

u21) = g ((h1 + r1) (h2 + r2)) = gH.

Conservation of mass gives A1u1 = A2u2 and so

A21
A22

u21 = 2gH, = u21 =

2gHA22
A21 A22

Note:
1. Bernoullis equation holds along a stream tube (or streamline) for steady flows and in the
absence of viscous (frictional) forces.
2. For many flows, the whole space occupied by the fluid may be regarded as a stream tube, e.g.
the wall of a pipe is a surface of a stream tube. In this case the solution is only approximate as
we have neglected viscous effects which will be important near the walls of the pipe, although
their overall effect on these integral properties may remain small.

17

Kinematics

Kinematics is the study of velocity and acceleration of fluid particles.

3.1

Velocity

In Cartesian coordinates, we write the velocity of the fluid particle at a point with coordinates (x, y,
z) as
u(x, y, z, t) = ux + vy + wz .
If the fluid velocity depends on time, t, we say the flow is unsteady, while if
u

= 0 = flow is steady. t

The fluid speed is given by

|u| = u2 + v2 + w2

1
2

Definitions
1. A point at which u = 0 is known as a stagnation point.
2. A flow is two-dimensional if
u
v
=
= 0,
z
z

and w = 0.

Examples
11- u = xx yy + 2zz steady flow with stagnation point at (0, 0, 0).
12- u = a(y t)x axy , a two-dimensional unsteady flow with stagnation point at x = 0, y = t.
Polar Coordinates for 2-D flows
For a general point P with Cartesian coordinates (x, y), the polar coordinates are (r, ) with

x = r cos ,
y = r sin ,

=cos x + sin y ,

= sin x + cos y .

18

The expression for r comes from r/r, while is obtained by noting that r = 0.

3.2 Describing flow fields

If the fluid velocity u is known everywhere, there are a number of ways in which we can visualise the
flow.
In 2-D we can draw an arrow at various points illustrating the direction and magnitude of the flow.
Example 13
u = ayx + axy .
In polars this is
u = ar ( sin x + cos y ) =
ar . Hence the flow can be visualised by

An alternative way of visualising the flow is by considering particle pathlines, streaklines and
streamlines. We provide a general definition of each line but only illustrate the ideas for 2-D flows.

3.2.1

Streamlines

A streamline is a curve which, at an instant of time, is tangent to the flow velocity at all points along
the curve.
Note that for an unsteady flow the pattern of streamlines changes with t.
In simple cases the pattern of streamlines can be deduced from the direction of flow at each point.
Example 13a see previous example.
u = ayx + axy = ar .
Streamlines are concentric circles.

19

Mathematical Formulation
Consider the streamline passing through the point x = xx + yy + zz = x0, and choose s to
be a coordinate along the streamline with s = 0 when x = x0.
NB s is any coordinate along the curve, not necessarily the distance from x0.

Since the tangent to the curve is parallel to the velocity field,

dx
= u, for some .
ds
Choosing the scaling on s correctly we can set = 1 and so we must solve
dx
= u(x),
ds

x(0) = x0.

Writing u = ux + vy + wz , an alternative way of writing this differential equation is

dx
dy
dz
=
=
(= ds),
u
v
w
from which it follows that

dy
v
= ,
dx
u

dz
w
= ,
dx
u

etc.

Example 13b see previous example.

u = ayx + axy .
Hence,
dy
v
ax
= = .
dx
u
ay
Solving by separation of variables
y dy =

x dx, =

y2
x2
= + C, = x2 + y2 = constant.
2
2

So the streamlines are concentric circles centred on the origin.

20

Example 14
u=x +
ty . The streamlines at time t = t0 are given by
dy
= t0
dx
ie straight lines with gradient t0.

t<1

3.2.2

t>1

Particle Pathlines

Consider a fluid particle at point P (x0, y0, z0) (position vector x0) at some fixed time t = 0 (say).
As t increases, the fluid particle moves through space the path taken is known as the pathline. By
plotting a set of pathlines, the flow pattern can be visualised.
NB As t increases, the picture only changes in that the pathlines get longer.

Example 15
u = ayx ,

21

t = t1

t = t2 > t1

Mathematical Formulation
dx
= u (x, t) ,
dt

x(t = 0) = x0.

For each x0 chosen, there is a particle

path.
Example 16
u = ayx axy .
The equation governing the particle path passing through the point (x0, y0) at t = 0 is
dx
= ay,
dt

dy
= ax,
dt

x(0) = x0,

y(0) = y0.

Solving these two linked differential equations is easiest by differentiating the first equation and
eliminating y using the second equation,
d2x
dy
=
a
= a2x.
dt2
dt
Solving
d2x
+ a2x = 0 = x = A cos(at) + B sin(at).
dt2
Then using the initial conditions we obtain
x=x0 cos(at) + y0 sin(at),
y=y0 cos(at) x0 sin(at),
Note that
x2 + y2 = x20 + y02,
and so the pathlines are circles.

22

Example 17
u = yx (x bt)y
. Prove that particle path through (x0, y0) are given by
x=(y0 b) sin t + x0 cos t + bt,
y=(y0 b) cos t x0 sin t + b.
Results from MATLAB for b = 1 for the particle paths through (0, 0.5) and (0, 0).
2

1.5

0.5

0
0

10

12

14

Example 18
u = x + ty .
Calculate the particle path for a particle at position (x0, y0) at time t = .
dx
=1, = x = t + C1 = t + x0, (using x() = x0)
dt
dy
2
2
=t, = y = 1 t2 +2 C2 = 1
2 (t ) + y0, (using y() = y0).
dt
Hence the particle path for a particle at the origin at t = 0 is
x = t,

3.2.3

y = 12 t2,

or equivalently y = 21 x2.

Streaklines

A common technique for visualising flow in experiments is to release smoke or dye at a point and see
how it is carried by the flow (eg smoke from the top of a chimney). The position of these marked
particles at a particular time is known as the streakline. For an unsteady flow, the streakline changes
position as well as lengthening as t increases. For a steady flow the streakline will lengthen as t
increases but not change shape.
Mathematical Formulation
Solving
dx
x(t = ) = x0.
= u(x, t),
dt
23

This gives the position of a particle x as a function of t, (release time) and x0 (release position)
x(x0 , t, ).
The streakline at time t0 is then given by the set of points
0 t0 .

x(x0 , t0 , ),
Example 19
From example 18 of previous section, if

u = x + ty,
then the particle pathlines for a particle released from the origin at time t = are
y = 12 (t2 2 ).

x = t ,

Thus the streakline at time t0 , originating from (0,0) is given by the parametric curve
x = t0 ,

y = 21 (t20 2 ),

0 t0 .

The parameter can be eliminated to give

y = xt0 21 x2 ,

0 x t0 .

The streaklines at t0 = 1 and t0 = 2 are plotted on the handout.

24

Streaklines and Pathlines

u = x + ty,

0.5
0.4
0.3
0.2
0.1
0
0

0.2

0.4

0.6

0.8

Streakline at t0 = 1, path lines for particles released at = 0, 0.5.

1.5

0.5

0
0

0.5

1.5

Streakline at t0 = 2, path lines for particles released at = 0, 0.5, 1.0, 1.5.

25

Example 20
u = sin(t)x + y.
Slope of streamlines is given by
dy
1
=
.
dx
sin(t)
Pathlines

Consider particle released from origin at time t = 0.

dx
=sin(t), x(0)
dt
dy
=1,
y(0)
dt

= 0,
= 0,

=
=

x(t)= 1cos(t)
,

y(t)=t.

This is plotted as a parametric curve below.

Streaklines
Consider the position of a particle at time t which was released from origin at
time t = .
cos( ) cos(t)
dx
,
=sin(t), x( ) = 0, = x(t)=
dt

dy
=1,
y( ) = 0, = y(t)=t .
dt
The streakline at t = 2/ is then given as a parametric equation,
x( ) =

cos( ) 1
,

y( ) =

2
.

1
0.8

1.5

0.6
1
0.4
0.5
0.2
0
0

0.05

0.1

0.15

0.2

0.25

0
0.35

0.3

Pathline for particle at origin at t = 0 with = 2.

26

0.3

0.25

0.2

0.15

0.1

0.05

Streakline at t = 2/ for particles

released from origin at 0 < < 1.

3.3

Convective Derivatives

Consider a fluid in motion with density and velocity u depending on time and position. If we
consider a small blob of fluid, we examine how the density of this blob changes as it moves through
the fluid. We first consider a 1-D flow field. Then a blob centred at P x at time t, is at point Q x
+ u(x, t)t at time t + t. Hence the density of the blob at time t + t can be written as

+ (u(x, t)t)
+ ...,
t
x

Q = (x + u(x, t)t, t + t) (x, t) + (t)

as t 0, using a Taylor expansion about (x, t). Hence the rate of change of as we move with
the fluid is

Q P
=
+ u(x, t)
+ Small Terms.
t
t
x
This is known as the convective derivative of (also known as the material derivative),
written as
D

=
+ u(x, t) .
Dt
t
x
This is readily extended to a 3-D velocity field,
D

=
+ u ,
Dt
t

(3.3.1)

where the convective derivative is the operator

D

=
+ u .
Dt
t

3.4

Mass Conservation and Compressibility

By considering flow in and out of a fixed volume, we can derive the Conservation of Mass
Equation

+ (u) = 0,
t
or equivalently,
D
+ u = 0.
Dt
Proof
Consider a volume of fluid V fixed in space, with surface S. In time t, the mass flow through a small
element of surface S with normal n is given by un S t, since un is the component of
velocity normal to the surface.
Then the total mass flow into V through the surface S is given by
t

un dS,
S

the minus sign arising because n is the normal out of the

volume. The change in mass contained in volume V is then
m =

dV = t
V

27

dV,
t

and so
V

dV =
t

un dS,
S

(u) dV,

(using divergence theorem).

