Fundamentals of Fluids Mechanics - Brujan

E.A.
Brujan Fundamentals of Fluid Mechanics

4
1. PROPERTIES OF FLUIDS

1.1. Density, specific weight, specific volume, and specific gravity
The density of a fluid is its mass per unit volume, while the specific weight is its weight
per unit volume. In SI units, density will be in kilograms per cubic meter, while specific gravity will
have the units of force per unit volume or newton per cubic meter.
Density and specific weight of a fluid are related as follows:

g
g

= = or . (1.1)

It should be noted that density is absolute since it depends on mass which is independent of
location. Specific weight, on the other hand, is not absolute for it depends on the value of
gravitational acceleration g which varies with location, primarily latitude and elevation above mean
sea level. The density of water at 20
o
C is 1000 kg/m
3
, while the density of air at the same
temperature and at a pressure of 1 bar is 1.29 kg/m
3
.
Specific volume v is the volume occupied by a unit mass of a fluid. It is commonly applied to
gases and is usually expressed in cubic meter per kilogram. Specific volume is the reciprocal of
density.
Specific gravity s of a liquid is the ratio of its density to that of pure water at a standard
temperature. In the metric system the density of water at 4
o
C is 1 g/cm
3
and hence the specific
gravity (which is dimensionless) has the same numerical value for a liquid in that system as its
density expressed in g/cm
3
.

1.2. Compressibility of liquids
Compressibility is the property of a fluid to change its volume with pressure. The
compressibility of a liquid is inversely proportional to its volume modulus of elasticity, also known
as the bulk modulus. This modulus is defined as

( )dp dv v dv dp v E
v
= = , (1.2)

where v is the specific volume and p the unit pressure. As v/dv is a dimensionless ratio, the units of
E
v
and p are the same. The bulk modulus is analogous to the modulus of elasticity for solids;
however, for fluids it is defined on a volume basis rather than in terms of the familiar one-
dimensional stress-strain relation for solid bodies.
The volume modulus of mild steel is about 170,000 MN/m
2
. Taking a typical value for the
volume modulus of cold water to be 2,200 MN/m
2
it is seen that water is about 80 times as
compressible as steel. The compressibility of liquids covers a wide range. Mercury, for example, is
approximately 8% as compressible as water, while the compressibility of nitric acid is nearly six
times greater than that of water.
The quantity reciprocal to bulk modulus is called volume compressibility defined as:

p v
v
=
0
, (1.3)

where v
0
is the initial volume, v is the change in volume, and p is the change in pressure.

1.3. Surface tension
The mechanical model of a liquid surface is that of a skin under tension. Any given patch of
the surface thus experiences an outward force tangential to the surface on the perimeter. The force
per unit length of an interfacial perimeter acting perpendicular to the perimeter is defined as the
E.A. Brujan Fundamentals of Fluid Mechanics

5
surface tension, . The interface between two immiscible liquids, such as oil and water, also has a
tension associated with it, which is generally referred to as the interfacial tension.
The interface between a solid and a fluid also has a surface tension associated with it. Figure
shows a liquid droplet at rest on a solid surface surrounded by air. The region of contact between
the gas, liquid, and solid is termed contact line. The liquid-gas surface meets the solid surface with
an angle measured through the liquid, which is known as the contact angle. The system shown in
Figure 1.1(a) has a smaller contact angle that shown in Figure 1.1(b). The smaller the contact angle,
the better the liquid is said to wet the solid surface. For = 0, the liquid is said to be perfectly
wetting.
When the system illustrated in Figure 1.2 is in static mechanical equilibrium, the contact line
between the three interfaces is motionless, meaning that the net force on the line is zero. Forces
acting on the contact line arise from the surface tensions of the converging solid-gas, solid-liquid,
and liquid-gas interfaces, denoted by
SG
,
SL
, and
LG
, respectively (Figure 1.2). The condition of
zero net force along the direction tangent to the solid surface gives the following relationship
between the surface tensions and contact angle :

cos
LG SL SG
+ = . (1.4)

This is known as Youngs equation.

Figure 1.1. Illustration of contact angle and wetting. The liquid in (a) wets better the solid than that in (b).

