A Fundamental Course in Fluid Mechanics and

Heat Transfer

Srikanth Vedantam

August 3, 2014
2
Contents

I Fluid Mechanics 5
1 Introduction 7

2 Fundamental concepts 11
2.1 Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Motion and deformation of a fluid particle . . . . . . . . . . . . . 14
2.2.1 Linear translation rate . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Rotation rate . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3 Linear strain rate . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.4 Shear strain rate . . . . . . . . . . . . . . . . . . . . . . . 20
2.3 Flow visualization . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 Pathlines . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.2 Streamlines . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.3 Streaklines . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.4 Timelines . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4 System and Control Volume . . . . . . . . . . . . . . . . . . . . . 22
2.4.1 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4.2 Control volume . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5 Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3 Balance laws 27
3.1 Balance laws in integral form . . . . . . . . . . . . . . . . . . . . 28
3.1.1 Balance laws for a system . . . . . . . . . . . . . . . . . . 28
3.1.2 Reynolds’ transport theorem . . . . . . . . . . . . . . . . 30
3.1.3 Balance laws for a control volume . . . . . . . . . . . . . 30
3.2 Balance laws in Local Form . . . . . . . . . . . . . . . . . . . . . 35
3.2.1 Mass balance . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.2 Linear momentum balance . . . . . . . . . . . . . . . . . . 37

4 Navier-Stokes equations 39
4.1 Constitutive equations . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Navier-Stokes equations . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3
4 CONTENTS

5 Fluid Statics 43
5.1 Force balance for a static liquid . . . . . . . . . . . . . . . . . . . 43
5.2 Variation of pressure in a static liquid . . . . . . . . . . . . . . . 44
5.2.1 Manometers . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.3 Surface tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.4 Hydrostatic forces on submerged surfaces . . . . . . . . . . . . . 47
5.4.1 Planar surfaces . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4.2 Curved surfaces . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4.3 Example problems . . . . . . . . . . . . . . . . . . . . . . 47
5.5 Buoyancy and stability . . . . . . . . . . . . . . . . . . . . . . . . 47
5.6 Fluids under rigid body motion . . . . . . . . . . . . . . . . . . . 47

6 Dimensional Analysis 49
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6.2 Buckingham Pi theorem . . . . . . . . . . . . . . . . . . . . . . . 50
6.3 Non-dimensionalization . . . . . . . . . . . . . . . . . . . . . . . 51
6.4 Similarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.5 Non-dimensionalization of differential equations . . . . . . . . . . 52

7 Incompressible inviscid flow 55

8 Incompressible viscous flow 57

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
8.2 Internal flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
8.3 External flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

9 Fluid machinery 59

10 Heat Transfer 61
10.1 Conduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
10.2 Convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
10.3 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
10.4 Steady state conduction in one dimension . . . . . . . . . . . . . 62
10.5 Fins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
10.6 Unsteady conduction in one dimension . . . . . . . . . . . . . . . 64
10.6.1 Lumped heat capacity system . . . . . . . . . . . . . . . . 65
10.7 Transient heat transfer from a semi-infinite solid . . . . . . . . . 65
10.8 Finite difference method . . . . . . . . . . . . . . . . . . . . . . . 65
 Part I

Fluid Mechanics

5
Chapter 1

Introduction

Suppose you have a glass bottle filled partially with water on a table. You might
recognize that the glass bottle system consists of three states of matter: solid
glass, liquid water and gaseous air. And further, you might classify the water
and air together as “fluids”.
What really makes the fluids “fluid”? At an atomic level, the glass will have
atoms arranged in an amorphous manner – that is, not crystalline (in fact, from
this point of view, glass is termed “supercooled liquid”). The amorphous distri-
bution of atoms extends to the liquid water and the gaseous vapour materials as
well. The atoms in the gas are generally quite far apart whereas in water and in
glass they tend to cluster quite close together. In fact, though we classify water
and the air as fluids, from an atomic viewpoint, glass and water look more alike.
So why do we classify water and air together as fluids?
We define fluids as materials which flow under applied loads. That is, they
take the shape of surrounding material. Don’t solids flow? In the presence of a
force (we need some force to be acting — for example, even water will form a
spherical drop in the absence of gravity), if we wait really long enough even for
small forces, everything – every material – flows. It is the timescale which is
different for different materials. For convenience, we tend to think of materials
which flow in a few seconds as fluids. Even a steel wire will noticeably flow
given a long enough time (this phenomenon is called creep).
What exactly do we mean by flow? If you compressed a liquid equally from
all directions, the liquid may contract by a certain amount. For a constant
compressive force, the amount of compression will be constant. And this is true
of a solid as well. However, if you sheared a liquid with a constant applied
stress it will continue to deform without stopping, but a solid will only deform
to a certain extent. After some amount of shearing, the stress in the solid
increases and this resists the applied load. In a liquid this does not happen.
This continuous deformation under shear stresses is called flow.
Mechanics is the study of forces and deformations of materials. Even among
fluids, the behaviour could be quite different among different fluids. Some fluids
tend to resist greater forces when they are stationary than when they are moving.

7
8 CHAPTER 1. INTRODUCTION

Other fluids do the opposite. Some fluid start to flow only after a certain level of
shear stress is applied. Characterising these sorts of behaviour is an important
aspect of Fluid Mechanics.
In fact, it is useful to take a step back and look at the way we characterise
the mechanical behaviour of any material. Why would we want to do this
characterisation in the first place? We would like to build models of material
behaviour. In the ideal scenario, we would like to develop a model by using
data from experiments in a simple setting (such as a one dimensional test) and
then predict what the material would do in a more complicated setting. This
will allow us to understand the physics better as well as develop more optimal
designs.
How do we develop models of materials which we can then use to describe
the material behaviour? Generally a few specific steps are followed which shall
be described below.
The first assumption we make is that the material is continuous. This seems
to be an error since we all know that all materials are composed of atoms with
significant empty spaces between them and matter is thus NOT continuous!
However, the assumption of continuity of matter makes life easy in terms of
developing models and has proved to be quite accurate for most materials, even
for nanomaterials which consist of only a few tens of thousands of atoms. The
assumption of continuity allows us to write differential equations to describe
materials. Thus the study of the mechanics of materials (including solids, liquids
and gases) under this assumption is called continuum mechanics.
The next step is to write down a description of the position and motion of
the body. We can write this down in a very general manner. However, if we
intend to study the material in very specific situations such as one dimensional
or two dimensional motions or small deformations, we can write down more
specific descriptions. This will simplify the model being developed greatly. This
description of the motion of the body is called kinematics.
The third step is to write down the physical laws governing the material.
These are some of the basic universal laws of nature such as the a. conservation
of mass, b. conservation of linear momentum, c. conservation of angular mo-
mentum, d. conservation of energy, e. conservation of charge which are assumed
to hold for every part of every body. Use of all the balance laws may not be
necessary for a particular problem. For example, unless the electrical charge is
expected to affect the forces or deformations of the body under consideration,
it may not be necessary to list and use it. In addition to the balance laws, we
also write down the second law of thermodynamics which is an imbalance law.
The balance laws and second law of thermodynamics are essential components
for developing models of material behaviour.
The balance laws introduce various field quantities — required to describe
the force, temperature, charge — which are functions of position in the body
and time. However, it will be seen that the number of balance laws introduced
is always smaller than the number of variables to be solved for. This is because
the balance laws are valid for any material in the universe and the information
describing a particular material has still not been introduced. Unless this specific
 9

information is given, the model does not “know” which material we are trying
to describe. This information is introduced in a continuum model through
constitutive equations. These give specific material relations and depend on
whether the material is a solid, liquid or gas. And even among solids, whether
the material is an elastic solid or inelastic. There are many possible types of
constitutive relations and choosing the appropriate one is the biggest challenge
in developing a material model. Appropriate constitutive equations can only be
written by studying the experimental behaviour of the material very carefully.
We provide enough constitutive equations to complete the system such that
there are as many equations as there are unknowns. The next step is to combine
all the above equations and obtain the governing equations.
For a specific geometry we can then provide the boundary conditions appro-
priate to the problem and solve the governing equations to describe the material.
However, the governing equations are usually complicated and we have to resort
to numerical computational techniques to solve the system. The commonly used
numerical methods are the finite difference and finite element methods.
To summarize, the basic steps of continuum mechanics are:

1.1. Describe the position and motion of the body; Kinematics.

1.2. Write down the basic laws of physics arising from the Balance laws and
second law of thermodynamics.
1.3. Write down equations describing specific material behaviour through Con-
stitutive equations.
1.4. Combine all the above equations to obtain the Governing equations.
1.5. In order to describe a specific problem, specify the Boundary condi-
tions.

1.6. Solve the governing equations and the boundary conditions, usually using
a Numerical solution technique.
In this text we will restrict our discussion to describing the kinematics of
fluid systems, writing down the balance laws of linear and angular momentum
and one simple constitutive equation. This simple equation describes what is
called a Newtonian fluid. We will combine this constitutive equation with the
balance laws of linear momentum to obtain the governing equations for fluids
called the Navier-Stokes equations. In the fully three dimensional form, these
equations are non-linear and complicated to solve.
We will then solve these equations in some simple geometries with simplifying
assumptions and try to understand the physics of fluid flow in some very specific
situations. The aim of this process is to develop an understanding of the flow
and forces in certain simple geometries.
10 CHAPTER 1. INTRODUCTION
Chapter 2

Fundamental concepts

In Chapter 1 we noted that the first steps of continuum mechanics required

the description of the positions and velocities of the fluid body which we called
kinematics. In this chapter we will lay out the concepts connected with the
kinematic description of the body. This will be connected to the deformation
of the body and its motion which we will describe in detail. In addition we will
also introduce the concept of stress. It is important to note that all the concepts
in this chapter are quite general are not restricted to fluids alone. They apply
to every form of matter.