This is true for all volumes hence result.

For an incompressible fluid, the density is constant as we move with the fluid, so
D
= 0, = u = 0.
Dt
Throughout this course, all fluids are assumed to be incompressible, so
u = 0.
This is a good approximation for water and also for air at low speeds (M = u/c
the Mach number and c is the speed of sound).

Example 21
u = (2x + y)x + (ay + 2yz)y +
bz2z , Find a and b such that the flow is incompressible.

a,b constants.

u = 2 + (a + 2z) + 2bz = (2 + a) + (2 + 2b)z.

Need u = 0 for all z and so need
2 + a = 0 = a = 2 AND 2 + 2b = 0 = b = 1.

3.5

Vorticity

Definition Vorticity
The vorticity of a flow is given by
= u.
Note that this is a vector quantity.

Example 22
Calculate the vorticity of the flow with velocity
u = 2yx + zy + 4xz .
y z

u= x
= (0 1)x 4y 2z .
x y z
2y z 4x

28

1, where M is

Vorticity of 2-D flow

Consider
u = u(x, y)x + v(x, y)y ,
then
=u=

u
v
z = z .
x y

So for any 2-D flow, the vorticity is normal to the plane of the flow.
Definition Irrotational

A flow is irrotational = 0,
u = 0.

Example 23
If u = yz2x + xz2y + 2xyzz , show that the flow is
y

irrotational.
u=

3.6

= (2xz 2xz)x (2yz 2yz)y + (z2 z2)z = 0.

x y z
yz2 xz2 2xyz

Velocity Potential for Irrotational Flows

Theorem from Vector Calculus

For any vector field f,
f = 0 There exists a scalar field such that f = .
Note that it is easy to show that f = implies f = 0.
The reverse implication (ie that a can be found) is proved by constructing in terms of the
components of u.
Hence for an irrotational flow, u, there exists a scalar field (x, y, z) such that
u = ,
and is known as the velocity potential.
Incompressible and Irrotational Flows
If the flow is irrotational, then u = . If the flow is also incompressible
u = 0, = () = 2 = 0,
so satisfies Laplaces Equation.
29

Example 24
If = xy + yz3, calculate u and show directly that flow is irrotational.
=

+
z = yx + (x + z3)y + 3yz2z .
x
x
y y +z

To show that flow is irrotational

y
z

u= x
= (3z2 3z2)x (0 0)y + (1 1)z = 0.
x y z
y x + z3 3yz2

Example 25
If = 3x2yz y3z calculate u and show that the flow is incompressible.
u = =6xyzx + (3x2z 3y2z)y + (3x2y y3)z ,
u=6yz 6yz + 0 = 0.
To show that flow is incompressible without first calculating u,
2 = 6yz 6yz = 0.

Example 25a
Given
u = 6xyzx + (3x2z 3y2z)y + (3x2y y3)z ,
calculate .
We have
u=

+
z = 6xyzx + (3x2z 3y2z)y + (3x2y y3)z ,
x x y y +z

and so

= 6xyz, = 3x2z 3y2z,

= 3x2y y3.
x
y
z
Integrating the first of these equations with respect to x we have
= 3x2yz + f(y, z),
where f(y, z) is an arbitrary function of y and z at this stage. Substituting into the second
equation,

3x2yz + f(y, z)
y

= 3x2z 3y2z, =

f
= 3y2z, = f(y, z) = y3z + g(z).
y

The function g(z) is determined by substituting = 3x2yz y3z + g(z) into the third equation,

3x
2 3 yz y z + g(z)
z

=3x2y y3,

g =3x2y y3,
3x2y y3 + z
30

and so g(z) is a constant, giving

= 3x2yz y3z + C, (compare with previous example)
(NB The velocity potential is only ever defined up to an additive constant. Whatever the value of
the constant chosen, u = is unchanged.)

31

Dynamics

4.1

Eulers Equation of Motion

Consider a blob of fluid occupying a small volume centred on the point P .

Acceleration of Fluid Blob =

Du
.
Dt

We then relate this acceleration to the forces acting on the blob using Newtons Second
Law

Mass x Acceleration=Total Force Applied.

We assume that the only forces applied are due to gravity and the pressure of the surrounding fluid,
and so

Du
dV =
Dt

pn dS
+

g dV,
V

p dV +

g dV,

using divergence theorem.

This is true for all volumes V and so

Du
=p + g,
Dt
u
1
+ (u)u= p + g.
t

g = gz

This is known as Eulers Equation (or Eulers Equation of Motion).

Notes:
1. We have ignored other forces acting on the fluid particles, in particular neglecting the effect of
viscosity (which can be thought of as friction between fluid particles due to their relative
motion). The full equation of motion including this effect and more general externally applied
forces is
u + (u)u = p + F + 2u,
t
where F is the body force per unit mass and is the coefficient of viscosity.
2. In many fluid flows viscosity is small. If the viscosity of the fluid is set to zero ( = 0), the flow
is said to be INVISCID, in which case Eulers equation applies. In this course we only
consider inviscid flows.
3. For an incompressible, inviscid flow we have four scalar equations
u
t

+ (u)u = 1p + g, (1 vector equation = 3 scalar equations)

u = 0, (1 scalar equation)

with four unknowns the three components of the velocity vector, and the pressure p.

32

4.2

Boundary Conditions

A fixed, impermeable surface is surface fixed in space through which no fluid can pass.
At a fixed, impermeable surface the normal component of fluid velocity on the surface is zero.
un = 0,

is normal to surface.
n

Hence flow must be tangential to a fixed, impermeable surface.

Example 26
If u = yx + xy show that the flow is incompressible and irrotational and satisfies the boundary
conditions on the impermeable surfaces y = x.

u = 0 + 0,

u = (1 1)z = 0.

The normal to the line y = x is

1
= (x + y ),
n
2
since the normal must be perpendicular to x + y (the tangent to the line).
Then on y = x,
un = x(x +
1
y ) (x + y ) =2 0.
Similarly can show that normal to other surface is parallel to x + y .

4.3

Bernoullis Equation

In an earlier section we suggested that fluid flow must obey some form of energy equation. We wrote
down
p + 12 |u|2 + gz = constant,
without fully explaining how the contribution to the energy from pressure arises, and without
explaining when the equation is valid.
33

We now derive this result for an inviscid flow starting from Eulers equation.
We first note that from 1, for any vector field f,
1 f2
2

f ( f) =

(f)f.

Hence setting f = u (fluid velocity) we have f = (vorticity) and

1 u2
2

(u)u =

u ,

and Eulers equation becomes

u
u =
t

p + 12 |u|2 + gz .

We now consider two separate

cases: Irrotational Flow
4.3.1
In this case = 0 and we can write the velocity field in terms of the velocity potential u = ,
so we obtain
p
()
+ 1 ()2 + gz .
=
t
2
Or equivalently
p
+ + 12 ()2 + gz
t

= 0,

which implies that

p + + 12 ()2 + gz
t

= constant,

throughout the fluid.

If instead we consider a general conservative body force, this is one such that F = then
p + + 12 ()2
t
Gravity is a conservative body force F = gz , = gz.

34

= constant.

(4.3.2)

Example 27
If u = (etx, ety, 2etz) and p = p0 at the point (0, 0, 0), show that the flow is irrotational and
incompressible, and find the pressure everywhere, in the absence of a body force.
y
u=et

x
x y z
x y 2z

= 0,

u=et(1 + 1 2) = 0.
Hence the flow is irrotational and incompressible. Next we must construct the velocity potential,
= 12 x2et + f(y, z, t),

etx= x =
=
e y= y
t

f
+ 0=ety,
y
f(y, z, t)= 12 y2et + g(z,
t),

=
2e z = z

= 12 x2et + 21y2et + g(z, t),

+ 0=2etz,

z
g(z, t)=z2et + h(t).

Combining these results gives

=

1 2
x + y2 2z2 et + h(t).
2

(NB. We can show that 2 = 0) The

unsteady pressure equation gives
1
+ F (t),
2 ()
t 2
1
1
= 0 (x2 + y2 + 4z2)e2t + (x2 + y2 2z2)et
2
2

p=

And p = p0 at (0, 0, 0) which fixes F (t) h (t) = p0.

35

+ F (t) h (t).

4.3.2

Steady Flow

If the flow is steady we have

p + 12 |u|2 + gz .

u =
Taking the dot product of this equation with u,

u p + 12 |u|2 + gz

= 0.

Recalling that is normal to the line =constant, we see that vector u is tangential to the line
=constant where
p
= + 12 |u|2 + gz,

and so this line is a streamline. Hence

p+

1 |u|2
2

+ gz = Constant along a Streamline,

for any steady flow. However if the vorticity is non-zero, then the constant is different along different
streamlines.

36

4.4

Momentum Integral Theorem for Steady Flows

Starting from Eulers equation for a fluid in a gravitational field

u + (u)u
t

= p gz .

We can integrate over a fixed control volume V to give

u
dV =
t

(p + (u)u + gz ) dV
V

pn dS (u)u dV Mgz ,

=
S

M is mass of fluid in volume.