Figure 1.2. Surface tension forces acting on the contact line

The surface tension of a droplet causes an increase in pressure in the droplet. This can be
understood by considering the forces acting on a curved section of surface as illustrated in Figure
1.3(a). Because of the curvature, the surface tension forces pull the surface toward the concave side
of the surface. For mechanical equilibrium, the pressure must then be greater on the concave side of
the surface. Figure 1.3(b) shows a saddle-shaped section of surface in which surface tension forces
oppose each other, thus reducing or eliminating the required pressure difference across the surface.
The mean curvature of a two-dimensional surface is specified in terms of the two principal radii of
curvature, R
1
and R
2
, which are measured in perpendicular directions. A detailed mechanical
analysis of curved tensile surfaces shows that the pressure change across the surface is directly
proportional to the surface tension and to the mean curvature of the surface:

6

|
|
.
|
\
|
+ =
2 1
1 1
R R
p p
B A
, (1.5)

where the quantity in brackets is twice the mean curvature. The sign of the radius of curvature is
positive if its center lies in phase A and negative if it lies in phase B. This equation is known as the
Young-Laplace equation, and the pressure change across the interface is termed the Laplace
pressure.

Figure 1.3. Mechanics of curved surfaces that have principal radii of curvature of: (a) the same sign, and (b)
the opposite sign

We conclude this section with the introduction of capillarity. Liquids have cohesion and
adhesion, both of which are forms of molecular attraction. Cohesion enables a liquid to resist tensile
stress, while adhesion enables it to adhere to another body. Capillarity is due to both cohesion and
adhesion. When the former is of less effect than the latter, the liquid will wet a solid surface with
which it is in contact and rise at the point of contact; if cohesion predominates, the liquid surface
will be depressed at the point of contact. For example, capillarity makes water rise in a glass tube,
while mercury is depressed below the true value. Capillary rise in a tube is depicted in Figure 1.4.

Figure 1.4. Illustrative example of capillarity

If a glass capillary tube is brought into contact with a liquid surface, and if the liquid wets the glass
with a contact angle of less than 90
o
, then the liquid is drawn up into the tube. The surface tension is
directly proportional to the height of rise, h, of the liquid in the tube relative to the flat liquid
surface in the larger container. By applying the Young-Laplace equation to the meniscus in the
capillary tube, the following relationship is obtained:


7
=
gb
h
cos 2
, (1.6)

where b is the radius of curvature at the center of the meniscus and is the density difference
between liquid and gas. For small capillary tubes, b is well approximated by the radius of the tube
itself, r, assuming that the contact angle of the liquid on the tube is zero. For larger tubes, the value
of h must be corrected for gravitational deformation of the meniscus.
Surface tension decreases slightly with increasing temperature. Room-temperature organic
liquids typically have surface tensions in the range of 20 mN/m to 40 mN/m, while pure water has a
value of 72 mN/m at 25
o
C. The surface tension of water at the atmospheric pressure can be
calculated with

(
(
|
|
.
|
\
|
|
|
.
|
\
|
=
c
c
c
c
T
T T
T
T T
625 , 0 1 2358 , 0
256 , 1
, (1.7)

where T is temperature and T
c
= 647.3 K.
Surface tension effects are generally negligible in most engineering situations; however,
they may be important in problems involving capillary rise, the motion of drops and bubbles, the
breakup of liquid jets, and in hydraulic model studies where the model is small.

1.4. Sound velocity
Consider a fluid at rest in a rigid pipe of cross-sectional area A (Figure 1.5). Suppose a
piston at one end is suddenly moved with a velocity V for a time dt. This will produce an increase in
pressure which will travel through the fluid with velocity c. While the piston moves the distance
Vdt, the wave front will move the distance cdt. During this time the piston will displace a mass of
fluid AVdt, and during this same time the increase in pressure dp will increase the density of the
portion between 1 and 2 by d. Equating the mass displaced by the piston to the gain in mass
between 1 and 2 due to the increased density, AVdt = Acdtd, from which

d
V
c = . (1.8)