2.1 Kinematics
Kinematics is the description of deformation and motion of matter. Motion is
time dependent deformation. There are several possible ways of describing the
deformation and motion of a body. These descriptions are very general and not
restricted to any one form of matter. However, some are more convenient than
the others for describing fluids.
First it is important to consider a “particle” of matter. A particle (not re-
lated to atoms or molecules) is an infinitesimal (vanishingly small) element of
matter. It is continuously connected to the neighbouring particles in accord
with the continuum viewpoint of matter. In order to give the position of a
particle we will first fix a coordinate frame. There are many possible coordinate
systems we can choose and the physical concepts do not depend on this choice.
The form of the equations look different depending on the choice but are com-
pletely interchangeable. For simplicity we will make the choice of a rectangular
Cartesian coordinate system. In the Appendix we will also list out the forms of
the equations in other coordinate systems.
In Fig. 2.1, we show a particle located at a point with coordinates (x, y, z)
in a Cartesian coordinate frame with origin at O. The shape of the particle is
chosen according to the coordinate frame with sides oriented along the axes of
the coordinate frame. The extent of the particle is dx in the x− direction and

11
12 CHAPTER 2. FUNDAMENTAL CONCEPTS

dz

dy O x
z

dx
Figure 2.1: Differential fluid element

dy, dz along the y− and z− directions respectively. For convenience we choose

the center of the particle at (x, y, z). However, we can alternatively choose one
corner of the cube to be at (x, y, z) and this choice is completely equivalent since
the eventual equations we derive will be for the limit of the particle size tending
to 0.
Before we consider a real material, let us first consider a special kind of
material composed of students taking this course this semester. The students
are “particles” of this material. Let us imagine a boundary which encloses all the
students as shown by the dotted line in Fig. 2.2. As the students move about,
the boundary of the body passes through the “outermost” students. The dashed
line and the crosses show the students in their first class and the dashed dotted
line and round particles shows the students as they are getting into their second
classroom. Since Mechanics is the study of forces and motions of a body, the
force on this body may be the “motivation” of the students to get to the next
class. The “motion” describes how fast they get there. Consider a case where
all the students are in one classroom for one class and have to all go to another
classroom for the next class. We are usually interested in asking two kinds of
questions in describing this motion:

2.1. How fast was a particular student A able to move from one classroom to
the next? In other words what was the velocity of student A with time,
vA (t)?
2.1. KINEMATICS 13

Classroom 1 Classroom 2

Corridor
A

Figure 2.2: Flow of a student “fluid” shown at two different classrooms.

2.2. How fast are students able to move through the a particular region along
the path, say, the doorway of the next classroom? What is the velocity of
the “particles” at this doorway: v(doorway, t)?
In this example of the students, since we know the names of the students in
the class, we are able to ask the first question meaningfully. In the case of fluids
we cannot label the particles except in special cases and so we generally ask the
second question. But, can we label particles in any material? Can we ask the
first question of any real material? It turns out that for solids undergoing small
deformations we can label particles of the solid by means of the initial position.
Then the first question is actually very convenient in this setting.
As we mentioned earlier, in the context of the above example, the “force”
is the motivational level of the class and the “acceleration” is the eagerness of
the students to step into the next classroom. The eagerness of a student at a
particular point such as the doorway to the classroom has two components: the
acceleration of a particular student in the doorway to move into the classroom
(from the innate eagerness for the class) and the rush of her classmates behind
who are almost “carrying” the student in the doorway into the classroom by
crowding close behind her. We will see that the acceleration of a fluid element
has two analogous components.
When we are tracking the velocity of a certain particle (as asked in the first
question above), we can write the student particle’s position at any given time
using a vector rA = xA (t)î + yA (t)ĵ. Then her velocity and acceleration are
simply
dxA dyA
vA = î + ĵ = vxA (t)î + vy A (t)ĵ (2.1)
dt dt
and
d2 xA d2 yA
aA = î + ĵ. (2.2)
dt2 dt2
This sort of description of the material is called a Lagrangian description.
On the other hand, describing the velocity of particles at every point in the
domain (such as in the second question above) is called the Eulerian description
14 CHAPTER 2. FUNDAMENTAL CONCEPTS

of motion of the material. We will look at it in detail next in the context of

fluid flow.
In the Eulerian description of fluid motion we consider the fluid velocity field.
That is, we specify the velocity of the fluid particles at every point r = xî+y ĵ+z k̂
and time t in the fluid by

v(x, y, z, t) = vx (x, y, z, t)î + vy (x, y, z, t)ĵ + vz (x, y, z, t)k̂. (2.3)

The velocity field describes the velocities of the particles flowing through the
various points in the domain.

2.2 Motion and deformation of a fluid particle

A rigid particle experiences two kinds of motions: linear motion under external
forces and angular motion under external torques. In order to obtain the particle
trajectory we solve the linear and angular momentum balance given by
X
F = ma, (2.4)
X
T = Iα, (2.5)

where a is the linear and α the P angular acceleration.

P The total external forces
and torques on the particle are F and T respectively.
In the case of particles in a deformable body, in addition to the rigid particle
modes the particles also experience two modes of deformation: stretching and
shearing. Thus a fluid particle can experience two types of motions and two
types of deformations in general as shown in Fig 2.3.
We will derive the rates of linear and angular accelerations as well as the
rates of stretching and shearing next. It is important to note that just knowing
the velocity field, we can obtain all the rates of motion and deformation.

2.2.1 Linear translation rate

In fluids the Lagrangian description can be employed to look at the position of
a drop of tracer dye in the fluid. In three dimensions, if the position of the drop
of tracer dye is given by rp (t) = xp (t)î + yp (t)ĵ + zp (t)k̂ then the velocity and
acceleration of the tracer dye drop are given by

dxp (t) dyp (t) dzp (t)

vp (t) = î + ĵ + k̂ (2.6)
dt dt dt
and
d2 xp (t) d2 yp (t) d2 zp (t)
ap (t) = î + ĵ + k̂. (2.7)
dt2 dt2 dt2
These arise from the elementary definitions of velocity and acceleration of a
particle.
2.2. MOTION AND DEFORMATION OF A FLUID PARTICLE 15

y y

O x O
x
Translation y Rotation
y

O O
x x
Shearing Stretching

Figure 2.3: The various modes of deformation of a fluid particle.

When we want to calculate the acceleration from an Eulerian velocity field,

we cannot just differentiate the velocity with respect to time. We need to look
at the acceleration of the particles passing through the point of interest at the
moment in time. At the time t let a particle p be at the point (x, y, z). Since
we have the label of the particle at this point, we know that its velocity is
(dxp /dt, dyp /dt, dzp /dt). But this should be the same as the velocity given at
the point p by the velocity field. Thus the velocity of the particle occupying the
point (x, y, z) is
v(x, y, z, t) = v(xp (t), yp (t), zp (t), t). (2.8)
Then the acceleration of the particle occupying (x, y, z) at the instant t is ob-
tained by applying the chain rule of differentiation on the velocity of the particle
flowing through the point
dvp ∂vp ∂vp dxp (t) ∂vp dxp (t) ∂vp dxp (t)
ap (x, y, z, t) = = + + + . (2.9)
dt ∂t ∂xp dt ∂yp dt ∂zp dt

But at the instant t, the components of the particle velocity and the flow field
16 CHAPTER 2. FUNDAMENTAL CONCEPTS

velocity match (this is the definition of the flow field velocity as described above).
Thus we can replace the expressions dxp (t)/dt by vx and dyp (t)/dt and dzp (t)/dt
by vy and vz respectively. Thus the acceleration in the Eulerian description is
given by

Dv ∂v ∂v ∂v ∂v
a(x, y, z, t) = = + vx + vy + vz . (2.10)
Dt ∂t ∂x ∂y ∂z

To emphasize the point that the derivative is of the velocities of the underlying
particles in the flow, the special symbol D/Dt is used. It is called the material
derivative in continuum mechanics. The vector representation is also written as

∂v
a= + (∇v)v. (2.11)
∂t

In component form the above equation can be written as

∂vx ∂vx ∂vx ∂vx

ax = + vx + vy + vz , (2.12)
∂t ∂x ∂y ∂z
∂vy ∂vy ∂vy ∂vy
ay = + vx + vy + vz , (2.13)
∂t ∂x ∂y ∂z
∂vz ∂vz ∂vz ∂vz
az = + vx + vy + vz . (2.14)
∂t ∂x ∂y ∂z

The reason for the presence of the last three terms is that the acceleration
at a point is given by the local acceleration of the fluid element as well as the
movement of surrounding fluid elements into the location. The movement of the
surrounding fluid elements into the location is called the convective component
of the acceleration.
2.2. MOTION AND DEFORMATION OF A FLUID PARTICLE 17

Example Problem 2.2.1. Let the velocity field in a tube be given by

U (x + L)
v(x, y, z, t) = î.
L
2.1. Find the acceleration of a particle moving along the x− axis.
2.2. Find the position xp (t) of a particle which was initially at x = 0 at t = 0.
Calculate its velocity vp (t) and acceleration ap (t).

Solution. 2.1. Substituting the given velocity field vx = U x/L, vy = 0, vz = 0

into eqs. (2.12)–(2.14), we obtain ax = U 2 (x + L)/L2 , ay = 0, az = 0.
2.2. Recall that the definition of the velocity field at each point in the domain
is nothing but the velocities of particles at those points. Therefore, when
the particle p is at a point xp its velocity is dxp /dt = vx . Thus

dxp U (xp + L)
=
dt L
Integrating the above equation and using the fact that the particle was
at x = 0 at t = 0, we get xp (t) = L(eU t/L − 1). This is the Lagrangian
description of the fluid flow. (Note that the acceleration of the particle is

d2 xp U 2 (x + L)
2
=
dt L2
which matches with the acceleration field from the Eulerian description.)
From this example we can see how, even if the Eulerian flow field is not
time dependent, the convective part still contributes to the acceleration
of the fluid.