If we now consider a steady flow with a constant everywhere,

pn dS = (u)u dV Mgz .
S

For any scalar field and any vector field f such that f = 0,
((f) f) dV,

f dV =
V

using divergence theorem.

fn dS,
S

From this it follows that for any g and f such that f = 0,

g(fn ) dS,

(f)g dV =
V

(taking the x , y and z components separately).

Thus for a steady flow with constant density, for any fixed volume
pn dS = u(un ) dS Mgz ,

F =
S

where F is the total force on the fixed volume.

NB The total force on a volume due to a constant pressure is zero since
pa dV = 0,

pan dS =
S

for constant pa.

Hence we can set atmospheric pressure to be zero when using this result for real problems.

37

Example 28
Consider a jet of water directed at right angles to a flat plate.
If the flow rate and velocity of the incoming jet are Q and U respectively, calculate the force on the
plate, ignoring the effect of gravity.
Let area of incoming jet be A and assume that the fluid velocity through this incoming jet is
constant, so Q = UA and the pressure in the jet is approximately constant and equal to atmospheric
pressure (which we take to be zero).

We split the surface of the control volume into 4 parts:

1. Circle with area A through incoming jet far from plate

2. The curved surface of the jet

3. A disc of radius R on the plate

4. The rim of the disc, radius R through the fluid spreading over the plate.
We take the radius of the disc large enough that the flow is virtually parallel to the plate as it passes
out of the control volume through surface (4).
Considering each of the surfaces in turn,
On 1
On 2
On 3

pn dS =0
(since p = pa = 0)
2
u( n ) dS =AU x (since u = Ux , n = x )
upn dS =0
(since p = pa = 0)
(since un = 0)
u( n ) dS =0
upn dS =F x
(F is force on disc, tends to total force on plate as R .)
u(un ) dS =0
(since un = 0)

On surface (4), the contributions of both integrals are zero by symmetry.

Hence using the momentum integral theorem we have
F = AU2 = QU.

38

4.5 Calculating Velocity Potential for general flows

If the flow is irrotational and incompressible, then we obtain Laplaces Equation,
2 = 0,
this follows since 0 = u = () = 2. So to determine we must solve this equation with the
appropriate boundary conditions.
Boundary conditions on
For a solid, fixed boundary un = 0, and so

=0
n

on S,

using the identity n = /n.

For example, if the surface is a cylinder, radius a, centred at the origin, then the normal to the
surface is in the radial direction and the boundary condition becomes

=0
r
4.5.1

on r = a.

Uniqueness of velocity potential

1. Uniqueness theorem
There cannot be two different forms of irrotational motion for a confined mass of fluid with
prescribed velocities on the boundaries.
Proof
Assume there are two different solutions 1 and 2. On the boundary surface S,
1
2
=
,
n
n
as
nis prescribed.
Let = 1 2, then
2 = 21 22 = 0 0 = 0.
i.e. 2 = 0 in V and

1
2

= 0,
n
n
n

on S. Now note
() dV =
V

() n dS by Divergence theorem,
S

( n ) dS,

=
S

dS.
S n

Note that
() = () + = 2 + ()2,
39

and so
2 dV +
V

()2 dV =

i.e.
V

dS
S n

()2 dV = dS,
V S n

2 dV, Greens theorem.

V

Thus,

()2 dV =
V

dS
n

2 dV,
V

=0.
As ()2 0 this integral implies that = 0.
= constant,
1 = 2 + constant.
i.e. solutions 1 and 2 are identical (to within an arbitrary constant) and hence the solution
for u is unique.
Corollary
If the boundaries S are at rest then irrotational motion is possible.
Proof
So

=0 on S
n
2=0 in V.
Therefore a solution is = constant, but by the above theorem this is the unique solution.
If = constant = u = = 0 so no flow.
2. Theorem
If there is a given distribution of vorticity in the fluid and the normal boundary velocities are
given then the flow is uniquely determined.
Proof
Assume there are two possible flows u1 and u2 (No potential as = 0). Let
u = u1 u2 then

u = u1 u2 = 0,

by conservation of mass for an incompressible fluid.

u = u1 u2 = 1 2 = 0,
40

as prescribed vorticity. Therefore u = and then 2 = u = 0.

On the boundary u n = u1 n u2 n = 0 as given on the

= 0 on S.
boundary.
n
Also 2 = 0 in V . Therefore represents an irrotational motion with boundaries at rest.
Hence by the above corollary, = constant in region V .
u = = 0,
and so
u1 = u2. i.e. unique.

41

Channel Flows

5.1 Steady flow under a Sluice Gate

Consider steady flow of depth h1 and flow speed u1 (independent of depth) far upstream of a sluice
gate.
We assume that downstream of the sluice gate the flow has depth h2 and uniform flow u2.

By mass conservation
u1h1 = u2h2.
Applying Bernoulli along the surface streamline we have
1 2
2u 1

+ gh1 = 21u22 + gh2,

since the pressure is atmospheric on the free surface.

We will use the momentum integral theorem, on the volume indicated so need to find pressure on
the surfaces (1) and (5). On surface (1) the flow is uniform and hence irrotational, so we can apply
the unsteady Bernoulli equation (4.3.2),
p+

1 u2
1
2

+ gz = pa +

1 u2
1
2

+ gh1 , = p = g(h1 z),

taking pa = 0 as explained earlier. Similarly on (5), p = g(h2 z).

Considering each of the surfaces in turn, we consider the component of
(u(un ) + pn ) dS,
S

in the x direction.

42

pn dS =x 0h1 g(h1 z) dz = 12 gh21 x

On 1

u(un ) dS =h1u21x
On 2 and 4

(since u = u1
n 0)
(since p = pa =

pn dS =0

(since un =x,0)

u(un ) dS =0
On 3

pn dS =T x

(T is force on gate.)

u(un ) dS =0

(since un = 0)

pn dS =x 0h2 g(h2 z) dz = 1 2 gh22

On 5

u(un ) dS =h2u22x
On 6

= x )

x
(since u = u2

=x )

(since n = z )
n

pn dS =0

u(un ) dS =0

(since un = 0)
x,
Substituting these results into the momentum integral theorem we find that the total force on the
gate is
T = 12 g(h21 h22) + h1u21 h2u22 .

43

5.2

Hydraulic Jumps

In many flows we see a rapid change in the surface accompanied by turbulent flow. For example,
water from a tap falling onto a horizontal plate forms a disc of shallow fast flow with a jump in the
surface at a particular radius. A similar effect is often seen at the bottom of weirs or downstream
of sluice gates.

These are known as hydraulic jumps and typically stay in a fixed position. When a strong tidal
surge enters a narrowing river estuary a similar jump occurs, but moves upstream.

These are usually known as hydraulic bores, but much of the theory is the same for the two cases.
Working out the position of such a jump depends on the physical problem and is typically very
complicated. Here we only consider the relationship between the fluid velocity and water depth
either side of the jump, using methods similar to the example of flow underneath a sluice gate.

44

The equation of mass conservation gives u1 h1 = u2 h2 .

Applying the momentum integral theorem, there is no external force acting on the water (compare
to sluice gate problem, where the gate exerted a force on the fluid). Hence
1
g(h21
2

h22 ) + h1 u21 h2 u22 = 0.

However, in this case we can not apply Bernoullis equation along the surface streamline, since we
know that the flow is turbulent in the region of the jump, and so energy is lost by the fluid (lost
in form of sound or heat).
Combining these equations we can obtain u2 and h2 in terms of u1 and h1 .

5.3

Flow over obstacle on river bed

Consider a flow of depth h and speed U encountering a small obstacle on the bed given by z = hb(x),
with 0 b(x)
1 and b(x) 0 as |x| .
Let the surface of the water be given by z = h + hs(x), with |s(x)|
1 and s(x) 0 as x .

Letting the velocity at x be u(x), the depth at x is h + hs(x) hb(x), so conservation of mass gives
U h = u(x)h (1 + s(x) b(x)) ,
45

= u = U (1 + s b)1 .

Applying Bernoulli on the surface streamline,

pa +

1 2
U
2

+ gh = pa +

1 2
u
2

+ gh(1 + s) ,

since the pressure is atmospheric at the surface. Eliminating u,

U 2 + 2gh = U 2 (1 + s b)2 + 2gh(1 + s),
where

U
F = ,
gh

= F 2 = F 2 (1 + s b)2 + 2s,

is known as the Froude Number.

Since s and b are both small, we can use the binomial expansion,
(1 + s b)2 = 1 2(s b) + . . . .
to finally give
F 2 = F 2 (1 2s + 2b + . . .) + 2s,

= (F 2 1)s = F 2 b.

Since b > 0, if F > 1 corresponding to a fast, shallow incoming flow, then s > 0 and the flow rises
over the bump. If F < 1 (a slower deep flow), the flow dips over the bump.

F >1
Supercritical

F <1
Subcritical

46

Two-dimensional Flows

When solving two-dimensional flow fields,

u = u(x, y)x + v(x, y)y ,
simplifications can be made to the analysis. In particular, we can find an easier way of calculating
streamlines.

6.1

The stream function

For any 2-D incompressible flow, we can write the velocity in the form
u=

,
y

v=

,
x

where (x, y) is known as the Stream Function.

NB
1. This definition is equivalent to u = ((x, y)z ).
2. The property of incompressibility is immediately satisfied,
u =

6.1.1

= 0.

Properties

1. =Constant is a streamline.
= (x,y, 0) = (v, u, 0).
Hence,
u= 0.