From mechanics the impulse of a force equals the resulting increase in momentum. The
impulse of the force produced by the piston is Adtdp. The mass Acdt is initially at rest, but as
the pressure wave travels through it, each element of it will have its velocity increased to V, so that
at the end of the time dt the entire mass up to section 2 will have the velocity V. Hence the increase
in momentum is AcVdt. Thus Adtdp = AcVdt, and

=
V
dp
c , (1.9)
or
d
dp
c =
2
. (1.10)

In section 1.2 the volume modulus of elasticity is defined as E
v
= (v/dv)dp. Since = 1/v,
v = 1 = constant, and thus dv + vd = 0, and v/dv = +/d. Hence E
v
= dp/d. Therefore, we
finally have

8
v
E
c = . (1.11)

This is the velocity of a pressure or sound wave, commonly referred to as the sound speed. For
incompressible liquids c . For normal air at 0
o
C, the sound velocity is 314 m/s while at 20
o
C,
the velocity is 340 m/s. The sound speed in water at 24
o
C is about 1496 m/s.

Figure 1.5. Illustrative example of sound velocity

1.5. Viscosity
An ideal fluid may be defined as one in which there is no friction. Thus the forces acting on
any internal section of the fluid are purely pressure forces, even during motion. In a real fluid,
shearing (tangential) and extensional forces always come into play whenever motion takes place,
thus given rise to fluid friction, because these forces oppose the movement of one particle relative
to another. These friction forces are due to a property of the fluid called viscosity. The friction
forces in fluid flow result from the cohesion and momentum interchange between the molecules in
the fluid.

1.5.1. Shear viscosity
The shear viscosity of a fluid is a measure of its resistance to tangential or shear
deformation. To better understand the concept of shear viscosity we assume the following model
(Figure 1.6): two solid parallel plates are set on the top of each other with a liquid film of thickness
Y between them. The lower plate is stationary, and the upper plate can be set in motion by a force F
resulting in velocity U. The movement of the upper plane first sets the immediately adjacent layer
of liquid molecules into motion; this layer transmits the action to the subsequent layers underneath
it because of the intermolecular forces between the liquid molecules. In a steady state, the velocities
of these layers range from U (the layer closest to the moving plate) to 0 (the layer closes to the
stationary plate). The applied force acts on an area, A, of the liquid surface (surface force), inducing
a so-called shear stress (F/A). The displacement of liquid at the top plate, x, relative to the
thickness of the film is called shear strain (x/L), and the shear strain per unit time is called the
shear rate (U/Y).

Figure 1.6. Illustrative example of shear viscosity


9
If the distance Y is not too great or the velocity U too high, the velocity gradient will be a
straight line. Experiment has shown that for a large class of fluids

Y
AU
F ~ . (1.12)

It may be seen from similar triangles in Figure 1.5 that U/Y can be replaced by the velocity gradient
du/dy. If a constant of proportionality is now introduced, the shearing stress between any two thin
sheets of fluid may be expressed by
dy
du
Y
U
A
F
= = = , (1.13)

which is called Newtons equation of viscosity. In transposed form it serves to define the
proportionality constant
dy du
= , (1.14)

which is called the dynamic coefficient of viscosity. The term du/dy = & is called the shear rate.

Figure 1.7. Rheological behaviour of materials Figure 1.8. Shear-thinning behaviour of two
polymer aqueous solutions

A fluid for which the constant of proportionality (i.e., the viscosity) does not change with
rate of deformation is said to be a Newtonian fluid and can be represented by a straight line in
Figure 1.7. The slope of this line is determined by the viscosity. The ideal fluid, with no viscosity, is
represented by the horizontal axis, while the true elastic solid is represented by the vertical axis. A
plastic which sustains a certain amount of stress before suffering a plastic flow can be shown by a
straight line intersecting the vertical axis at the yield stress. There are certain non-Newtonian fluids
in which varies with the shear rate. Typical representatives of non-Newtonian fluids are liquids
which are formed either partly or wholly of macromolecules (polymers), and two phase materials,
like, for example, high concentration suspensions of solid particles in a liquid carrier solution. For
most of these fluids, the shear viscosity decreases with increasing shear rate, and we call them
shear-thinning fluids. Here the shear viscosity can decreases by many orders of magnitude. This is a
phenomenon which is very important in the plastics industry, since the aim is to process plastics at
high shear rates in order to keep the dissipated energy small. An example is given in Figure 1.8 for
the case of two polymer aqueous solutions. If the shear viscosity increases with shear rate, we speak
of shear-thickening fluids. Note that this notation is not unique, and shear-thinning fluids are often
called pseudoplastic, and shear-thickening fluids are called dilatant.