2.2.2 Rotation rate

The rotation of the fluid particle and its corresponding angular velocity of the
particle ω = ωx î + ωy ĵ + ωz k̂ are important in characterizing the flow. The
angular velocity is defined as the average of the angular velocities of two per-
pendicular lines on the fluid particles. From Fig 2.4 showing a fluid particle at
times t and t + dt, we can see that the average of the angular velocities of the
sides of the fluid element oriented along the x− and y− directions is
      
1 dα dβ 1 1 ∂vx 1 ∂vy
ωz = + = − δydt + δxdt
2 dt dt 2dt δy ∂y δx ∂x
 
1 ∂vy ∂vx
= − . (2.15)
2 ∂x ∂y
The negative sign in the term vx − (∂vx /∂x)δx arises since the line A0 B 0 is to
the left of AB in order to account for the positive (counter-clockwise) angular
18 CHAPTER 2. FUNDAMENTAL CONCEPTS
 
∂vx
vx − ∂y δy dt

B′
dβ

C′
A′ dα
∂vx
vx + ∂y δy B D

 
∂vy
vy + ∂x δx dt
δy vy dt

vx δx
A C
vy

vx dt
At t At t + dt

Figure 2.4: Angular rotation of two perpendicular lines in a fluid particle.

velocity.
You can see for yourself that the other two components of the angular ve-
locity are given by
 
1 ∂vz ∂vy
ωx = − , (2.16)
2 ∂y ∂z
 
1 ∂vx ∂vz
ωy = − . (2.17)
2 ∂z ∂x

The vorticity Ω is defined to be twice the angular velocity. Thus the vorticity
of a fluid particle is the curl of the velocity vector:

Ω = 2ω = ∇ × v. (2.18)

2.2.3 Linear strain rate

The linear deformation (strain) rate of a fluid element is described by the ve-
locity of stretching or contraction of perpendicular lines on the fluid element.
As can be seen in Fig. 2.6, A0 B 0 = AB + BB 0 − AA0 . Thus the rate of change
of length per unit length of a fluid element along the x− axis is

1 d 1 A0 B 0 − AB
(δx) =
δx dt dt AB 
1 1 ∂vx ∂vx
= dx + δxdt − δx = . (2.19)
dt δx ∂x ∂x
2.2. MOTION AND DEFORMATION OF A FLUID PARTICLE 19

Example Problem 2.2.2. Let the velocity field in a two dimensional domain
be given by v(x, y) = √ 2x 2 î + √ 2y 2 ĵ. Find the angular velocity of fluid
x +y x +y
particles in this flow.
Solution. UsingnEq. (2.15), we seethat the z− component o of the angular ve-
1 1 y 1 x
locity, ωz = 2 − 2 (x2 +y2 )3/2 2x − − 2 (x2 +y2 )3/2 2y = 0. The x− and y−
components can easily be verified to be 0 as well. Thus the flow has zero angu-
lar velocity. How is that possible? It can be seen in Fig. 2.5, that the flow field
is radial: that is every particle is moving outward from the centre. In such a
flow the angular velocity is zero.

Figure 2.5: The velocity vectors of the flow field v(x, y).

Similar arguments in the y− and z− directions give the linear strain rates in
∂v
those directions as ∂yy and ∂v
∂z .
z

The volumetric strain rate is the rate of change of volume of the fluid element
per unit volume of the particle is given by
∂vx ∂vy ∂vz
+ + . (2.20)
∂x ∂y ∂z
To see this, note that the volume of the fluid particle is δV = δxδyδz and
the volumetric strain rate is
1 d 1 d
(δV) = (δxδyδz)
δV dt δxδyδz dt
1 d 1 d 1 d
= (δx) + (δy) + (δz), (2.21)
δx dt δy dt δz dt
20 CHAPTER 2. FUNDAMENTAL CONCEPTS
 
∂vx
vx + ∂x δx dt
vx dt
A′ B′
A B
∂vx
vx vx + ∂x δx

δx
At t At t + dt

Figure 2.6: Linear strain rate.

which, from the linear strain rates derived above is the expression given in (2.20).

2.2.4 Shear strain rate

The change in shape of a fluid particle is given by the shear strain rate. This
is defined as the rate of decrease of the angle between two initially mutually
perpendicular lines. Going back to Fig 2.4 we can see that the change in angle
in a time dt is equal to
d π  dα dβ ∂vy ∂vx
−γ = − = + (2.22)
dt 2 dt dt ∂x ∂y
If α = β, there is no shear and only a rigid rotation occurs. Conversely, if
α = −β, then the element does not rotate and is purely under shear.

Example Problem 2.2.3. Let the velocity field in a two dimensional domain
be given by v(x, y) = √ 2x 2 î + √ 2y 2 ĵ as in Ex. (2.2.2). Find the shear
x +y x +y
strain rate for this flow.

n Using Eq. (2.22),we see that the o

Solution. component of the shear strain rate,
1 y 1 x
γxy = − 2 (x2 +y2 )3/2 2x + − 2 (x2 +y2 )3/2 2y = (x2−2xy
+y 2 )3/2
. The x− and y−
components can easily be verified to be 0. The flow has zero angular velocity
but a finite shear rate.

2.3 Flow visualization

In the previous section, we studied the modes of motion and deformation of
a fluid particle in a flow. While the quantitative information of the kinematic
description is necessary for developing a model of the material, other kinds of
information is useful to develop an understanding of the flow behaviour. This
sort of overall information about the flow field can come from observing patterns
in the fluid flow. These patterns can be observed by several methods. We will
describe these in the following subsections.
2.3. FLOW VISUALIZATION 21

2.3.1 Pathlines
Pathlines are the lines traced out by individual particles through the flow. These
are Lagrangian descriptors of the flow of discrete particles in the fluid. The
pathline can be obtained from the Eulerian velocity field v(x, y, z, t) by
Z t
xp = xinit + vdt (2.23)
ts

where xinit is the initial position of the at the starting time ts .

Pathlines are the easiest and most common visualization technique. For
example, soot particles from a candle flame give us an idea of the complex flow
fields above the flame.
As mentioned before, the pathline of a single particle is a Lagrangian de-
scription of the flow. Conversely, the pathlines of many particles introduced into
a flow allow us to experimentally obtain the Eulerian flow field in a technique
called PIV (particle image velocimetry). In this technique, many particles are
introduced into the flow field and images are taken at short time intervals. The
positions of the same particles in the two images give us an indication of the
velocity with which the particles moved. Since many particles are introduced
at the same time, the velocities in the entire domain can be calculated.

2.3.2 Streamlines
Streamlines are defined as a curve whose tangent at any point is the instan-
taneous Eulerian velocity vector. They indicate the instantaneous direction of
fluid motion at any point. Since the velocity vector is tangent to the curve, the
equation of the streamline is given by
dx dy dz
= = (2.24)
vx vy vz
As we can see, the above equation represents a family of curves and the constant
of integration gives the location at which the streamline starts.

2.3.3 Streaklines
If a dye is introduced into a fluid at a particular point, the pattern that is
observed is the continuous tracing of the particles that are passing through that
point. This is termed as a streakline. In other words, a streakline is the locus
of fluid particles that have passed sequentially through a prescribed point in
time. This is the most common flow visualization technique in which a dye or
a differently coloured liquid is introduced into the flow and the lines formed are
visualized as streaklines.

2.3.4 Timelines
Let a set of adjacent particles are marked be coloured at an instant of time. The
positions of these particles at later times gives the timeline of the particles in
22 CHAPTER 2. FUNDAMENTAL CONCEPTS

the flow. If the particles marked are close enough, a continuous line will appear
to deform in the fluid as it flows. This helps us visualize the flow, particularly
to determine the regions in which the flow is changing rapidly.

2.4 System and Control Volume

In the last section we described the velocity of a fluid at a point. In many cases,
it is very useful to look at an entire volume of fluid. In this case we have two
choices, we can follow a fixed mass of fluid or we can consider a fixed region of
space through which fluid flows. These will be described in this section.
It is worth emphasizing the difference between this section and the last once
again. In the last section we looked at infinitesimal fluid particles and described
the position, velocity and acceleration of such particles. Depending on whether
we followed a particular particle or looked at the overall velocity field in the
fluid domain, the description was called Lagrangian or Eulerian respectively.
Here we are looking at finite volumes of fluid domains. Depending on whether
the boundary allows exchange of fluid mass or not, we call the domain a system
or control volume respectively. In both cases, the domain itself may have an
overall motion or deformation into a different shape.

2.4.1 System
A system is defined a fixed, identifiable mass of material. That is, we cannot
exchange a part of the mass of the fluid with another of equal mass. Such a
situation arises naturally in some situations such as helium in a party balloon.
The rubber balloon serves as the system boundary in this case. As the balloon
floats around, the system moves. However, the mass of helium is fixed and
identifiable. The balloon may be squeezed into different shapes and the system
boundaries may deform.

2.4.2 Control volume

When describing fluids, there are few situations in which a well defined system is
identifiable. In most cases of interest, the mass of material is constantly flowing
in and out of regions of space. A region of space into which material is flowing
in and out is called a control volume. The boundaries of a control volume —
called control surfaces — may be mobile as well. These control surfaces may
be real surfaces such as the surface of a solid over which liquid is flowing or
imaginary.

2.5 Stress
In mechanics we are interested in forces on a body and how they affect the
motion of the body. If we want to consider how forces are transmitted inside
the body, it is easier to think about a particle such as that in Fig. 2.7 and its
2.5. STRESS 23

fy

O fx
x
fz
z

Figure 2.7: Two differential elements showing the resolved forces on the inter-
face.

adjacent particles. As shown in the figure let us consider the adjacent particle
along the x−axis first. In general, the particle on the right applies a force on the
particle on the left on the interface between the two particles. The components
of the vector force can be resolved in the x−, y− and z− directions as seen in the
figure. The x− component of the forces affects the adjacent particles differently
than the y− and z− component of the forces. The fx component is trying to
pull apart (or push together depending on the sign) the surfaces whereas fy and
fz are trying to slide the surfaces apart (in different directions). These forces
are acting on the surface area Ax = dydz whose normal is in the x− direction.
Similarly, on the surface of the particle whose normal is in the y− direction with
surface area Ay = dxdz, fy is the force component which pulling the surface
apart whereas fx and fz are the sliding components. Finally, fz is the pulling
apart force on the area Az = dxdy with the normal in the z− direction whereas
the sliding components are fx and fy .

When we are looking at the force on an interface and its effect on the particle,
it is not the actual magnitude of the force acting on the surface, but rather the
force per unit area which is important. After all we can have a relatively small
force acting on a very small area which may cause a greater effect than a larger
force on a much larger area. Thus we consider the force per unit area which we
term stress as the important quantity in the mechanics of deformable bodies. As
we saw in the previous paragraph we can identify several important components
of forces for each surface: those that act to pull apart surfaces and those that
act to slide apart surfaces. Thus the stresses acting on the surface with normal
in the x− direction has stresses σxx = fx /Ax and τxy = fy /Ax , τxz = fz /Ax of
which σxx is the normal stress and τxy , τxz are the shear stresses. Similarly, on
the surface with normal in the y− direction, σyy = fy /Ay is the normal stress
and τyx = fx /Ay ,τyz = fz /Ay are the shear stresses. To complete the full set
of stresses, the surface with normal in the z− direction has σzz = fz /Az as
the normal stress and τzx = fx /Az ,τzy = fy /Az the shear stresses. These nine
24 CHAPTER 2. FUNDAMENTAL CONCEPTS

stresses are written in matrix form as

 
σxx τxy τxz
σ =  τyx σyy τyz  (2.25)
τzx τzy σzz

Depending on the coordinate frame chosen, the values of the stresses may be
different but the matrices of stresses in the various frames are related through
a simple transformation rule. Any such matrix object which transforms under
these geometric rules is called a tensor. The nine components of the stress tensor
are shown on the differential element in Fig. 2.8.
y
σyy

σyx
σyz
σxy
σzy
σxx
O x
σzz σzx σxz
z

Figure 2.8: Differential element showing the nine stress components.