(6.1.3)

Now, is always perpendicular to the line=constant (see MAT1005). Also, the vectors ,
u and the line=constant all lie on the x, y plane.

47

From (6.1.3), u is perpendicular to and hence

u is tangent to =constant.
2.

1
2

|| = (y2 +x2)2 = (u2 + v2)

= |u|.

3. The magnitude of u is inversely proportional to the distance between streamlines.

= c1

u1

d1

d2

u2

= c 2

1
d
This can be stated more precisely. For any line, , joining two points x and y,
Conservation of mass = u1d1 = u2d2, i.e. u .

Flux of volume across =(y) (x).

Proof

Flux=

un dl,

,
y
x

dl = (dy, dx),
n

(dy, dx) =

dy +
dx ,
y
x

d=(y) (x).

4. In plane polars,(r, ),
u = ur + u ,
r

ur =

48

1
,u =
r

.
r

6.1.2 Governing Equations and Boundary Conditions

Recall, for a two-dimensional flow, the vorticity is given by
= 0, 0,

v
x

u
y

so we can write = z , where

v
u

.
x y

Substituting the expressions for u and v in terms of from (6.1) we have

=

2
2

,
x2
y2

and hence, the stream functionsatisfies,

2= .

Boundary Condition on
Solid fixed boundary

6.1.3

=
=
=

No flow normal to surface,

Surface is a streamline,
=Constant on surface.

Examples

Given the velocity potential , find the stream functionand plot the streamlines.
(29)

1 2
(x y2),
2

(30)

K
log r.
2

Example 29
=
Also, u =

,
y
x

1 2
(x y2),
2

u = (u, v) = = (x, y).

and so

= x
y

v=
= y
x

u=

=xy + f(x),
=xy + g(y),

where f(x) and g(y) are undetermined functions. However, for both these results to be true
f(x) = g(y) =constant,
= xy + constant.
An alternative method is to show that= xy + f(x) from the first equality, then substitute this into the
second equation, hence proving that f (x) = 0.
Streamlines are given by =Constant. Hence the streamlines are rectangular hyperbola, xy
=Constant.
49

For the direction of flow along the streamlines, use the expression for u. In this case u = x, so u > 0
when x > 0 and u < 0 when x < 0. Always try to mark direction of flow on streamlines with arrows.
NB Same flow is possible if any streamline is replaced by a solid, fixed boundary. In this case, x = 0
is a streamline, so the flow in the upper half plane is like that due to a jet directed normal to a flat
wall (x = 0).

Example 30
A 2-D source, using polar coordinates (r, ).
K
log r,
2
K
.
u = =
2r r
=

Writing the velocity in terms of the stream function,

u=
Hence

1
K
=
r 2r

.
r r r

K
=
(+Constant).
2

=0
r
The streamlines,=Constant, correspond to the lines =Constant, which are straight lines directed
out from the origin.

50

Flow directions on the streamlines are marked for K > 0, ie outward flow from a source.
The flux across the circular contour r = a is given by
K
2
2
d = K.
Flux = u n dl = r ra K
d =
2 0
0
2a
So the velocity potential represents a source of strength K at the origin.

6.2

Line Vortices

6.2.1 Velocity potential of line vortex

Consider a flow with velocity potential
=

K
,
2

K = constant.

Then

1
+
,
r
r
r
K
=
.
2r

u=

Hence the velocity is all in the direction and |u| as r 0.

Calculating the stream function, since
u=
we have

,
r r r

1
= 0,
r

and hence

K
=
,
r
2r

K
log r.
2
The streamlines are given by=constant which corresponds to r =constant, i.e. concentric circles
centred on the origin. This flow is described as being due to a line vortex at the origin. For the case
K > 0 the streamlines are:
=

51

6.2.2

Circulation

Definition
Given a velocity field u and a curve C in space, the circulation of the flow around the curve C is
defined to be
=
udl.
C

Example 31
For the line vortex
with

K
,
2r
calculate the circulation around the curve C given by r = a (ie a circle radius a centred on the
origin).
On C we have dl = (a d) , and so
u=

2
0

K
2a

(a ) d = K.

Thus the circulation round a circle centred on the origin is independent of the radius of the circle.
The velocity potential

= , = constant.
2
represents flow due to a vortex of strength (or circulation) at the origin.

6.2.3

Relation between vorticity and circulation

For a 2-D flow, consider a closed curve C on the x, y plane, with A the area enclosed by C, using the
definition of circulation and Stokes Theorem
=

( u) z dS =

udl =
C

dS,
A

since for a 2-D flow = u = z .

52

Example 32
Consider line vortex strength 1 at origin. Flow is irrotational except at the origin where |u| .
Hence by previous result the circulation round any curve not including the origin is zero.
We showed in previous example that circulation round circle of radius a is 1 .
Hence we can consider the line vortex to be the limit of a circle of radius a and vorticity 1 with
a 0 and 1 , with a2 1 = 1 .
This serves as a model for a concentrated patch of vorticity as may be seen in a river flow.
NB
1. It can easily be shown that the circulation round any simple closed curve surrounding the a
vortex of strength 1 at the origin is 1 .
2. For a system of line vortices of strength i located at (xi , yi ), then the circulation round a
curve C is
=
i ,
sum over vortices inside curve C.
The analogue of circulation in 3D is, given curve C in 3-D space, spanned by any surface S, the
Circulation round C is
= udl =
n dS.
6.2.4

Kelvins Circulation Theorem Inviscid Flows

Consider a flow with constant density and let C be a closed simple curve at time t = 0. As time
increases the particles lying on the curve will move we let the C(t) be the curve described by
these particles at time t.
Kelvins Circulation Theorem states that the circulation round the curve C(t) is independent of
time.
For the proof of this theorem see appendix D
6.2.5

Significance of Irrotational Flows

If at some time a flow is irrotational, and the subsequent flow is inviscid, then the flow will be
irrotational at later times.
Proof
Consider a small closed curve C at time t = 0. Since the flow is irrotational at t = 0, then the
circulation round C is zero. Since the circulation is constant, the circulation round the curve
carried with the fluid is zero at later times and so the vorticity inside the small curve is zero.
Note
Including viscosity can lead to generation of vorticity in a flow. For example consider a flat plate
in a uniform flow initially in line with flow, then rotated. Vortices shed due to viscous effects at
sharp edges. Same principle for spoon impulsively moved in liquid.

6.3

Combining simple examples of 2D velocity potentials

As the equation of motion (ie Laplaces Equation) is linear, we can superpose solutions. We can
make more complex flow fields by combining some of velocity potentials which we have already
seen. Below are just a few examples, and more will be seen during the course.
53

1. A source at the origin in a uniform flow

= Ux +

k
log((x2 + y2)1/2).
2

2. A source at (x0, y0) in a uniform flow

= Ux +

k
log(((x x0)2 + (y y0)2)1/2).
2

i.e. shifted the origin.

3. Source at r1, strength k and a sink at r2, strength k.
k
log R1 (Source at r1) R1 = |r r1|
2
k

log R2 (Sink at r2) R2 = |r r2|.

2

Here r is a general position vector, e.g. r = xx + yy in Cartesians.

Example 33
Consider a source, strength K at (x0, 0) and a sink strength K at (x0, 0).
Write down the velocity potential (x, y).
Find the velocity at (0, 2x0).
Find the approximate value of (x, y) as x0 0.
Velocity potential
=

K
K
log (x x0)2 + y2
log (x + x0)2 + y2 .
4
4

Velocity is given by u = so
u=

K
x + x0
x x0
K

2
2
x
2 (x x0) + y
2 (x + x0)2 + y2
y
K
y
+ K

y .
2 (x x0)2 + y2
2 (x + x0)2 + y2

So at (0, 2x0)
u=

K
2

x0

5x20
K
=
5x0 .x

x0
5x20

K
+
x 2

2x0
5x20

2x0
y ,
5x20

As x0 0
(x x0)2 + y2 =(x2 + y2) 2x0 + x20,

x20
2x0x
2
2
+
=(x + y ) 1 2
x + y2 x2 + y2
54

Similarly
(x + x0)2 + y2 = (x2 + y2)

1+

x20
2x0x
+
x2 + y2 x2 + y2

So,
=

K
x20
2x0x
log(x2 + y2) + log 1 2
+
x + y2 x2 + y2
4
K
x20
2x0x

+
log(x2 + y2) + log 1 + 2
x + y2 x2 + y2
4

Using log(1 + X) X + ...... for |X|

1 this simplifies to

K
2x0x
2x0x
2
2
2
x +y
x + y2
4
Kx0
x

.
2
x + y2

+ ....

6.4 General solution of velocity potential in plane polar coordinates

Consider a velocity potential (r, ),

1
r
= r+
r= 0. r
2

1 2
r2 2

This can be solved using separation of variables. Try = f(r)g(),

gf
(rf ) g
+ 2 =0,
r
r
r(rf )
g
+ =0.
f
g
The first term depends only on r, the second only on and hence each term must be equal to a
constant (independent of r and ). So
g
= 2,
g

g = A cos + B sin ,

constant

but g() must be 2-periodic and so = n, an integer. [NB If g /g is taken as equal to a positive
constant, g() would not be 2-periodic.]
The r equation then becomes
r(rf ) n2f = 0.
Trying f rm we obtain m2 = n2, ie m = n. The general solution is then given by

rn (An cos n + Bn sin n) + rn (Cn cos n + Dn sin n) ,

(r, )=
n=1

+A0 + B0 log r,
where the final two terms correspond to an additional flow due to a line vortex and a point source.
55

Example 34
Find the velocity potential describing flow past a cylinder r = a, with zero circulation, subject to
uniform flow speed, U, far from the cylinder.
Boundary conditions
Ux = Ur cos ,r

=0
r = a.
r
Using the boundary condition at r , together with the zero circulation condition, the only term
with a positive power of r is Ur cos and so

(r, ) = Ur cos +

(Cn cos n + Dn sin n) .

n=1

Applying the boundary condition at r = a gives

C1 = Ua2, D1 = 0, Ci = Di = 0 (i 2). Hence

=U

r+

a2
r

cos .