10
The dimensions of dynamic viscosity are force per unit area divided by velocity gradient or
shear rate. In the metric system the dimensions of dynamic viscosity are

s Pa
m
s N
s
N/m
of dimensions
of dimensions
of Dimensions
2 1
2
=
= = =

dy du

A widely used unit for viscosity in the metric system is the poise (P). The poise = 0.1 Ns/m
2
.
The centipoise (cP) (= 0.01 P = mNs/m
2
) is frequently a more convenient unit. It has a further
advantage that the viscosity of water at 20
o
C is 1 cP. Thus the values of the viscosity in centipoises
is an indication of the viscosity of the fluid relative to that of water at 20
o
C.
The dynamic viscosity of water can be calculated with

( ) ( ) ( )
(
=

= =
5
0
6
0
* * * *
0
1 1 exp
i j
j i
ij
T H T , (1.15)

where H
ij
is given by

i \ j 0 1 2 3 4 5 6
0 5.132 2.152 2.818 1.778 0.417 0 0
1 3.205 7.318 10.707 4.605 0 0.158 0
2 0 12.41 12.632 2.34 0 0 0
3 0 14.767 0 4.924 1.6 0 0.036
4 7.782 0 0 0 0 0 0
5 1.885 0 0 0 0 0 0

and
( )
1
3
0
*
* *
0
071 . 55

=
|
.
|
\
|
=

i
i
i
T
H
T T , (1.16)

with H
i
= (1, 0.978, 0.58, -0.202), T
*
= T / T
c
,
c
/
*
= , T
c
= 647.3 K, and
c
= 317.76 kg/m
3
.
In many problems involving viscosity there frequently appears the value of viscosity divided
by density. This is defined as kinematic viscosity , so called because force is not involved, the only
dimensions being length and time, as in kinematics. Thus

= . (1.17)

In SI units, kinematic viscosity is measured in m
2
/s while in the metric system the common units are
cm
2
/s, also called the stoke (St). The centistoke (cSt) (0.01 St) is often a more convenient unit.
The dynamic viscosity of all fluids is practically independent of pressure for the range that is
ordinarily encountered in engineering work. The kinematic viscosity of gases varies with pressure
because of changes in density.

1.5.2. Extensional viscosity
The extensional viscosity of a fluid is a measure of its resistance to extensional or
elongational deformation. To better understand the concept of extensional viscosity we assume the
following model (Figure 1.9): a spherical bubble is collapsing in a liquid of infinite extent. The
motion is thus spherical symmetric and can be described by only one spatial coordinate, the radial
distance from the bubble center, r. The maximum velocity of the liquid is attained at the bubble

11
wall while the liquid velocity is zero at infinity. By analogy with the results presented in section
1.5.1 we can define the extensional viscosity as

dr du
=
E
(1.18)

where
E
is the extensional viscosity and du/dr = & is the constant strain rate. In most applications,
the extensional viscosity is presented in terms of a Trouton ratio which is defined conveniently to
be the ratio of extensional viscosity to the shear viscosity, Tr =
E
/. The Trouton ratio which takes
the constant value 3 for Newtonian liquids and shear-thinning inelastic liquids, is found to be a
strong function of strain rate & in many non-Newtonian elastic liquids (a material is said to be
elastic if it deforms under stress (e.g., external forces), but then returns to its original shape when
the stress is removed), with very high values (~10
4
) possible in extreme cases (see, for example,
Figure 1.10). This observation has emphasized the potential importance of extensional viscosity in
such areas as fibre spinning, lubrication, drag reduction and enhanced oil recovery. It must be noted
here that generating a purely extensional flow in the case of mobile liquids is virtually impossible.
The most that one can hope to do is to generate flows with a high extensional component and to
interpret the data in a way which captures that extensional component in a convenient and
consistent way through a suitable defined extensional viscosity and strain rate.