Problems
2.1. Describe an example of a Langrangian description of a material and an
example of an Eulerian description of a material.

2.2. For a velocity field given by v = xî − y ĵ at what time does a particle which
started at x = 1 at t = 0 reach x = 1.5? At what time does a particle
which started at y = 2 at t = 0 reach y = 1.5?

2.3. For a velocity field given by v = y î − xĵ what is the location of a particle
which started at (1, 1) at t = 0 at time t = 1.5?

2.4. Consider four particles initially located at the corners of a unit square
(1, 1), (1, 2), (2, 1), (2, 2) in a flow field given by

(a) v = xî + y ĵ
(b) v = xî − y ĵ
(c) v = y î + xĵ
(d) v = y î − xĵ
2.5. STRESS 25

What are the locations of these particles at t = 1? What is the shape

obtained by joining the particles with straight lines? What is the area
of this shape? What are the angles between the sides of this shape? Do
these answers match with the calculation of the volumetric strain rate and
shearing rate from the given velocity field?

2.5. For the following two dimensional flows, find the acceleration, angular
velocity components, volumetric strain rate and shearing rates.1

(a) v = (3x + y)î + (2x − 3y)ĵ

(b) v = t2 î + t3 ĵ
(c) v = t(x + y)î + t(x − y)ĵ

2.6. For the following three dimensional flows, find the acceleration, angular
velocity components, volumetric strain rate and shearing rates.

(a) v = (at + bx)î + (ct + dy)ĵ + (et + f z)k̂

(b) v = axt2 î + byt3 ĵ + czt4 k̂
(c) v = t(x + y − 3z)î + t(x − y + z)ĵ + (x2 + y 2 + z 2 )k̂

2.7. A flow field in the xy−

 plane has a velocity field given by v(x, y) =
α √ 2y 2 î − √ 2x 2 ĵ . What is the angular velocity of particles in this
x +y x +y
flow? What is the shear rate?

2.8. If the y− component of the velocity in a flow field is given by vy = (x − y)

give all the possible steady (i.e. time-independent) x− components such
that the flow is irrotational (i.e. angular velocity is zero).

2.9. What is an incompressible flow? Can an apparently easily compressible

fluid (such as a gas) experience incompressible flow?

2.10. Which of the following flow fields represent two-dimensional incompress-

ible flows?

(a) v = (4x + y)î + (x − 4y)ĵ

(b) v = (xt2 + y)î + (xt2 + x)ĵ
(c) v = (x + 2y)tî + (x − 2y)tĵ

2.11. Which of the following flow fields represent three-dimensional incompress-

ible flows?

(a) v = (4x + y)î + (x − 4y)ĵ + xy k̂

(b) v = (x + y + zt)î + (xt + y + z)ĵ + (x + yt + z)k̂
1 In this and following problems, you must assume that there are unit coefficients which

allow each term of the velocity components to have the units of velocity.
26 CHAPTER 2. FUNDAMENTAL CONCEPTS

(c) v = (x + 2y)tî + (x − 2y)tĵ

2.12. The three components of velocity in a flow field are given by vx = Ax +
By + Cz, vy = Dx + Ey + F z and vz = Gx + Hy + Jz. Determine a
relationship between the constant coefficients A, B, . . . , J such that the
flow field would represent an incompressible flow.

2.13. If the x− component of the velocity in a flow field is vx = Ax(B −y), find a
possible steady y− component of the velocity such that the flow is incom-
pressible. How many y− components are possible to allow incompressible
flow? Are there unsteady y− components for incompressible flow?
2.14. Would the case of the “student body” example discussed in Section 2.1,
be a system or control volume?
2.15. Verify that the acceleration of a fluid particle can also be written as
 
∂v 1 2
a= +∇ |v| + ω × v
∂t 2
Chapter 3

Balance laws

The basic balance laws or conservation laws of nature are merely statements
of bookkeeping. They do the job of an accountant for a company who tries to
track income and expenditure. Money is not really conserved due to the fickle
notion of value, but the principle that is followed is that whatever money is
earned minus the money spent must be a part of the company’s assets. The
complexity comes from the fact that money comes in different forms such as
cash, credit and assets. Similarly, the conservation laws are, conceptually very
simple but careful bookkeeping is required.
As we mentioned in the introduction, the basic quantities which are nei-
ther created nor destroyed in the universe are mass, linear momentum, angular
momentum, energy and charge. In this text we will write down the balance
laws for the first four quantities. Actually, you may be aware that the general
theory of relativity allows for interchange between mass and energy. So really
we must write a balance law for mass-energy. However, in most applications
that we deal with, these relativistic effects do not play a significant role and
the mass-energy interchange does not occur. Hence we can write separate equa-
tions for mass and energy. Charge also does not affect the flow for the types of
fluids we will be studying in this text. It may be necessary to write the conser-
vation of charge equations in situations where it affects the fluid flow such as
magnetohydrodynamics.
In some instances we may recognize that other higher order quantities may
be conserved for the time duration of interest. In such a case it may be more
useful to write down the balance laws for these higher order quantities. For
example, the ushers for a museum recognize that if a certain number of people
went into the museum since it opened, the same number should exit by closing
time. The museum guards do not weigh each person and write down the law of
conservation of mass for the museum.
This bookkeeping can be performed for an entire system or at a particle
level. When done at the system level or control volume level, the balance laws
arise in the form of integrals over the volume and surface areas of the system
or control volume. In this chapter we will derive the equations for the basic

27
28 CHAPTER 3. BALANCE LAWS

conservation laws of nature in integral form.

3.1 Balance laws in integral form

When local variations in fluid velocities and accelerations are not important, the
global or integral forms of the balance laws become important. However, effects
such as the influence of a wall on the neighbouring fluid cannot be accounted
for by this approach. There are many important problems that can be solved
using the integral form as we will see in this section.

3.1.1 Balance laws for a system

Recall that a system is a fixed, identifiable mass of material. It is useful to
understand the balance laws as described in words first and only then look at
the equations describing these statements.

3.1. Balance of mass: The rate of change of mass of a system is zero.

dM
=0 (3.1)
dt system

where M is the mass of the system

Z Z
Msystem = dm = ρdV. (3.2)
M V

where M is the system mass and V is the system volume. We used the
standard definition of density to relate dm = ρdV . Thus,
Z Z
d d
dm = ρdV = 0 (3.3)
dt M dt V

3.2. Balance of momentum: The rate of change of momentum of a system is

equal to the sum of all the external forces acting on the system.

dP
=F (3.4)
dt system

where the linear momentum of a system is defined as

Z Z
Psystem = vdm = ρvdV. (3.5)
M V

Putting these two together, the balance of linear momentum is given by

Z Z
d d
F= vdm = ρvdV (3.6)
dt M dt V
3.1. BALANCE LAWS IN INTEGRAL FORM 29

3.3. Balance of angular momentum: The rate of change of angular momentum

of a system is equal to the sum of all the external torques acting on the
system.
dH
=T (3.7)
dt system
where the angular momentum of a system H is defined as
Z Z
Hsystem = r × vdm = ρr × vdV. (3.8)
M V

Thus the balance of angular momentum is

Z Z
d d
T= r × vdm = ρr × vdV. (3.9)
dt M dt V

3.4. Balance of energy: The rate of change of energy of a system is equal to

the total heat power supplied to the system minus the work power done by
the system.
dE
= Q̇ − Ẇ (3.10)
dt system
where the total energy of the system E is the sum of the internal energy,
and kinetic and potential energies
Z   Z  
|v|2 |v|2
Esystem = u+ + gz dm = ρ u+ + gz dV. (3.11)
M 2 V 2

Thus the balance of energy is given by

Z   Z  
d |v|2 d |v|2
Q̇ − Ẇ = u+ + gz dm = ρ u+ + gz dV.
dt M 2 dt V 2
(3.12)

3.5. Imbalance of entropy: The rate of increase of entropy of a system is greater

than or equal to the rate of heat supplied divided by absolute temperature.

dS Q̇
≥ (3.13)
dt system T

where the total entropy of the system is given by

Z Z
Ssystem = sdm = ρsdV. (3.14)
M V

Thus the second law of thermodynamics can be written as

Z Z
d d Q̇
sdm = ρsdV ≥ (3.15)
dt M dt V T
30 CHAPTER 3. BALANCE LAWS

In most cases in Fluid Mechanics, a system of fixed identifiable mass cannot

be identified easily. Instead, the control volume is a preferred approach. How-
ever, since mass enters and leaves a control volume, the bookkeeping should
include the flux of the quantity carried in and out by the fluid mass. We will
see how this is done next. For converting the equations for a system to a con-
trol volume, it is necessary to apply an important theorem called the Reynolds’
transport theorem.

3.1.2 Reynolds’ transport theorem

3.1.3 Balance laws for a control volume
We first note that in each of the balance laws for the system we had an extensive
property which was the integral of an intensive property over the system. Thus
we had Z Z
Nsystem = ηdm = ρηdV, (3.16)
M V

where, by comparing with Eqs. (3.2), (3.29), (3.8), (3.11) and (3.14) we see that
if,

N = M, then η = 1 (3.17)
N = P, then η = v (3.18)
N = H, then η = r × v (3.19)
 
|v|2
N = E, then η = u + + gz (3.20)
2
N = S, then η = s (3.21)

The strategy to derive the balance laws for a control volume is to compare
the system and control volume at an instant in which both coincide. The rate
of change of any quantity in a system is related to the change in the quantity
in a corresponding control volume by
Z Z
dN ∂
= ηρdV + ηρv · da (3.22)
dt system ∂t CV CS

This equation is called the Reynolds Transport Theorem. It is very important

to note that here the velocity v is measured relative to the control surface. We
will now look at the specific forms for each of the balance laws.