(6.4.4)

6.5 Complex Potentials for incompressible, irrotational 2-D flows

6.5.1 Definition
For a two-dimensional flow which is both incompressible and irrotational, we have seen that

,
=u=
x
y

.
=v=
y
x

(6.5.5)

These equations are known as the Cauchy-Riemann equations (see MAT3034, Functions of a
complex variable, if selected next year).
As andsatisfy these equations, we can construct a complex function
w(z) = + i,

z = x + iy

(6.5.6)

known as the complex potential. This function is an analytic function of the complex variable z,
which means it has a derivative at every value of z. Some complex functions are not analytic and
some points in the complex plane have a different derivative depending upon how you approach that
point.
NB
1. Any analytic function satisfies the Cauchy-Riemann equations.
2. Throughout this section
z = x + iy NOT spatial coordinate in third direction, w
= + iNOT velocity component in third direction.
3. Any analytic function represents some two-dimensional, irrotational, incompressible flow, but
this flow is often physically unrealistic.
56

6.5.2

Boundary Conditions on w(z)

On a solid boundary, =Constant.

Hence,

6.5.3

Im(w) = Constant

on solid boundary.

(6.5.7)

Velocity in terms of Complex Potential

It is possible to show that the derivative of a complex function can be written in terms of its real
and imaginary parts,

dw
=
+i .
x
x
dz
As the function is analytic we can take the derivative along the x-axis. Thus
dw
= u iv,
dz

(6.5.8)

is the key result to remember. Hence, if w(z) is given, the velocity components u and v can be
obtained immediately, without explicitly calculating or in terms of x and y. The magnitude of
velocity at any point is given by

dw
= u2 + v2 = |u|.
dz

6.5.4

(6.5.9)

Simple Examples

Here we choose some elementary functions for w(z) and consider the resulting flow field. Hopefully
this provides us with the necessary insight in order to work out the required complex potential for
more complicated flows.
Example 35

w = Az, A = Uei, U, real.

u iv =

dw
= A = U(cos + i sin ).
dz

Hence u = U cos and v = U sin and w(z) represents uniform flow, speed U, inclined at angle
to the positive x direction.
Alternatively we can obtain the velocity potential and stream function from the complex potential,
+ i=w = Ueiz,
=U(cos + i sin )(x + iy),
=U ((x cos y sin ) + i(y cos + x sin
)) ,
=U(x cos y sin ),
=U(y cos + x sin ).
The streamlines given by=Constant, have equation
y = x tan + C.
57

The directions on the streamlines are for U > 0, in which case the velocity in the x-direction is
positive everywhere.

Example 36

w = Bz2,

B is real and positive.

+ i= w = B(x + iy)2 = B(x2 + 2ixy y2).

Hence
= B(x2 y2),= 2Bxy,
and the streamlines are given by xy = c (c a constant). ie rectangular hyperbola as in 6.1.3,
Example 29.

Example 37
w = Bzn,
B is real.
This time it is easier to calculate the flow field using polar coordinates, and hence we write z = rei.
+ i= w = Brnein = Brn(cos n + i sin n),
58

which gives
dw
= Bnrn1.
dz
n
The streamlines are given by r sin n =Constant. If we now consider the streamline= 0, then
2
= 0 = sin n = 0 = = 0, , . . . ,
n
n
and hence these streamlines are straight lines. We now let = /n and restrict ourselves to the
region of the plane 0 .
Since = 0 and = are streamlines, we can replace them by solid boundaries and flow is
unchanged. The resulting flow field depends on the value of n chosen.
= Brn sin n, |u| =

(a) n > 2

0 < < 2

This gives the streamlines for flow into an acute corner. The direction of the streamlines
corresponds to B < 0 (since u < 0 on = 0).
As r 0, |u| = Bnrn1 0. Hence, O is a stagnation point.
(b) 1 < n < 2

<<

59

This corresponds to flow in the vicinity of an obtuse angle. As above, O is a stagnation

point.(c) 1 < n < 1 < <
2

This corresponds to flow outside of a corner.

Since n < 1, rn1 as r 0, hence this flow has infinite velocity at the point O.(d)
n = 1 = 22

This corresponds to flow round the edge of a semi-infinite plate. As before, infinite velocities
would occur at the tip of the plate.
The appearance of infinite velocities in these last two examples is clearly physically unrealistic.
Indeed, using Bernouillis equation, we see that if |u| , then p and so we have infinite
suction. However, the solutions obtained are still useful away from the vertex. Close to the vertex,
we can not ignore viscous effects.
60

Flow (d) is particularly useful, corresponding to the limiting case of flow round a rounded edge. This
is important in aerofoil theory.
Also the solutions above are useful as local solutions near corners in more complicated flows.
Example 38

w = B log z,

B is real and positive.

+ i= B log(rei) = B(log r + i).

[NOTE: This is a general result log(z) = log |z| + i arg(z).]
Hence = B log r,= B and the streamlines are given by =Constant, ie lines radially outwards
from the origin. also,
dw
B
|u| =
=
.
dz
|z|
Comparing with that in Example 30 in 6.1.3,
w=

K
log z,
2

is a source, strength K at O.

Example 39 w = iB log z, B is real and positive.

Following example 38, = B,= B log r and the streamlines are r =Constant, ie concentric
circles centred at the origin. Comparing with the Example in 6.2.1,
w = i

log z,
2

is a line vortex at O, circulation .

NB All these examples can be generalised, eg.

K
w = 2 log(z z0),
is a source at z = z0.
6.5.5

Combining Simple Examples

As with the velocity potentials in 6.3, we can superpose solutions.

Example 40

Source in Uniform Stream

K
log z,
2
K
=Uy +
,
2
K
Cartesian
=Uy +
tan1 y ,
x
2
K
,
Polars
=Ur sin +
2
dw
K
.
=U +
u iv =
dz
2z
w=Uz +

61

At a stagnation point the velcity is zero,

= z =

u=v=0

K
.
2U

Hence the stagnation point is located at

K
,0 .
2U

The streamlines are given by

y=

K
( ),
2U

= Constant.

Plotting the streamline passing through the stagnation point is usually a good idea for visualising
the flow. In this case, the stagnation point is at = , y = 0 and so = . The easiest way to sketch
this streamline is to tabulate y for various ,
0

y 12 K
U

3
4

3K 1K 1K
8U 4U 8U

3 2
0 14

KU

Construction of further streamlines gives a flow field like,

62

Since streamlines can be replaced by solid surfaces, the same flow is obtained if a solid body is
inserted inside the streamline= K/2U.

Example 41

Dipole or Doublet

Consider a sink, strength K at z = 0, and a source of strength K at z = ei,

K
K
K
i
w = log(z
2 e ) log
2z = log 1
2.

ei
z

For small ,
log 1

ei
z

ei
,
z

and so, if 0 and K in such a way that K is kept fixed, then

w=

ei
,
z

K
.
2

This field is known as a dipole (or doublet) of strength , orientation

ei

= Im(w) = Im = sin(
).
i
re
r
The streamlines are then given by
r = sin( ),

= Constant.

Exercise: Show that when = 0, the streamlines are circles, centred on the y-axis.
We can then show that the pattern of streamlines is given by,

63

Example 42

Uniform Flow, Dipole and Line Vortex.

w=

Uz

Ua2
z

i
log z, real
2

Line Vortex
at Origin

Uniform Dipole
Flow
at Origin

This corresponds to flow around a cylinder with circulation .

See 6.6.

6.6

Calculating Forces on Body using Pressure

Recall from 2.2 that the force on a body is given by

F =p S (n ),
F = pn dS.
For two-dimensional flows this simplifies to
Force =

pn dl.

(6.6.1)

For irrotational flows, the pressure is then calculated using the unsteady Bernoulli equation (4.3.2)
or by using Bernoullis equation along the streamlines which lie on the surface of the body.

Example 43
Consider a cylinder, radius a, in uniform flow (zero circulation), in a gravitational field. If the
cylinder is centred on the origin, and p p0 on y = 0 as |x| , find the pressure on the cylinder.

64

From (6.4.4), the flow is given by

=U

a2
r

r+

cos ,

or in terms of the complex potential

Ua2
w = Uz +
.
z
The streamlines look like

The speed anywhere in the fluid is given by

|u| =

dw
a2
=U 1 2 .
dz
z

Far from the cylinder, |z| and |u| U. On the cylinder, z = aei and
1
2
= 2U sin .
|u| = U 1 e2i = U (1 cos 2)2 + sin2 2
Using Bernoulli on the streamline which passes along y = 0 and the surface of the cylinder,
p0 1
1
p + |u|2 + gy,
+
2

2U = 2
Conditions at y = 0, x , Conditions on surface of cylinder.