Figure 1.9. Illustrative example of an Figure 1.10. Extensional viscosity of two
extensional flow polymer aqueous solutions

1.5.3. Engler viscosity
The Engler viscosity is defined by the formula

w
1
EV = , (1.19)

where
1
is the time during which 200 cm
3
of the fluid under investigation flow through the gauged
orifice of a viscometer at a given temperature, T, and
w
is the time during which 200 cm
3
of
distilled water flow through it at 20
o
C. Kinematic viscosity can be determined from the Engler
viscosity with the help of the following formula:

4
10
EV
063 . 0
EV 073 . 0

|
.
|
\
|
= . (1.20)

For the Engler viscosity exceeding 16 Engler degrees, the formula EV 10 4 . 7
6
= should be used.

12

1.6. Viscoelasticity
When a fluid is suddenly strained and then the strain is maintained constant afterward, the
corresponding stresses induced in the fluid decrease with time. This phenomenon is called stress
relaxation, or relaxation for short. If the fluid is suddenly stressed and then the stress is maintained
constant afterward, the fluid continues to deform, and the phenomenon is called creep. If the fluid is
subjected to a cycling loading, the stress-strain relationship in the loading process is usually
somewhat different from that in the unloading process, and the phenomenon is called hysteresis.
The features of hysteresis, relaxation, and creep are found in many materials. Collectively, they are
called features of viscoelasticity.
Mechanical models are often used to discuss the viscoelastic behaviour of fluids. Figure
1.11 illustrates the simplest model composed of a combination of a linear spring with spring
constant and a dashpot with coefficient of viscosity (Maxwell linear model). A linear spring is
supposed to produce instantaneously a deformation proportional to the load. A dashpot is supposed
to produce a velocity proportional to the load at any instant. Thus, if F is the force acting in a spring
and u is its extension, then F = . If the force F acts on a dashpot, it will produce a velocity of
deflection u& , and u F & = . In the Maxwell model, the same force is transmitted from the spring to
the dashpot. This force produces a displacement F/ in the spring and a velocity F/ in the
dashpot. The velocity of the spring extension is / F
&
if we denote a differentiation with respect to
time by a dot. The total velocity is the sum of these two:

F F
u + =
&
& . (1.21)

Furthermore, if the force is suddenly applied at the instant of time t = 0, the spring will be suddenly
deformed to u(0) = F(0)/, but the initial dashpot deflection would be zero, because there is no time
to deform. Thus the initial condition for the Maxwell model is

) 0 (
) 0 (
F
u = . (1.22)

For the Maxwell fluid, the sudden application of a load induces an immediate deflection by the
elastic spring, which is followed by creep of the dashpot. On the other hand, a sudden
deformation produces an immediate reaction by the spring, which is followed by stress relaxation
Figure 1.11. The mechanical model of
the Maxwell linear fluid

13
according to an exponential law. The factor / = , with the dimension of time, is called the
relaxation time because it characterize the rate of decay of the force.
In terms of stress and strain rate the Maxwell model reads as

e& & = + . (1.23)

This equation is called the equation of linear Maxwell viscoelastic fluid. We can call the
characteristic time the memory span of the fluid. As 0, we obtain the constitutive relation
valid for Newtonian fluids.
The materials mentioned until now have been pure fluids, that is materials where the
shearing forces vanish when the rate of deformation vanishes. However, we often have to deal with
substances which have a dual character. Of these substances, we shall mention here the Bingham
material, which can serve as a model for the material behaviour of paint, or more generally, for high
concentrations of solid particles in Newtonian fluids.