Mass balance
Using Eqs. (3.3) and (3.22) and setting η = 1 from (3.17) we can obtain the
balance of mass for a control volume
Z Z
∂
ρdV + ρv · da = 0 (3.23)
∂t CV CS
3.1. BALANCE LAWS IN INTEGRAL FORM 31

R
R ṁ = CS ρv · da is the mass flow rate through the
It should be noted that
control surface and q̇ = CS v · da is the volume flow rate through the control
surface.
When the flow is steady there are no time varying terms in the problem and
the mass balance becomes Z
ρv · da = 0. (3.24)
CS
When the fluid is incompressible, ρ = constant and the mass balance becomes
Z
∂V
+ v · da = 0 (3.25)
∂t CS
R
where V = CV dV is the volume of the control volume.

Example Problem 3.1.1. Consider the steady flow of water through the mul-
tiport device shown in Fig. 3.1. The areas are A1 = 20 cm2 , A2 = 15 cm2 ,
A3 = 12 cm2 , and A4 = 5 cm2 . The mass flow rate into the device through A1
is 2 kg/s and the volume flow rate out of the device through A3 =0.001 m3 /s.
The velocity of water through A2 is v2 = 4î m/s. If the velocity of water may be
assumed to be in the direction of the outlet ports, calculate the velocity of water
through A4 .

Solution. The control volume is shown using a dotted line in the Fig. 3.1. Since
the flow is steady, we can use Eq. (3.24). Since the fluid is water, we will also
assume that its density is constant. If, as usual, we assume that the velocity
profile across each inlet/outlet port is constant, the integrals reduce to

ρv1 · A1 + ρv2 · A2 + ρv3 · A3 + ρv4 · A4 = 0

The quantity ṁ1 = ρv1 · A1 is the mass flow rate through area A1 which is
given as 2 kg/s.
Since the velocity through A2 is given as v2 = 4î m/s and the area A2 = −15î
cm2 , the mass flow rate is ṁ2 = ρv2 · A2 = −6 kg/s. (The density of water is
1000 kg/m3 )
The volume flow rate out of A3 is q̇3 =0.001 m3 /s (given). Thus the mass flow
rate out of A3 is ρq̇3 = 1 kg/s.
Substituting all the mass flow rates through Areas 1,2 and 3, we obtain the mass
flow rate through 4 to be ṁ4 = 5 kg/s into the device. Since the area vector
A4 = 5(cos 30î + sin 30ĵ) cm2 , the velocity vector is v4 = −1(cos 30î + sin 30ĵ)
m/s.

Linear momentum balance

Using Eqs. (3.6) and (3.22) and setting η = v from (3.18) we can obtain the
balance of linear momentum for a control volume
Z Z
∂
ρvdV + vρv · da = F (3.26)
∂t CV CS
32 CHAPTER 3. BALANCE LAWS

A3

45◦

60◦

A4

A2

A1

Figure 3.1: Problem 3.1.1

where F is the total external force acting on the system. The external force
may be composed of a body force FB and a surface force FS where the body
force can be written as
Z Z
FB = bdm = ρbdV (3.27)
CV

When the force of gravity is the only body force then b = g.

It is very important to note that the velocity v is measured relative to the
control volume and that Eq. (3.26) holds only for inertial or nonaccelerating
control volumes. In the case of rectilinear acceleration of the control volume,
the derivation must be done with more care. We begin again Newton’s second
law for a system
dP
=F (3.28)
dt system
where the linear momentum is
Z Z
Psystem = vdm = ρvdV. (3.29)
M V

This equation is valid only for velocities measured relative to an inertial reference
frame. If we denote an inertial reference frame by XY Z, then Newton’s second
laws is
Z Z
dPXY Z d dvXY Z
F= = vXY Z dm = dm (3.30)
dt system dt M M dt

The velocities of fluid particles measure relative to the inertial reference frame
vXY Z and the accelerating CV reference frame vxyz can be related by
vXY Z = vxyz + vCV (3.31)
3.1. BALANCE LAWS IN INTEGRAL FORM 33

where vCV is the velocity of the control volume relative to the XY Z frame.
Differentiating the above equation with time we get

dvXY Z dvxyz dvCV

= aXY Z = + = axyz + aCV (3.32)
dt dt dt
Substituting Eq. (3.32) into Eq. (3.30) gives
Z Z
dvCV dvxyz
F= dm + dm (3.33)
M dt M dt
or Z
dvCV dPxyz
F− dm = (3.34)
M dt dt system

Now we can use the Reynolds Transport Theorem for the term on the RHS of
the above equation to obtain
Z Z Z
∂
F− ρaCV dV = ρvxyz dV + vxyz ρvxyz · da (3.35)
CV ∂t CV CS

where we have defined aCV = dvdtCV and used the fact that the system and CV
are coinciding at the time of consideration.

Angular momentum balance

Using Eqs. (3.9) and (3.22) and setting η = r × v from (3.19) we obtain the
balance of angular momentum for a control volume
Z Z
∂
T= r × vρdV + r × vρv · da (3.36)
∂t CV CS

where the total external torque on the control volume may be written as
Z
T = r × Fs + r × bρdV + Tshaf t (3.37)
CV

the first two terms on the RHS being the torques due to surface and body forces
respectively. Combining the above equations we get the angular momentum
balance for an inertial control volume
Z Z Z
∂
r × Fs + r × bρdV + Tshaf t = r × vρdV + r × vρv · da (3.38)
CV ∂t CV CS

Again it must be emphasized that this equation is valid only for an inertial
control volume.
We will now present the angular momentum balance for a rotating control
volume. Consider a fixed inertial fram XY Z and a rotating and translating
reference frame xyz as shown in Fig. ??. Let the angular velocity of xyz be ω
and r be the position vector of a fluid particle in the rotating control volume.
34 CHAPTER 3. BALANCE LAWS

The balance of angular momentum is in the rotating control volume xyz is

Z
r × Fs + r × bρdV + Tshaf t
CV
Z
− r × [2ω × vxyz + ω × (ω × r + ω̇ × r]ρdV
CV
Z Z
∂
= r × vxyz ρdV + r × vxyz ρvxyz · da (3.39)
∂t CV CS

Energy balance
The first law of thermodynamics for the system states that
dE
Q̇ − Ẇ = (3.40)
dt system

where the total energy of the system is given by

Z Z
Esystem = edm = eρdV (3.41)
M V

with
|v|2
e=u+ + gz (3.42)
2
As usual we use the Reynold’s Transport Theorem with N = E and η = e to
obtain Z Z
dE ∂
= eρdV + eρv · da (3.43)
dt system ∂t CV CS

Thus the first law of thermodynamics may be stated as

Z Z
∂
Q̇ − Ẇ = eρdV + eρv · da (3.44)
∂t CV CS

For convenience of problem solving we write Ẇ = Ẇshaf t + Ẇnormal + Ẇshear +

Ẇother . Here Ẇshaf t is the work transferred in and out of the control volume
through a shaft. Ẇnormal is the work done by normal stresses acting on the
boundary of the control volume. This can be rewritten as
Z
Ẇnormal = − σnn v · a (3.45)
CS

Viscous stresses can make σnn different from the hydrostatic pressure p as we
will see in the constitutive equations. However for most cases of engineering
interest σnn ≈ p. Thus the normal stresses term becomes
Z Z
p
Ẇnormal = − σnn v · a = − ρv · a (3.46)
CS CS ρ

in which we just multiplied and divided by ρ.

3.2. BALANCE LAWS IN LOCAL FORM 35

Note that if we choose the control surfaces perpendicular to the flow then
the shear work Ẇshear = 0. This is because the shear stresses act tangential to
the control surfaces whereas the fluid flow is normal. Thus there will no power
added by the shear stresses. For a general choice of a control surface, there will
be a shear power input term.
Putting everything together we obtain the energy balance for a control vol-
ume
Z Z  
∂ |v|2 p
Q̇− Ẇshaf t − Ẇshear − Ẇother = eρdV + u+ + gz + ρv·da
∂t CV CS 2 ρ
(3.47)

Entropy imbalance
The second law for a system is
dS 1
≥ Q̇ (3.48)
dt CV T
where the total entropy of the system is
Z Z
Ssystem = sdm = sρdV (3.49)
M V

The Reynolds Transport Theorem to the system formulation of the second

law to obtain Z Z
dS ∂
= sρdV + sρv · da (3.50)
dt CV ∂t CV CS
Since the system and CV coincide at the time instant under consideration
Z !
1 1 1 Q̇
Q̇ = Q̇ = dA (3.51)
T system T CV CS T A

Combining all of the above we get the second law of thermodynamics for a
control volume
Z Z Z !
∂ 1 Q̇
sρdV + sρv · da ≥ dA (3.52)
∂t CV CS CS T A

3.2 Balance laws in Local Form

The integral balance laws are useful when we are looking at the overall behaviour
of the interaction of a fluid with a control volume. However, we cannot get
detailed information about the fluid from point to point. Besides the interaction
with surfaces cannot be treated in a proper fashion. For this we need to look
at the differential form of the balance laws. In this section we will derive the
balance laws for mass and linear momentum.
36 CHAPTER 3. BALANCE LAWS

dz

dy O x
z

dx
Figure 3.2: Differential control volume

3.2.1 Mass balance

Consider the differential control volume shown in Fig. 3.2. Fluid is entering and
leaving all the surfaces. Using Taylor’s series expansion about the center of the
cubical element, the density and velocities of the fluid on the various surfaces
are

∂ρ dx
ρ(x − dx/2) = ρ − (3.53)
∂x 2
∂ρ dx
ρ(x + dx/2) = ρ + (3.54)
∂x 2
∂vx dx
vx (x − dx/2) = vx − (3.55)
∂x 2
∂vx dx
vx (x + dx/2) = vx + (3.56)
∂x 2
(3.57)

We state the law of conservation of mass as the net rate of mass flux out of
the control volume plus the rate of increase of mass of the control volume equals
zero. The net rate of mass flux out of the control volume can be seen to be
 
∂ρvx ∂ρvy ∂ρvz
+ + dxdydz (3.58)
∂x ∂y ∂z
3.2. BALANCE LAWS IN LOCAL FORM 37

The rate of change of mass in the control volume is given by

∂ρ
dxdydz (3.59)
∂t
Combining these two terms into the word state of the law of conservation of
mass we obtain
∂ρ ∂ρvx ∂ρvy ∂ρvz
+ + + =0 (3.60)
∂t ∂x ∂y ∂z
which can be written in vector form as ∇ · ρv = 0. For an incompressible fluid
ρ =constant in space and time and we get
∂vx ∂vy ∂vz
+ + =0 (3.61)
∂x ∂y ∂z
which is ∇ · v = 0 in vector form.