Hence, the pressure on the surface of the cylinder is given by

1 U2 2U2 sin2 ga sin
2

p = p0 +

On the surface of the cylinder, the normal is in the radial direction,

element of arc length is dl = ad. Hence the force on the body is
F =

pn dl =

.
=n
cos x + sin y and the

p(cos x + sin y )a d,

=(p0 +

1 2
U )a
2
2

+2U2a

(cos x + sin y )d

sin2 (cos x + sin y

+ga2

sin )d
(cos x + sin y )d.

65

Evaluating all these integrals, the only non-zero contribution comes from
hence
F = ga2y ,

2
0

sin2 d = and

which is the buoyancy force (independent of fluid flow).

Important: There is no force in the x-direction ie No Drag.
NB
In this case we could guess that there is no drag on a cylinder in inviscid flow by fore-aft symmetry.
BUT a more general result exists:
Any body placed in a uniform, inviscid flow with zero
circulation experiences no drag.
This result is known as DAlemberts Paradox.

Example 44
Find the force on a half cylinder attached to a plane wall, normal to a uniform flow.

See Problem Sheet.

66

Example 45

Cylinder in Uniform Flow with circulation

Ua2
i
w = Uz + log z.
z
2

Sketch the streamlines for various values of and calculate the force on the cylinder. I
Show that z = aei is a streamline.

i (log a + i),
w=Ua(ei + ei) 2

log a
= 2Ua cos +
i
.
2
2
Hence Im(w) =Constant, and the circle |z| = a is a streamline. II
Show that the flow is uniform far from the cylinder.

u iv =

dw
Ua2
i 1
,
=U 2
dz
z
2 z
U

as z .

III Find the Stagnation points in the flow. At the

stagnation points, u = v = 0 and hence dw

dz

z2 2aiz a2, = 0,

= 0. Solving for z,
z = a i

1 2 ,

where = /4Ua. Thus there are two stagnation points z1 and z2. From the quadratic equation for
z, the product of the roots is a2,
z1z2 = a2 = |z1||z2| = a2,
and so, either one stagnation point is inside the cylinder and the other is outside, or both lie on
the surface |z| = a.
We now consider the separate cases:
=0
There is no circulation and the stagnation points are at a.
2

(i)

2
2

0
x

67

[NB These plots are obtained using MATLAB. See Appendix E. The streamlines inside the
cylinder are also plotted, though they are not relevant to flow considered here.]
(ii)

4Ua < < 0

In this case 1 < < 0 and
|z1,2| = a 2 + (1 2)

1
2

= a,

and both stagnation points lie on the surface of the cylinder.

1.5
1

0.5
0
0.5
1
1.5
2
2

(iii)

0
x

= 4Ua ( = 1)
In this case there is only one stagnation point, z1 = z2 = ai.
1.5
1

0.5
0
0.5
1
1.5
2
2

0
x

< 4Ua
In this case, < 1,

z = a 2 1 i,

so both stagnation points lie on the imaginary axis, but only one is outside the cylinder.
1.5
1
0.5
0
y

(iv)

0.5
1
1.5
2
2

0
x

68

IV
Find the force on the cylinder
On the cylinder z = aei, so
2

|u|2 = U Ue2i 2Uiei ,

=U2 |(1 cos 2 2 sin ) + i (sin 2 2 cos )|2 ,
=U2 (1 cos 2 2 sin )2 + (sin 2 2 cos )2 ,
=4U2 sin2 2 sin + 2 .
Using Bernoulli along a streamline from which encircles r = a (neglecting gravity)
p0 + 12U2 =p + 2 U2 sin2 2 sin + 2 ,
2
2
2
p=p0 + 1U
2 1 4 4 sin + 8 sin

Therefore,
pn dl,

F =

p (cos x + sin y ) a d,

=4U2ay
, =Uy .

So
(i) Positive circulation

= Downward Force

(ii) Negative circulation

= Upward Force (Lift)
A physical illustration of these effects is seen when balls are spinning (eg Table Tennis).
Top Spin balls drop shorter case (i)
Back Spin balls carry further case (ii).
In football, side spin allows balls to be bent round a defensive wall.
Swing of cricket balls is considerably more complicated.

6.7
6.7.1

Induced Velocity and Image Systems

Induced Velocity

Consider two vortices, one of strength 1 at z = 0, and one of strength 2 at z = 2,

w=
u iv =

dw
=
dz

i1
log z
2

i1 1
2 z

i2
log (z 2),
2

i2 1
.
2 z 2

Velocity due to Velocity due to

vortex at z = 0 vortex at z = 2
= (u1 iv1) + (u2 iv2).
As z 0, |u1| and u2 (0, 2/4).
69

The velocity u2 is known as induced velocity of the vortex at z = 0 due to the vortex at z = 2. If
the vortices are free to move through the fluid, then each vortex moves with the velocity induced
by the other vortex.
For example, if 1 , 2 > 0,

For pairs of vortices, the induced velocity is always perpendicular to the line between the vortices.
If 1 = 2 = , the two vortices move on a circular path.

If 1 = 2 = > 0,

70

In this case the separation of the vortices stays constant, and the vortices move as a pair with speed
/2a, where a is the distance between the two vortices. This explains why two vortices generated
by a spoon in a cup of tea move across the mug.
These ideas can be extended to systems with more than two vortices. If n vortices are present, the
induced velocity at vortex 1 is due to the effect of vortices 2, 3, . . . n.
6.7.2

Image Systems

Consider a vortex of strength at z = ai and a vortex of strength at z = ai.

i
i
log(z ai) + log(z + ai),
2
2
i
log z ai ,
=
z + ai
2
dw
i
1
1
2a

=
.
u iv =
=
dz
2 z ai z + ai
2 z2 + a2
w=

(6.7.2)

On the line y = 0, z = x and

2a
u iv = 2 2 .
2 x + a
Hence v = 0 which proves that y = 0 is a streamline (no normal velocity component). As discussed
previously, we can replace the line y = 0 with a solid boundary.

71

The induced velocity at vortex 1 is

.
4a
Hence the flow due to a single vortex in the presence of a plane wall y = 0 is given by (6.7.2). The
vortex at z = ai is known as the image vortex. It is located at the mirror image of the original
vortex.
Two more general results extend the concept of image systems.
u iv =

A Schwartz Reflection Principle

Let w = f(z) describe a flow everywhere in the complex plane, with no singularities (points where
f(z) ) in y < 0. If a solid wall is now introduced at y = 0, the new complex potential is given by
w = f(z) + f(z),
where (.) denotes the complex conjugate.
This can be proved by showing that the line y = 0 is a streamline. Setting y = 0, z = z and
w = f(z) + f(z) = 2Re(f). Thus w is real, and so = Im(w) = 0, a constant on y = 0.

Example 46
From the example introduced above, without the wall, the complex potential for a vortex at z = z0
is
i
w = f(z) = log(z z0).
2
With the boundary included, the complex potential is given by
i
log(z z0)
2
i
=
log(z z0) +
2

w = f(z) + f(z)=

i log(z z0),
2
i
log(z z0),
2

where we have used the identity, log(Z) = log(Z).

Hence for a vortex at z0, the image vortex must be located at z0, ie at the mirror image location.

72

This method can also be used to find the flow due to a source near a wall, or for more than one vortex
near a wall.

Example 47
Tea-cup
Vortices

The same motion will occur for a single vortex in a right-angled corner. In this case there are three
image vortices. See problem sheet.

B Images for Circle

There is an equivalent principle to the Schwartz reflection principle which defines an image system
for a two-dimensional flow in the presence of a circle, radius a, centred at the origin.
If w = f(z) describes a flow everywhere in the complex plane, with no singularities in |z| < a, then
w = f(z) + f
73

a2 ,
z

is the complex potential for the flow in the presence of the circle.
Proof:
Put z = aei and show that Im(w) = 0, to demonstrate that the surface of the circle is a streamline.
Hence for a source or vortex at z = z0, (|z0| > a), an image is located at z = z1, where z1 = a2/z0.
Clearly |z1| < a and the image lies inside the circle.

Example 48
If the flow is uniform in the absence of the circle, at angle to the positive x-axis, w = Ueiz, where
U is real. With the circle present,
Ueia2
w(z) = Ueiz +
z

= Ueiz +

Ueia2
.
z

This can be compared with earlier examples.

Example 49
If a vortex is located at z = 2a, with a circle |z| = a, then
w=

i log(z 2a) i log a2 2a

z

i
2
i
u iv=
2
=

2
log(z 2a) log a 2a
z
1
1
2a
+ + 2
.
z 2a z a 2az

,
,

Hence the induced velocity at the vortex located at z = 2a is

u1 iv1 =

i
2

2a
1 +
2
2a a 4a2

74

i
.
12a

7
7.1

Surface Waves
Introduction

We now consider a fluid layer of average depth h with surface given by z = (x, y, t) and fixed
impermeable boundary z = h. We wish to consider how the position of the surface can change with
time eg move surface with a paddle (or hand)

7.2

Basic wave theory

We consider a 2-D problem with surface given

by

z = (x, t) = A cos(k(x ct)),

k known as wave number.

At t = 0, (x, 0) = A cos(kx)

Wave crests located at

x = 0,

2
,
k
75

4
,...
k

so distance between crests is

=

2
.
k

This is known as the wavelength.