If the solid particles and the fluid are dielectrics, that is do not conduct electrically, then these
dispersions can take on Bingham character under a strong electric field, even if they show only pure
fluid behaviour without electric field. These electrorheological fluids, whose material behaviour
can be changed very quickly and without much effort, can find applications, for example, in the
damping of unwanted oscillations. Through appropriate measures the material can be made to self-
adjust to changing requirements and may be formed into intelligent materials, which are found
increasingly interesting. Even the behaviour of grease used as a means of lubricating ball bearings,
can be described with the Bingham model.
The constitutive equation of the Bingham material in the case of a simple shearing flow is: if
the material flows, we have for the shear stress

& + =
0
for
0
, (1.24)

otherwise the material behaves like an elastic solid, i.e. the shear stress is

G = for
0
< , (1.25)

where
0
is the yield stress and G is the shear modulus.
An important constitutive equation for materials with yield stress is the Casson equation
which it is used to describe the rheology of human blood at a hematocrit content of less than 40%.
The Casson equation reads as
& + =
0
(1.26)
Figure 1.12. Behaviour of Bingham materials

14
where the dynamic viscosity is constant. Note that, for human blood,
0
is very small: of the
order of 0.05 dyn/cm
2
, and is almost independent of the temperature in the range 10 37
o
C.
0
is
markedly influenced by the macromolecular composition of the suspending fluid. A suspension of
red cells in saline plus albumin has a zero yield stress; a suspension of red cells in plasma
containing fibrinogen has a finite yield stress.


15
2. FLUID STATICS

There are no shear stresses in fluids at rest; hence only normal forces are present. The
average pressure intensity is defined as the force exerted on a unit area. If F represents the total
force on some finite area A, while dF represents the force on an infinitesimal area dA, the pressure
is
dA
dF
p = (2.1)

If the pressure is uniform over the total area, then p = F/A. In SI units, pressure is expressed in N/m
2

(Pascal).
In a fluid at rest no tangential stresses can exist and the only forces between adjacent
surfaces are pressure forces normal to the surfaces. Therefore the pressure at any point in a fluid at
rest is the same in every direction.

2.1. Variation of pressure in a static fluid
Consider the differential element of static fluid shown in Figure 2.1. Since the element is
very small, we can assume that the density of the fluid within the element is constant. Assume that
the pressure at the center of the element is p and that the dimensions of the element are

x,

y , and

z. The forces acting on the fluid element in the vertical direction are the body forces, the action of
gravity on the mass within the element, and the surface forces transmitted from the surrounding
fluid and acting at right angles against the top, bottom, and sides of the element. Since the fluid is at
rest, the element is in equilibrium and the summation of forces acting on the element in any
direction must be zero. If forces are summed up in the horizontal direction, that is, x or y, the only
forces acting are the pressure forces on the vertical faces of the element. To satisfy
= 0
x
F and
= 0
y
F , the pressure on the opposite vertical faces must be equal. Thus, p/x = p/y = 0 for the
case of the fluid at rest.

Summing up the forces in the vertical direction and setting equal to zero

z y x y x
z
z
p
p y x
z
z
p
p F
z

2 2
(2.2)
y x
z
z
p
p
+
2

y x
z
z
p
p
2

z y x

z
y
x
x
y
z
Figure 2.1. A differential element of static fluid

16

This results in = z p , which, since p is independent of x and y, can be written as

=
dz
dp
(2.3)

This is the general expression that relates variation of pressure in a static fluid to vertical position.
The minus sign indicates that as z gets larger (increasing elevation), the pressure gets smaller.
To evaluate pressure variation in a fluid at rest one must integrate the previous equation
between appropriately chosen limits. For incompressible fluids ( = constant) we have

=
p
p
z
z
dz dp
1 1
, ( )
1 1
z z p p = (2.4a,b)

This expression is generally applicable to liquids since they are only slightly compressible. For
compressible fluids, however, must be expressed algebraically as a function of z and p if one is to
determine pressure accurately as a function of elevation.
For the case of a liquid at rest, it is convenient to measure distances vertically downwards
from the free liquid surface. If h is the distance below the free liquid surface and if the pressure of
air and vapour on the surface is arbitrarily taken to be zero we can write

p = h (2.5)

As there must always be some pressure on the surface of any liquid, the total pressure at any depth h
is given by this equation plus the pressure on the surface p
0
:

p = p
0
+ h (2.6)

This expression is called the fundamental equation of hydrostatics.
It is now easy to conclude that all points in a connected body of constant density fluid at
rest are under the same pressure if they are at the same depth below the liquid surface (Pascals
law). This indicates that a surface of equal pressure for a liquid at rest is a horizontal plane. For
gases, the term h can be generally neglected for small h-values because of the relatively small
density of the gas.