3.2.2 Linear momentum balance

The balance of linear momentum of the differential control volume gives
 
∂vx ∂vx ∂vx ∂vx ∂σxx ∂τxy ∂τxz
ρ + vx + vy + vz = ρgx + + + (3.62)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
 
∂vy ∂vy ∂vy ∂vy ∂τyx ∂σyy ∂τyz
ρ + vx + vy + vz = ρgy + + + (3.63)
∂t ∂x ∂y ∂z ∂x ∂y ∂z
 
∂vz ∂vz ∂vz ∂vz ∂τzx ∂τzy ∂σzz
ρ + vx + vy + vz = ρgz + + + (3.64)
∂t ∂x ∂y ∂z ∂x ∂y ∂z

where gx , gy andgz are body force components in the x, y, and z directions re-
spectively.
Angular momentum balance tell us that the stress matrix is symmetric:
τxy = τyx , τxz = τzx , τyz = τzy
38 CHAPTER 3. BALANCE LAWS
Chapter 4

Navier-Stokes equations

We derived the mass balance and linear momentum balance equations for a
differential control volume in the previous chapter. We have six unknowns of
the stress components and three unknowns of the velocity components and the
pressure. We have just four equations — mass balance and three components of
the linear momentum balance. In order to be able to solve the problems using
this approach we need to apply constitutive equations.

4.1 Constitutive equations

For a Newtonian fluid the stresses may be expressed in terms of the velocity
gradients thus:
 
∂vy ∂vx
τxy = τyx = µ + (4.1)
∂x ∂y
 
∂vz ∂vx
τxz = τzx = µ + (4.2)
∂x ∂z
 
∂vy ∂vz
τyz = τzy = µ + (4.3)
∂z ∂y
2 ∂vx
σxx = −p − µ∇ · v + 2µ (4.4)
3 ∂x
2 ∂vy
σyy = −p − µ∇ · v + 2µ (4.5)
3 ∂y
2 ∂vz
σzz = −p − µ∇ · v + 2µ (4.6)
3 ∂z
(4.7)

where µ is the viscosity and p is the thermodynamic pressure.

39
40 CHAPTER 4. NAVIER-STOKES EQUATIONS

4.2 Navier-Stokes equations

Substituting the constitutive equations in the linear momentum balance stated
in the previous chapter, we obtain
    
∂vx ∂vx ∂vx ∂vx ∂p ∂ ∂vx 2
ρ + vx + vy + vz = ρgx − + µ 2 − ∇·v
∂t ∂x ∂y ∂z ∂x ∂x ∂x 3
     
∂ ∂vy ∂vx ∂ ∂vz ∂vx
+ µ + + µ + (4.8)
∂y ∂x ∂y ∂z ∂x ∂z
    
∂vy ∂vy ∂vy ∂vy ∂p ∂ ∂vy ∂vx
ρ + vx + vy + vz = ρgy − + µ +
∂t ∂x ∂y ∂z ∂y ∂x ∂x ∂y
     
∂ ∂vy 2 ∂ ∂vy ∂vz
+ µ 2 − ∇·v + µ + (4.9)
∂y ∂y 3 ∂z ∂z ∂y
    
∂vz ∂vz ∂vz ∂vz ∂p ∂ ∂vz ∂vx
ρ + vx + vy + vz = ρgz − + µ +
∂t ∂x ∂y ∂z ∂z ∂x ∂x ∂z
     
∂ ∂vy ∂vz ∂ ∂vz 2
+ µ + + µ 2 − ∇ · v (4.10)
∂y ∂z ∂y ∂z ∂z 3

In this course we will focus only on the incompressible Navier Stokes equa-
tions:
   2 
∂vx ∂vx ∂vx ∂vx ∂p ∂ vx ∂ 2 vx ∂ 2 vx
ρ + vx + vy + vz = ρgx − +µ + + (4.11)
∂t ∂x ∂y ∂z ∂x ∂x2 ∂y 2 ∂z 2
   2 
∂vy ∂vy ∂vy ∂vy ∂p ∂ vy ∂ 2 vy ∂ 2 vy
ρ + vx + vy + vz = ρgy − +µ + + (4.12)
∂t ∂x ∂y ∂z ∂y ∂x2 ∂y 2 ∂z 2
   2 
∂vz ∂vz ∂vz ∂vz ∂p ∂ vz ∂ 2 vz ∂ 2 vz
ρ + vx + vy + vz = ρgz − +µ + + (4.13)
∂t ∂x ∂y ∂z ∂z ∂x2 ∂y 2 ∂z 2

These equations in addition to the incompressible mass balance equation

∂vx ∂vy ∂vz
+ + =0 (4.14)
∂x ∂y ∂z
constitute the governing equations for incompressible fluid flow. In conjunction
with appropriate boundary conditions the velocity and pressure field in a fluid
may be solved.

4.3 Problems
4.1. Solve for the fully developed, steady, two dimensional velocity field for
a fluid flow between two plane horizontal surfaces. The bottom plane is
at rest whereas the top surface is moving at a constant speed U in the
x-direction. The separation between the plates is h.

(a) Calculate the shear stress at the top and bottom plates.
4.3. PROBLEMS 41

(b) What is the force required to drag the top plate at the constant
velocity U if the total area of the plate is A?
4.2. Solve for the fully developed, steady, two dimensional velocity field for a
fluid flow down an incline. The incline is at rest and the top surface of
the fluid of layer is exposed to vacuum. If the mass flow rate is ṁ what is
the height of the fluid layer?
4.3. Solve for the fully developed, steady, two dimensional velocity field for a
fluid flow between two plane horizontal surfaces. Both the top and bottom
plates are at rest but there is constant pressure gradient ∂p/∂x driving
the flow. If the outlet is to the atmosphere, what is the power required to
drive the fluid at a mass flow rate of ṁ?
42 CHAPTER 4. NAVIER-STOKES EQUATIONS
Chapter 5

Fluid Statics

There are many instances in which structures that we design and build will
need to take into account fluid pressures. An example would be any under sea
structure on which water exerts tremendous pressure. There are instances where
surface tension driven phenomena need to be accounted for. In this chapter we
will discuss these aspects of fluid mechanics.
A note on the notation: vectors will be given in boldface. That is, a vector
→
a will be written in the text as a.

5.1 Force balance for a static liquid

Consider a fluid which is not deforming but which may be in rigid body motion.
In such a motion, if you put a marker dye in any location of the fluid, the
dye will not move within the fluid. Let us derive the equation governing such
motion.
For a differential fluid element such as in Fig 2.1, the body force is given by

dfB = gdm = gρdV = ρgdxdydz (5.1)

The pressure in the y-direction on the left face of the differential element is
 
∂p −dy ∂p dy
py − = p + =p− (5.2)
∂y 2 ∂y 2

Similarly the pressure acting on the right face of the differential element is
 
∂p dy ∂p dy
py + = p + =p+ (5.3)
∂y 2 ∂y 2

The force due to the pressure on the left face is

 
∂p dy
dfpy − = p − dxdz ey (5.4)
∂y 2

43
44 CHAPTER 5. FLUID STATICS

and the force due to the pressure on the right face is

 
∂p dy
dfpy + = p − dxdz (−ey ) (5.5)
∂y 2

Thus the net force due to pressure in the y-direction is given by

∂p
dfpy = − dxdydz ey (5.6)
∂y

Similarly the other components due to pressure force can be derived and the
total pressure force in vector form is given by
 
∂p ∂p ∂p
dfp = − ex + ey + ez dxdydz (5.7)
∂x ∂y ∂z

Recall from Physics that grad p is written as ∇p and given by

 
∂p ∂p ∂p
∇p = ex + ey + ez (5.8)
∂x ∂y ∂z

Thus the balance of forces on the fluid element per unit volume can be written
as
−∇p + ρg = ρa (5.9)
where a is the acceleration experienced by the fluid element and is a rigid body
acceleration. Since we are not considering relative motion between fluid elements
a is a constant.

5.2 Variation of pressure in a static liquid

If we neglect all other body forces other than gravity and consider a static liquid,
the variation of pressure in a liquid is given by

dp
= −ρg (5.10)
dz
where g is the acceleration due to gravity.
This principle is used in measuring the pressure difference between two fluids
in a manometer.
Integrating eq. (5.10) using a presssure p0 at some reference level z0 , we can
obtain
p − p0 = −ρg(z − z0 ). (5.11)
It is convenient to measure the distance downward from the free surface of
the liquid and thus
p − p0 = ρgh (5.12)
where h is measured positive down from the free surface into the liquid.
5.2. VARIATION OF PRESSURE IN A STATIC LIQUID 45

5.2.1 Manometers

Consider a simple U-tube manometer shown in fig. 5.1. A is connected to

a tank containing liquid with density ρA and B is connected to another tank
containing liquid with density ρB . The manometeric liquid is of density ρM .
The difference in pressure between A and P is given by pP − pA = ρA ga. The
pressure difference between P and Q is given by pP − pQ = ρM gb. Finally the
pressure difference between B and Q is given by pQ − pB = ρB gc. Putting all
these together we get

pA − pB = g(cρB + bρM − aρA ) (5.13)

The problem with these simple manometers is that the sensitivity is not very
good. That is, small difference in pressures translates to small changes in height
and this makes it hard to read. In order to increase the sensitivity, inclined tube
reservoir manometers are usually used. How do these work? Consider the
inclined tube manometer shown in fig. ??.

c
6cm

Q
a
5cm

b
3.5cm

Figure 5.1: U-tube manometer

46 CHAPTER 5. FLUID STATICS

5.3 Surface tension

When a static liquid comes in contact with a solid surface, it makes an angle
to the surface as shown in Fig. ??. This contact angle is determined by the
relative energies of the liquid-vapour interfaces, solid-liquid interface and the
solid-vapour interface. The drop contact angle is such that it minimizes the
total free energy of the system.
Consider a hemispherical drop on a solid surface with contact angle θ. The
footprint of the drop is a circle of radius, say, R. The volume of the drop is
given by
1 (2 − 3 cos θ + cos3 θ)
V = πR3 . (5.14)
3 sin3 θ
The liquid vapour surface area of the drop is given by ALV = 2πR2 (1 −
cos θ)/ sin2 θ and the solid-liquid surface area is given by ASL = πR2 .
The total gibbs free energy of the system is given by

G = ASL γSL + ALV γLV + ASV γSV . (5.15)

To minimize the Gibbs free energy, we require

dG = dASL γSL + dALV γLV + dASV γSV = 0 (5.16)