The maximum height of the wave above its average value is A known as wave amplitude.
At later time, the crest originally at x = 0 is located at x = ct. So wave moves to right at
c.
c is known as wave speed or phase speed.
speed
Consider

=ckA sin(k(x ct))

t
= + c = 0,
t
x

=kA sin(k(x ct))

x
This is known as the 1-D wave equation for right travelling waves with speed c.
Note that a more general 1-D wave equation is
2
2
2

c
= 0,
t2
x2

with general solution

(x, t) = f(x ct) + g(x + ct),
corresponding to right and left travelling waves.
Angular frequency
An alternative way of writing the right travelling wave solution
is
= ck.
(x, t) = A cos(kx t),
Here is the angular frequency, and since (x, t + 2/) = (x, t), we say that
Time Period =

2
.

Note that in the case above we could have considered sin(kx t).

7.3

Potential Flow with a Free Surface

We return to the problem introduced above and take the flow to be irrotational so
2 = 0,

u = ,

where (x, y, z, t) is the unsteady velocity potential.

To solve this equation we need boundary conditions on . One boundary is the base z = h, the
other boundary is the surface, but the position of the surface is not known and emerges as part of the
solution of the problem. This is known as a free surface.
Dynamic Boundary Condition
By the unsteady Bernoulli equation for irrotational flows,
p + + 12 |u|2 + gz
t

= f(t),

but we can absorb f(t) into the definition of and so set f(t) = 0.
76

As explained earlier, we can take the constant atmospheric pressure to be zero, and so using
the unsteady pressure equation on the surface,
1
+ 2 ||2 + g = 0,
t

on z = (x, y, t).

Kinematic Boundary Condition

For a fixed impermeable boundary, we saw that the normal velocity on the surface was zero.
An alternative way of viewing this is that a fluid particle on a solid surface stays on the
surface. The same applies on the free surface of the fluid. Hence the vertical fluid velocity on
the surface must equal the vertical velocity of the surface
D
=z ,
Dt


+
+
= ,
on z = (x, y, t).
t
x x y y z
Base Condition
On z = h, the normal velocity is zero so,
= 0.
uz = z = z

7.4

Small amplitude disturbances

The equation 2 = 0 and the full boundary conditions can not be solved exactly, so we concentrate
on disturbances with small amplitude, (A h).
Both and are proportional to A so first we discard all terms which are quadratic in A, in the
kinematic and dynamic boundary conditions at the free surface

+ 2 || + g=0
= t + g=0,

t
on z = (x, y, t).

+
+
= , =
= ,
t
x x y y z
t z
Both of these conditions apply at the surface z = , an unknown location, but close to the fixed
surface z = 0. Using the Taylor expansions we re-write the boundary conditions so they apply at z =
0 rather than z = . If the boundary condition takes the form f(x, y, z, t) = 0 on z = , then
f(x, y, , t)=0,
f(x, y, 0, t) + fz(x, y, 0, t) + ...=0,
or at leading order f(x, y, 0, t) = 0.
Hence the free surface boundary conditions transferred to z = 0 become

+ g=0,
t

= .
t z
77

Thus the linearised problem is

2 = 0,

+ g = 0
t

and

= 0,
z

7.5

h < z < 0,

=
,
t
z

on z = 0,

on z = h.

Small-amplitude travelling wave solutions

We now look for a solution of the linearised problem in two-dimensions (no y-dependency) with
= a sin(kx t).

Thus the problem to be solved for is

h < z < 0,

xx + zz = 0,
with

=ag sin(kx t),

t

on z = 0,

=a cos(kx t),
z

on z = 0,

=0,
z

on z = h.

The boundary conditions on z suggest that we should use the method of separation of variables
and look for a solution for of the form
= f(z) cos(kx t), with f (0) = a, f (h) = 0.
78

Substituting into Laplaces equation

(f k2f) cos(kx t) = 0 =

d2f
k2f = 0.
dz2

The general solution of this equation can be written in a number of ways

f(z)=Aekz + Bekz,
=C cosh(kz) + D sinh(kz),
=E cosh(kz + ) + F sinh(kz + ),
where A, B, C, D, E, F, are constants determined by the boundary conditions, which form is
most convenient depends on the form of the boundary condition. Here the condition f (h) = 0
suggests
f (0) = a, = kE sinh(kh) = a.

f(z) = E cosh (k(z + h)) ,

Hence
=

a cosh (k(z + h))

cos(kx t),
k sinh(kh)

satisfies Laplaces equation and the two boundary conditions on z. To satisfy the final boundary
condition on z = 0 requires

a cosh (kh)
( sin(kx t)) = ag sin(kx t),
k sinh(kh)

which is only true if

2 = gk tanh(kh).
Thus we have obtained a travelling wave solution = a sin(kxt) as long as takes a particular
value for a given k,
= (k),
here (k) = (gk tanh(kh))1/2 .
This is known as the dispersion relation for the problem.
Note that the wave crests travel at different speed for different wave numbers,
c=

=
k

g tanh(kh)
k

1/2

Long waves (small k), travel faster than short waves (large k).

7.6

Linear Superposition and Standing Waves

The same method can be used to show that

sin (kx t + ) ,

and

cos (kx t + ) ,

a constant.

are also possible solutions of the linearised problem as long as satisfies the dispersion relation.
Moreover, since
= gk tanh(kh),
if we write 1 =

gk tanh(kh) then

sin(kx 1t),

cos(kx 1t),

sin(kx + 1t),
79

cos(kx + 1t),

are also solutions. So for a given wave number, right and left travelling waves with speed
g tanh(kh)
k

1/2 c

are possible.
Since the problem is linear, if 1 and 2 are solutions, the A1 + B2 is a solution for any constant A,
B. Thus
= A sin (kx 1t) + B sin (kx + 1t) = (A B) sin(kx) cos(1t) + (A + B) cos(kx) sin(1t),
is a solution. And so by suitable choice of A and B,
sin(kx) cos(1t),

cos(kx) sin(1t),

are also solutions.

Considering
= sin(kx) cos(1t),
when t = 0, = sin(kx) and so plotting for different times and /k < x < /k.

0.5

0.5

(x, t) for t = 0, /41, /31, 2/31, 3/41, /1

Thus we see that for x = 0, /k then = 0 for all times. These positions are known as nodes. For x
= /2k the amplitude of oscillation is one, the maximum amplitude for any position. This is
known as a standing wave.
If we consider the oscillations in a tank with solid walls at x = 0 and x = L, then x = 0 at x = 0, L,
and the solution is a standing wave, but only certain values of k (k1, k2, k3, . . .) are possible.

7.7 Phase Velocity and Group Velocity

Definitions
For a given dispersion relation, (k):
80

1. the phase velocity is

cp =

;
k

2. the group velocity is

d
.
dk
An illustration of the relevance of the group velocity is given below. A physical interpretation
of the group velocity is the speed at which energy is propagated (see appendix F for the proof
for deep water waves).
cg =

Example 50
Use dispersion relation for waves on water of depth
h
= (gk tanh(kh))1/2 ,
so the phase velocity is
cp =

g tanh(kh)
k

1/2

as evaluated earlier.
The group velocity is given
by
d
1
cg =
= (gk tanh(kh))1/2 g tanh(kh) + gkhsech2(kh) .
dk
2
For waves on deep water (kh
1),
= gk,

g
,
k

cp =

81

cg =

1
2

g
.
k

Illustration of Group Velocity

Consider a wave with amplitude a, wave number k and frequency , together with a wave of
amplitude a, wave number k + k and frequency + ,
= a [sin (kx t) + sin ((k + k)x ( + )t)] .
Using the identity sin(2A) + sin(2B) = 2 sin(A + B) cos(A B) this can be re-written as
=2a sin (k +2 k)x ( +2 )t
1 1 cos
2a sin (kx t) cos 12 k(x cgt) ,

1
2

kx 2 1 t ,
for k 0

where

d
,
k0
dk
which is known as the group velocity.
This can be considered to be a wave of wavelength 2/k moving at speed /k in an envelope with
wavelength 2/k moving at speed cg.
cg = lim

Numerical Example 51
Consider a wave given by
f(x, t) = sin x ct cos (x 21 ct) ,

= 0.1,

c = 0.2.

The wave envelope has wavelength 2/ = 20 and propagates at speed21 c = 0.1, while the wave
crests travel at twice the speed so the wave crests travel twice as fast as the envelope. Hence the
shorter waves (wavelength 2) appear at the back and propagate through the envelope.
wave crests

envelope
=2

c/2
c

0.5

0.5

1
50

env

=200

50

A similar effect is seen with wave packets the packet or envelope travels at the group velocity while
the wave crests travel at the phase velocity.

82

7.8

Particle paths for water waves

Consider the velocity potential

=

a cosh (k(z + h))

cos(kx t),
k sinh(kh)

(7.8.3)

and so
a cosh (k(z + h))
sin(kx t),
sinh(kh)
a sinh (k(z + h))
v=z =
cos(kx t).
sinh(kh)

u=x =

For short waves (large k) or deep water waves (large h), we have h
(7.8.3), (7.8.4) and (7.8.5) all simplify as

therefore kh

(7.8.4)
(7.8.5)
1 and then

ek(z+h) + ek(z+h)
cosh (k(z + h))
=
ekz.
ekh ekh
sinh (kh)
Therefore (7.8.4) and (7.8.5) imply that
u=a sin(kx t)ekz,
v=a cos(kx t)ekz.
Assuming that any fluid particle departs only a small amount (x (t), z (t)) from its mean position
(x, z) (constant) then
dx
=u = a sin(kx t)ekz,
dt
dz
=v = a cos(kx t)ekz.
dt
Hence

and so

x =a cos(kx
t)ekz, z =a sin(kx
t)ekz,
x 2 + z 2 = a2e2kz,

which are circles with radius aekz.