2.2. Absolute and gage pressure
If pressure is measured relative to absolute zero, it is called absolute pressure; when
measured relative to atmospheric pressure as a base, it is called gage pressure. This is because
practically all pressure gages register zero when open to the atmosphere and hence measure the
Figure 2.2. Illustrating absolute pressure,
and positive and negative gage
pressure

17
difference between the pressure of the fluid to which they are connected and that of the surrounding
air.
If the pressure is below that of the atmosphere, it is called vacuum and its gage value is the
amount by which it is below that of the atmosphere. Gage pressures are positive if they are above
that of the atmosphere and negative if they are below that of the atmosphere (vacuum) (Figure 2.2).

2.3. Barometer
The absolute pressure of the atmosphere is measured by the barometer. If a tube such as in
Figure 2.3 has its lower end immersed in a liquid which is exposed to the atmospheric pressure, and
if air is exhausted from the tube, the liquid will rise in it. If the air was completely exhausted, the
only pressure on the surface of the liquid in the tube would then be that of its own vapour pressure
and the liquid would reached its maximum height.
From concepts developed in section 2.1 the pressure at O within the tube and at a at the
surface of the liquid outside the tube must be the same; that is p
O
= p
a
. We may write

vapour atm
p h p + = (2.7)

and if the vapour pressure on the surface of the liquid in the tube were negligible, then we should
have:

h p
atm
= (2.8)

The liquid employed for barometers is usually mercury, because its density is sufficiently
large to enable a reasonably short tube to be used, and also because its vapour pressure is negligibly
small at ordinary temperatures. If some other liquid was used, the tube necessarily would be so high
as to be inconvenient and its vapour pressure at ordinary temperatures would be appreciable; hence
a nearly perfect vacuum at the top of the column would not be attainable.
The sea-level atmospheric pressure may be expressed as follows:

101.3 kN/m
2
, abs (1,013 mbars, abs) , 760 mm Hg , 10.3 m H
2
O

2.4. Measurement of pressure
There are many ways by which pressure in a fluid may be measured. Some of them are
discussed below.

Figure 2.3. Illustrative example of barometer


18
Figure 2.4. Piezometer column Figure 2.5. A simple manometer

2.4.1. Piezometer column
A piezometer column is a simple device for measuring moderate pressures of liquids. It consists of
a tube (Figure 2.4) in which the liquid can freely rise without overflowing. The pressure is given by
p = h. To reduce capillary error the tube should be at least 12 mm in diameter.
If the pressure of a flowing liquid is to be measured, special precautions should be taken in
making the connection. The hole must be drilled absolutely normal to the interior surface of the
wall, and the surface roughness near the hole must be removed. Also, the hole should be small,
preferable not larger than 3 mm.

2.4.2. Simple manometer
Since the open piezometer tube is cumbersome for use with liquids under high pressure and cannot
be used with gases, the simple manometer or U-tube of Figure 2.5 is a convenient device for
measuring pressure. To determine the pressure in A we may write:

C B
p p = (2.9)
where
z p p
A B
=
1
and y p p
atm C
+ =
2
(2.10a,b)
and, therefore,
y z p p
atm A
+ + =
2 1
(2.11)

Although mercury is generally used as the measuring fluid in the simple manometer, other
liquids can be used. As the density of the measuring fluid approaches that of the fluid whose
pressure is being measured, the reading becomes larger for a given pressure, thus increasing the
accuracy of the instrument, provided the density is accurately known.