Differentiating the above expressions for ALV and ASL we get

 
(1 − cos θ) 1 sin θ
dALV = 4πR dR + 2πR2 − dθ (5.17)
sin2 θ sin θ 1 + cos θ

and
dASL = 2πRdR. (5.18)

Note that
dASL = −dASV , (5.19)

that is, a change in solid-liquid surface area can only come at the expense of the
change in solid-vapour surface area. Finally by differentiating the expression for
the volume of the drop (and setting it to 0 since there is no change in volume)
we get,
Rdθ = −2(sin θ + sin 2θ/4)dR. (5.20)

Substituting the above expression into (5.17) we see that

dALV = dASL cos θ. (5.21)

Using
5.4. HYDROSTATIC FORCES ON SUBMERGED SURFACES 47

5.4 Hydrostatic forces on submerged surfaces

5.4.1 Planar surfaces
5.4.2 Curved surfaces
5.4.3 Example problems

5.5 Buoyancy and stability

5.6 Fluids under rigid body motion
48 CHAPTER 5. FLUID STATICS
Chapter 6

Dimensional Analysis

6.1 Introduction
Consider a simple spring mass system shown in Figure ??. From high school
physics we know that the equation governing the system is given by

d2 x
m + kx = 0, x(0) = x0 (6.1)
dt2
were m is the mass, k is the spring constant and x is the elongation of the
sping from its rest length at any time t and x0 is its initial extension. Then the
extension of the spring at any time t is given by
r
k
x(t) = x0 sin t (6.2)
m
If we want to plot this function we have to choose a mass, a spring constant,
initial extension and the units to express these quantities in. For various choices
of the parameters we will get various
p curves as seen in Figure ??(a). However,
if we define x̄ = x/x0 and t̄ = t k/m as non-dimensional position and time, we
see that whatever the choice of m, k, x0 we get the curve given by x̄(t̄) = sin t̄
(Figure ??(b)). From this we see the underlying form of the solution much
better.
In fact, we did not need the solution of this equation to be able to obtain the
non-dimensional forms for x and t. We know that the solution of the equation
(6.1) can only depend on m, k, t, x0 and therefore

x = f (m, k, t, x0 ). (6.3)

Now let us look at the units of the quantities in the above equation.
We know that we need to have the same dimensions for every additive term
of an equation. That is, it makes no sense to say 5 m = 4 s +1 kg or anything
like that. The dimensions of x and x0 are those of length which we will indicate

49
50 CHAPTER 6. DIMENSIONAL ANALYSIS

as [L]. Similarly the dimensions of m is [M ], k has dimensions of [M ]/[t]2 and

t has dimension [t].
Since the left side of (6.3) has dimension [L] so must the overall dimensions
of right side. All other dimensions must somehow cancel out. Since only x0 has
[L] in it, f must be a linear function of x0 . Thus f (m, k, t, x0 ) = x0 g(m, k, t)
where g is itself a dimensionless quantity.
p We can see that the only way g can
be dimensionless is if g(m, k, t) = h(t k/m). If any other combination comes
up we will be have some dimensional quantities left over. Thus we see that
p
x/x0 = h(t k/m). (6.4)
We, of course, won’t know what functional form h is from this analysis but now
we can do experiments to find out.
This analysis gives us a way of reducing our equation to a simpler form whose
nature we can see more clearly, but more importantly it allows us to reduce the
number of experiments we need to perform to get our function.
If we had started out with the original problem and tried to find out f , we
would need to vary m, k, t, x0 independently and get the value of x. If we did
10 experiments for each parameter, we will need to do 104 experiments. And
we would still not be able to see the underlying form easily. Using (6.4) we need
to do only about 10 experiments to start seeing the form.
Lastly, and perhaps the most important reason for non-dimensionalization,
is the fact that we can perform tests with masses and stiffnesses which are ex-
perimentally convenient. For example, if we are asked to measure the frequency
of oscillation of a micron sized spring mass oscillator, we can do a scaled up test
with convenient masses of and spring stiffnesses. If we want to design a new
car with good aerodynamic characteristics, we can build and work with scale
models.

6.2 Buckingham Pi theorem

Suppose for some physical problem we have a dependent variable x1 which is a
function of n−1 independent variables x2 , x3 , . . . , xn , we can write the functional
form as
x1 = f (x2 , x3 , . . . , xn ). (6.5)
Alternatively we can write this as
g(x1 , x2 , . . . , xn ) = 0. (6.6)
The Buckingham Pi theorem states that we can group these dimensional forms
into n − m independent dimensionless ratios, or Π parameters which can be
written in a functional form
G(Π1 , Π2 , . . . , Πn−m ) = 0. (6.7)
The number m is the number of independent dimensions required to specify the
problem. The dimensions mentioned here need not be only the primary dimen-
sions [M ], [L], [t], temperature [T ] or charge [Q]. They can be combinations of
6.3. NON-DIMENSIONALIZATION 51

the primary dimensions if only those combinations occur. The next section will
explain a systematic process of obtaining the Π parameters in detail.

6.3 Non-dimensionalization
There are several methods of obtaining the dimensionless parameters. I prefer
the simple and straightforward method described below.

6.1. List all the dimensional parameters involved.

6.2. Select a fundamental list of dimensions. Make sure that these dimensions
appear uniquely.

6.3. List the dimensions of the parameters in terms of the dimensions involved.

6.4. Select a set of parameters which contain the dimensions involved uniquely.

6.5. Divide the remaining parameters with the chosen parameters such that
non-dimensional groups are obtained.

The above procedure is demonstrated using two examples. In the first the
primary dimensions are independent, whereas in the second they are not.

Example Problem 6.3.1. Consider the drag force on sphere moving through
air. The drag force on the sphere can be written as a function of the ball diameter
D, velocity V , density of air ρ and viscosity µ. What are the dimensionless
groups for this problem?

6.4 Similarity
We mentioned in the introduction that an important reason for nondimensional-
ization is the ability to work with scale models. In order to enforce a correlation
between scale model tests and life sized models, we need to ensure similarity.
The similarity between the scale models and life size prototype models is of
three kinds:

6.1. Geometric similarity

6.2. Kinematic similarity

6.3. Dynamic similarity

Geometric similarity requires that the model and the prototype have similar
shape and dimensions ratios. Kinematic similarity requires that velocities in
corresponding points should be scaled by the same scale factor. Finally, dynamic
similarity requires that the forces at corresponding points should be scaled by
the same ratio. Dynamic similarity ensures kinematic similarity but not the
other way around.
52 CHAPTER 6. DIMENSIONAL ANALYSIS

In actual tests, it may not be possible to achieve complete dynamically

similar cases. However, it may still be useful to perform the scale model tests
since they show qualitative trends about how increasing one parameter affects
other parameters of interest. Sometimes, a correction factor is calculated and
applied to get quantitatively useful results.

6.5 Non-dimensionalization of differential equa-

tions
So far we’ve looked at physical problems and tried to guess the important de-
pendent and independent parameters. In describing most common fluids, the
governing equations are known: the Navier-Stokes equations. It is possible and
necessary to non-dimensionalize the governing equations even before solving
them. It is then possible to consistently simplify the non-dimensional equa-
tions.
We will non-dimensionalize the Navier Stokes equations now. The conserva-
tion of mass is
∂vx ∂vy ∂vz
+ + =0 (6.8)
∂x ∂y ∂z

and the incompressible Navier-Stokes equations are

   2 
∂vx ∂vx ∂vx ∂vx ∂p ∂ vx ∂ 2 vx ∂ 2 vx
ρ + vx + vy + vz = ρgx − +µ + + (6.9)
∂t ∂x ∂y ∂z ∂x ∂x2 ∂y 2 ∂z 2
   2 
∂vy ∂vy ∂vy ∂vy ∂p ∂ vy ∂ 2 vy ∂ 2 vy
ρ + vx + vy + vz = ρgy − +µ + + (6.10)
∂t ∂x ∂y ∂z ∂y ∂x2 ∂y 2 ∂z 2
   2 
∂vz ∂vz ∂vz ∂vz ∂p ∂ vz ∂ 2 vz ∂ 2 vz
ρ + vx + vy + vz = ρgz − +µ + + (6.11)
∂t ∂x ∂y ∂z ∂z ∂x2 ∂y 2 ∂z 2

To non-dimensionalize these equations we will take a reference length L

and a reference velocity V∞ . The reference length and velocity could be some
characteristic of the system such as the diameter and inlet velocity into a pipe.
Then the basic quantities can be non-dimensionalized as

x y z vx vy vz tL p
x̄ = , ȳ = , z̄ = , v̄x = , v̄y = , v̄z = , t̄ = , and, p̄ = 2
L L L V∞ V∞ V∞ V∞ ρV∞
(6.12)
The non-dimensional counterparts to the above equations can clearly be seen
as
∂v̄x ∂v̄y ∂v̄z
+ + =0 (6.13)
∂ x̄ ∂ ȳ ∂ z̄
6.5. NON-DIMENSIONALIZATION OF DIFFERENTIAL EQUATIONS 53

and the incompressible Navier-Stokes equations are

 2 
∂v̄x ∂v̄x ∂v̄x ∂v̄x gx L ∂ p̄ 1 ∂ v̄x ∂ 2 v̄x ∂ 2 v̄x
+ v̄x + v̄y + v̄z = 2 − + + + (6.14)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ V∞ ∂ x̄ Re ∂ x̄2 ∂ ȳ 2 ∂ z̄ 2
 2 
∂v̄y ∂v̄y ∂v̄y ∂v̄y gy L ∂ p̄ 1 ∂ v̄y ∂ 2 v̄y ∂ 2 v̄y
+ v̄x + v̄y + v̄z = 2 − + + + (6.15)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ V∞ ∂y Re ∂ x̄2 ∂ ȳ 2 ∂ z̄ 2
 2 
∂v̄z ∂v̄z ∂v̄z ∂v̄z gz L ∂ p̄ 1 ∂ v̄z ∂ 2 v̄z ∂ 2 v̄z
+ v̄x + v̄y + v̄z = 2 − + + + (6.16)
∂ t̄ ∂ x̄ ∂ ȳ ∂ z̄ V∞ ∂ z̄ Re ∂ x̄2 ∂ ȳ 2 ∂ z̄ 2

where we have set Re = ρVµ∞ L . Re is an important parameter called the

Reynolds number. We can see that for a model system and a prototype to
be dynamically similar, we only require the non-dimensional parameters, the
2
Reynolds’ number and the Froude numbers, F r = gi L/V∞ , for i = x, y, z to be
the same. In situations where the body forces are not important (such situations
are characterised by small values of F r), only the Reynolds number needs to be
matched.
It should be emphasised that the boundary conditions also need to be similar
in order for the model and prototype flows to be similar. In this process, we
may get some further non-dimensional numbers which need to be matched. For
example, if the boundary condition is given by

vbc = V∞ sin ωt (6.17)

then the non-dimensional form is

 
vbc ωL
v̄bc = = sin t̄ (6.18)
V∞ V∞

Thus for the model and prototype to be similar, the Strouhal number St =
ωL/V∞ should also match.
54 CHAPTER 6. DIMENSIONAL ANALYSIS
Chapter 7