83

Now at z = 12 ,

4
.
100
So at a depth of half a wavelength, the wave disturbance has decreased to about 4% of the surface
velocity.
ekz = e2z/ = e

84

Can extend analysis of surface waves to various other situations:

1. Waves on surface of cup solve Laplaces Equation in cylindrical polar coordinates with
r = 0 on r = a. Solutions involve Bessel functions.
2. Waves behind a ship known as Kelvin Wedge (see WWW or Waves in Fluids by Lighthill).

85

Cylindrical Polar coordinates

Cylindrical polar coordinates (r, , z) are coordinates such that

x = r cos ,

y = r sin ,

z = z.

z
r

The unit vectors in the r, , z directions are

r=cos x + sin y,
= sin x + cos y,
z = z.

A.1

Vector operators
A=Ar r + A + Az z
=
A=
A=
2 =

1
+
r+
z
r
r
z
1
1 A Az
(rAr ) +
+
r r
r
z
1 Az A

r
z
1
r r

r+

Ar Az

z
r

1
r

(rA ) Ar

1 2 2
+ 2
r2 2
z

NB You should know the form of in cylindrical polars, but all other identities would be given
in exam questions.

86

Divergence theorem

Given a once differentiable vector field A(x) = A1 x + A2 y + A3 z for a domain D and a volume V
bounded by a piecewise regular surface S and contained in D. Then
A dV =
V

A n dS,

(B.0.6)

where dV is a volume element, n is a unit outward normal to S at a current position x.

Proof
Consider A1 , A2 , A3 separately. Start by considering a convex region V . i.e. any line intersects
the surface at most twice, except when the line runs along an edge.
We have a typical diagram
z

n
z=f 2(x,y)
y
S2
C

x
S1
z=f1 (x,y)

where is the projection of S1 and S2 onto the plane z = 0.

A dV =
V

A1 A2 A3
+
+
x
y
z

dxdydz.

Consider

A3
dxdydz =
z

f2 (x,y)

f1 (x,y)

A3
dz
z

dxdy,

f (x,y)

[A3 (x, y, z)]f21 (x,y) dxdy,

(A3 (x, y, f2 (x, y)) A3 (x, y, f1 (x, y))) dxdy.

On S2 : dSn z = dS = dxdy. On S1 : dSn z = dS = dxdy. On cylinder walls n z = 0.

(A3 (x, y, f2 (x, y)) + A3 (x, y, f1 (x, y))) dS(n z).

The surface contributions add to give

contribution from the sides.

A3 n z dS, where S is the closed surface as there is zero

87

Similarly
A1
dxdydz,
x
A2
dxdydz,
y

A1 n x dS =
S

A2 n y dS =
S

Adding these we get

A1 A2 A3
+
+
x
y
z

(A1 x + A2 y + A3 z) n dS,

dxdydz =
S

hence (B.0.6).
For non-convex regions cut them up into convex pieces for which (B.0.6) holds. The volume
integrals simply add. Surface flux integrals provide all the curved outer parts but we need to
consider the faces produced by cuts. For these faces they occur in pairs, with normals in directly
opposite directions. So these paired contributions cancel. So theorem is not restricted to convex
V.

88

Stokes theorem

Given a once differentiable vector field A(x) = A1 x + A2 y + A3 z in some region R, then for a
simple closed curve C spanned by a regular surface S
A n dS =
S

A dx,

(C.0.7)

where the direction of n and the traverse of C are correspondingly related. A simple curve consists
of smooth arcs and does not intersect itself, likewise the spanning surface S does not intersect
itself. Regular implies continuously varying normal.
Proof
For simplicity we consider a spanning surface S which has a singly-covered projection S3 on the
Oxy plane, and a curve C with a simple projection curve lying in z = 0. (If not, either cut into
pieces for which this is true, or project onto planes x = 0 or y = 0.)
z
y
S
C

S3
O

Suppose S has equation z = f (x, y). Consider

(A1 x) n dS,

I1 =
S

=
S

Note: n

A1
A1
z+
y n dS.
y
z

(fx , fy , 1), hence n y = fy (n z).

I1 =
S

A1
A1
+ fy
y
z

n z dS.

On S A1 = A1 (x, y, f (x, y)) = P (x, y), so

P
A1 A1 f
=
+
,
y
y
z y
and
I1 =
S3

P
dS ,
y
P
dxdy.
S3 y
89

By Greens theorem
P dx =

I1 =

A1 dx.

Consider
A2
A2
zn
xn
x
z

(A2 y) n dS =

I2 =
S

x n=fx n z,
A1
A2
+ fx
I2 =
x
z
S

dS,

n z dS.

On the surface
A2 =A2 (x, y, f (x, y)) = Q(x, y),
Q A2 A2 f
=
+
,
x
x
z x
Q
Q
dS =
dxdy.
I2 =
S3 x
S3 x
=

Q dy =

A2 dy.

(A3 z) n dS =

I3 =

=
=
S

A3
A3
yn
xn
x
y

dS,

A3
A3
fx
fy n z dS,
y
x
S

(A3 fy )
(A3 fx ) n z dS.
x
y

Put
= A3 (x, y, f (x, y))fy P = A3 (x, y, f (x, y))fx ,
Q

Q
A3
A3
=
fy +
fx fy + A3 fxy ,
x
x
z
so
I3 =
S

=
S3

(A3 fy ) + (A3 fy )fx

x
z

Q P
dS =

x
y
S3

dy =
P dx + Q

(A3 fx ) (A3 fx )fy n z dS,

y
z

Q P
dxdy,

x
y

A3 fx dx + A3 fy dy.

On C dz = fx dx + fy dy
I3 =

A3 dz.
C

Adding contributions we have (C.0.7) for the figure considered.

A n dS =
S

A dx.
C

90

Proof of Kelvins Circulation Theorem

At time t, consider two points P and Q on curve C, with position vectors xP and xQ . Taking
these point as being close together
dl = xQ xP .
At time t + t, the points have moved to P and Q with position vectors xP and xQ .
We now need to calculate the change in udl in this time.
Letting uP = u(xP ) etc we have
uQ =u(xP + dl, t) = u(xP ) + (dl)u + . . . ,
(using Taylor expansion)
=uP + (dl)u,
xP =xP + uP t,
xQ =xQ + uQ t = (xP + dl) + (uP + (dl)u)t,
dl =xQ xP = dl + (dl)ut,
DuP
t,
(by definition of material derivative).
uP =uP +
Dt
Hence we can find the change in udl in time t.
u dl =udl + t

Du
dl + u(dl)u ,
Dt

u dl udl Du
=
dl + (dl) 12 uu ,
t
Dt
Du
D
(udl)=
+ 21 uu dl,
Dt
Dt
p
= gz + 12 uu dl,

Hence

D
=
Dt

using Eulers Equation

Du
=
Dt

p
gz + 12 uu dl = 0,

using Stokes Theorem (since (f ) = 0 for any f ). Hence is constant.

91

p
+ gz .

Flow around a cylinder

The complex potential for a flow of speed U , angle , around a cylinder of radius a with circulation
is given by
i
U a2 ei
i

log z.
w = U ze
+
z
2
To plot the streamlines we need to plot the lines Im(z) = =constant. A MATLAB script for
doing this is given below.
clear
clc
[x,y]=meshgrid(-2:0.01:2);
alpha=0;
U=1;
a=1;
gamma=-3*pi;
z=x+i*y;
psi=imag(U*(z*exp(-i*alpha)+a^2*exp(i*alpha)./z)-i*gamma/(2.0*pi)*log(z));
v=-2.0:0.05:2;
hold on
contour(x,y,psi,v,b)
contour(x,y,psi,[0,0],k,LineWidth,2)
axis tight
xlabel(x)
ylabel(y)

92

Propagation of energy for deep water waves

Here we prove that the energy propagation in water waves occurs at the group velocity.

Rate of transmission of energy to right=rate at which fluid on left does work on fluid on right,

per unit width (i.e. force=(pz)x velocity).

pu dz

But in linear form the pressure can be written as

p = p0 t gz

and u = x ,

Therefore

(p0 t gz) x dz.

R = Rate of transmission of energy =

We also require a time average over one period of the flow where we define the mean value of R as
=
R
2
Note that

2/

R dt.
0

2/

cos(kx t) dt = 0,
0

2/
0

cos2 (kx t) dt = 12 .

So

R=
2

2/

t x

0
2/

=
0
2/

=
0

and =

gA kz
e sin(kx t),

g A2 k 2kz
e cos2 (kx t) dzdt,
2

2 2

g A
cos2 (kx t) dt,
4

g 2 A2
=
= Mean rate of transmission of energy.
4
Now the total energy per unit area of surface T is the sum of the potential and kinetic energies
T=
S

g dS +
V

1
|u|2
2

dV,

1 1
1 0
2
1
g
dx
+
(u2 + w2 ) dxdz,
0 2
0 2
1
1
g 2 A2 k 2
= 21 gA2 + 12
,

2
2
2k

using u2 + w2 = g 2 A2 k 2 / 2 e2kz and = A cos(kx t). Therefore

T = 12 gA2

using 2 = gk.

93

Now,
Speed of energy propagation=
=
=

R
Mean rate of energy transmission
= ,
mean energy/unit area
T
1 A2 g 2

4
1
gA2
2

g
2

= 21

g
= cg .
k

In fact for all waves energy travels with the group velocity.

94