Incompressible inviscid flow

55
56 CHAPTER 7. INCOMPRESSIBLE INVISCID FLOW
Chapter 8

Incompressible viscous flow

8.1 Introduction
8.2 Internal flows
8.3 External flows

57
58 CHAPTER 8. INCOMPRESSIBLE VISCOUS FLOW
Chapter 9

Fluid machinery

59
60 CHAPTER 9. FLUID MACHINERY
Chapter 10

Heat Transfer

10.1 Conduction
In a body in which there is a temperature gradient, heat flows from regions of
high temperature to the low temperature regions. This rate of heat transfer
per unit area is empirically determined to be proportional to the temperature
gradient. Thus,
∂T
Q = −kA (10.1)
∂x
where k is the thermal conductivity of the body (W/m.K) which always pos-
itive as a consequence of the second law of thermodynamics. Thus when the
temperature gradient is negative, there is heat transfer in the positive direction.
The problem of conduction heat transfer is to determine the temperature
field in a body given either fixed temperatures or heat fluxes at the boundaries
of the body. Energy balance in this case gives the heat conduction equation.
Consider a one-dimensional infinitesimal element as shown in Fig. ??. The
balance of energy can be stated as:
Energy conducted in left face + heat generated within element
= change in internal energy + energy conducted out right face
These quantities may be represented mathematically as:
∂T
Energy conducted in left face = Qx = −kA (10.2)
∂x
Heat generated within element = q̇Adx (10.3)
∂T
Change in internal energy = ρcA dx (10.4)
∂t    
∂T ∂ ∂T
Energy out right face = Qx+dx = −A k + k dx (10.5)
∂x ∂x ∂x
where q̇ is the heat generated per unit volume (W/m3 ), c is the specific heat of
the material, (J/kg.K) and ρ is the density (kg/m3 ).

61
62 CHAPTER 10. HEAT TRANSFER

Combining the above terms and simplifying we get

 
∂ ∂T ∂T
k + q̇ = ρc (10.6)
∂x ∂x ∂t
If we do a similar calculation for a three-dimensional element we get the
three-dimensional heat conduction equation
     
∂ ∂T ∂ ∂T ∂ ∂T ∂T
k + k + k + q̇ = ρc (10.7)
∂x ∂x ∂y ∂y ∂z ∂z ∂t
If the thermal conductivity is constant we can rewrite this equation as
∂2T ∂2T ∂2T q̇ 1 ∂T
2
+ 2
+ 2
+ = , (10.8)
∂x ∂y ∂z k α ∂t
where α = k/ρc is called the thermal diffusivity (m2 /s). The larger the value
of α the faster the heat will diffuse through the material. This is because high
α can come through high k in which case, the heat will conduct away faster
or through low heat capacity ρc in which case less energy is used up to store
heat and raise temperature and more energy can be conducted away. Thus a
material with higher α feels cooler to touch than a material with a lower α even
if the two materials are at the same temperature.

10.2 Convection
The convection heat transfer from a solid surface is given by
Q = hA(Tw − T∞ ) (10.9)
where h is the convection coefficient (W/m2 .K), Tw is the temperature of the
solid wall and T∞ is the temperature of the bulk liquid far from the surface.

10.3 Radiation
The net heat exchange between two solid grey surfaces at temperatures T1 and
T2 is given by
Q = σ(T14 − T24 ) (10.10)
Thus when T1 > T2 heat is transferred from high temperature surface to the
low temperature surface.

10.4 Steady state conduction in one dimension

Consider a plane wall of constant thermal conductivity k, area A, width L and
surfaces temperatures T1 and T2 . The heat transfer in the wall is
kA
Q=− (T2 − T1 ) (10.11)
L
10.5. FINS 63

If there are n such walls each of width Li and thermal conductivity ki as in

the figure, the heat transfer is

k1 A k2 A kn A
Q=− (Ti2 − T1 ) = − (Ti3 − Ti2 ) = . . . = − (T2 − Tin ) (10.12)
L1 L2 Ln

The intermediate temperatures Ti2 , Ti3 , . . . , Tin are unknown and have to be
solved from the above equations. If the intermediate temperatures are not
required but only the overall heat transfer rate, then we can use

T1 − T2
Q = P Li . (10.13)
ki A

P Li
ki Ais the thermal resistance of the system. If the end walls are not at fixed
temperature but experience convection with convection coefficients h1 and h2
with bulk fluid temperatures T∞1 and T∞2 respectively, then the heat transfer
is given by
T∞1 − T∞2
Q= 1 P Li 1
. (10.14)
h1 A + ki A + h2 A

10.5 Fins
A fin is a device which enhances heat transfer by increasing the surface area
available for convection. Fins are only effective when the conduction heat trans-
fer rate is far greater then the convection heat transfer rate. If the opposite is
true (i.e. convection heat transfer is greater then conduction), then fins decrease
the overall heat transfer rate.
Consider a fin shown in Fig. ??. The energy balance gives

Energy in left face = energy out right face + energy lost by convection

Noting that the energy lost by convection is hP dx(T − T∞ ), where P is the

perimeter of the cross-section and the conduction in left face is −kA(dT /dx)
and out right face is −kA(dT /dx + dxd2 T /dx2 ), the fin equation can be written
as
d2 T hP
− (T − T∞ ) = 0 (10.15)
dx2 kA
Letting θ = T − T∞ and m2 = hP/kA, the equation becomes

d2 θ
− m2 θ = 0 (10.16)
dx2

The solution of the fin equation can be written as

θ = C1 exp(−mx) + C2 exp(mx) (10.17)

64 CHAPTER 10. HEAT TRANSFER

At the base of the fin, let the temperature be To so that θ = θo = To − T∞ .

The boundary conditions at the fin tip can be of several types: (1) The fin may
be considered long in which case the boundary conditions are

θ = θo at x = 0 (10.18)
θ is bounded at x = ∞ (10.19)

The solution to the fin equation is

θ
= exp(−mx) (10.20)
θo

(2) The fin tip is insulated and so dT /dx = 0 at x = L so that the boundary
conditions are

θ = θo at x = 0 (10.21)
dθ
= 0 at x = L (10.22)
dx
The solution is
θ exp(−mx) exp(mx)
= + (10.23)
θo 1 + exp(−2mL) 1 + exp(2mL)
which can be written more conveniently using the hyperbolic functions, cosh x =
(e−x + ex )/2,
θ cosh m(L − x)
= (10.24)
θo cosh mL
(3) Fin loses heat from the tip through convection so that the boundary condi-
tions are

θ = θo at x = 0 (10.25)
dθ
−k = hθL at x = L (10.26)
dx x=L

The solution in this case is

θ cosh m(L − x) + (h/mk) sinh m(L − x)

= (10.27)
θo cosh mL + (h/mk) sinh mL

10.6 Unsteady conduction in one dimension

The partial differential equation governing unsteady conduction in one dimen-
sion is given by
∂2T 1 ∂T
= (10.28)
∂x2 α ∂t
10.7. TRANSIENT HEAT TRANSFER FROM A SEMI-INFINITE SOLID65

10.6.1 Lumped heat capacity system

If a solid object is immersed in a fluid and if the heat conduction rate is much
higher than the convection from the surface we can assume that the entire body
is at the same temperature T . Then the heat transfer from the body is given
by
dT
Q = hA(T − T∞ ) = −ρcV (10.29)
dt
for which the solution is
T − T∞
= e−[hA/ρcV ]t (10.30)
To − T∞
It is common to write this solution as
T − T∞
= e−BiF o (10.31)
To − T∞
where Bi = hV /kA is the Biot number and F o = kA2 t/ρcV 2 is the Fourier
number.
The lumped capacity of the system is valid when the Biot number is small:
h(V /A)
Bi = < 0.1
k

10.7 Transient heat transfer from a semi-infinite

solid
When the surface of a semi-infinite solid at initial temperature Ti is suddenly
changed to To its temperature is given by
T (x, t) − To x
= erf √ (10.32)
Ti − To 2 αt
where erf(y) is the error function defined by
Z y
2 2
erf(y) = √ e−η dη (10.33)
π 0

10.8 Finite difference method

The finite difference method is a useful numerical technique to solve partial
differential equations whose solutions cannot be obtained in closed form. In
order to establish a difference solution, we divide the domain into (usually equal)
discrete pieces. We then look to evaluate the solution of the equation only at
the discrete locations. The differential terms are replaced by difference terms
using Taylor’s series expansions such as
dT Tk+1 − Tk
= (10.34)
dt ∆t
66 CHAPTER 10. HEAT TRANSFER

where the subscript k represents the time t = k∆t. Similarly, the second order
time derivative can be represented by

d2 T Tk+1 − 2Tk + Tk−1

2
= (10.35)
dt ∆t2
which can be obtained by using Eq.( 10.36) at two time steps ∆t apart.
Similarly the spatial derivatives
dT Ti+1 − Ti
= (10.36)
dx ∆x
where the subscript i represents the point x = i∆x. The second order derivative
can be represented by
d2 T Ti+1 − 2Ti + Ti−1
= (10.37)
dx2 ∆x2
Convince yourself that the finite difference form of the equation

∂2T 1 ∂T
= (10.38)
∂x2 α ∂t
can be written as
Ti+1,k − 2Ti,k + Ti−1,k 1 Ti,k+1 − Ti,k
2
= (10.39)
∆x α ∆t
If the values at a particular time step are known then the above method is
explicit since only one Ti,k+1 is the unknown that needs to be calculated. On
the other hand if the finite difference form is written as
Ti+1,k+1 − 2Ti,k+1 + Ti−1,k+1 1 Ti,k+1 − Ti,k
2
= (10.40)
∆x α ∆t
the finite difference form is called implicit since all the quantities the LHS are
unknown.