2.4.3. Differential manometer
In many cases only the difference between two pressures is desired and for this purpose differential
manometers, such as shown in Figure 2.6, may be used. For the geometry illustrated in the left-hand
side of Figure 2.6, the measuring fluid is of greater density than that of the fluid whose pressure
difference is involved. We may write

D C
p p = (2.12)
where
A A C
z p p =
1
and y z p p
B B D 2 1
+ = (2.13a,b)
and thus
y z p z p
B B A A 2 1 1
+ = where y z z
B A
= (2.14)

19
or
( ) y p p
B A
=
1 2
(2.15)
and finally
y
p p
B A
1
1
2
1

(2.16)
Figure 2.6. A differential manometer

This equation is applicable only if A and B are at the same elevation. If their elevations are
different, an elevation-difference term must be added to this equation. The differential manometer,
when used with a heavy liquid such as mercury, is suitable for measuring large pressure differences.
For a small pressure difference a light fluid, such as oil, or even air, may be used, and in this case
the manometer is arranged as in the right-hand side of Figure 2.6. Naturally, the fluid must be one
that will not mix with the fluid in A or B. By the same method of analysis as the preceding, it may
be shown that for the simple case with identical liquids in A or B, and with both A and B at the same
elevation
y
p p
B A
1
2
1
1
(2.17)

where the ratio
2
/
1
has a value less than 1. As the density of the measuring fluid approaches that
of the fluid being measured, (1 -
2
/
1
) approaches zero and larger values of y are obtained for small
pressure differences, thus increasing the sensitivity of the gage. Once again, this equation must be
modified if A and B are not at the same elevation.

2.5. Fluid masses subjected to acceleration
Under certain conditions there may be no relative motion between the particles of a fluid
mass yet the mass itself may be in motion. If a body of fluid in a tank is transported at a uniform
velocity, the conditions are those of ordinary fluid statics. But if it is subjected to acceleration,
special treatment is required.
Consider the case of a liquid mass in an open tank moving horizontally with a linear
acceleration a
x
, as shown in Figure 2.7(a). A free body diagram (Figure 2.7(b)) of a small particle
(mass m) of liquid on the surface indicates that the forces exerted by the surrounding fluid on the
particle are F
z
= W and F
x
= ma
x
; the latter is required to produce acceleration a
x
of the particle.
Equal and opposite to these forces are F
x
and F
z
of Figure 2.7(a), the forces exerted by the particle
on the surrounding fluid. The resultant of these forces is F. The liquid surface must be at right
angles to F, for it were not, the particle would not maintain its relative position in the liquid. Hence

20
g a
x
= tan . The liquid surface and all planes of equal hydrostatic pressure must be inclined at
angle with the horizontal.

In Figure 2.7(c) is shown a free-body diagram of an elemental cube of the liquid at the
center of which the pressure is p. Applying the equation of motion in the x-direction

=
x x
a m F (2.18)
x
a z y x
g
z y
x
x
p
p z y
x
x
p
p =
2 2
(2.19)

which reduces to
x x
a a
g x
p
= =
(2.20)

In the vertical direction
=
z z
a m F (2.21)
0
2 2
=
z y x y x
z
z
p
p y x
z
z
p
p
(2.22)
Figure 2.7. A fluid mass subjected to horizontal acceleration

21

which yields
g
z
p
= =
(2.23)

Thus for the case where the fluid is subjected to a horizontal acceleration a
x

( )
2 / 1
2 2
g a
n
p
x
+ =
(2.24)

where n is the direction at right angles to and outward from the liquid surface.
In the foregoing discussion the acceleration of the fluid mass was at right angles to the
direction of the gravitational acceleration. When the acceleration of the fluid mass is in some other
direction, the same general approach can be used. For a fluid subject to both horizontal and vertical
acceleration a
x
and a
z
, it can be shown that

( ) [ ]
2 / 1
2 2 2
g a a
n
p
z x
+ + =
(2.25)

When a
x
= a
z
= 0, this equation reduces to = n p , which is essentially the same as the basic
hydrostatic equation. The last equation indicates that, if fluid in a container is subjected to an
upward acceleration, there will be an increase of pressure within the fluid. A downward acceleration
results in a decrease in pressure.

Figure 2.8. Force on a plane area

2.6. Force on plane area
If the pressure is uniformly distributed over an area, the force is equal to the pressure times
the area, and the point of application of the force is at the centroid of the area. This is the case of
gases where the pressure variation with vertical distance is very small because of the low specific
weight. In the case of liquids the distribution of pressure is not uniform; hence further analysis are
necessary.
In Figure 2.8 let MN be the trace of a plane area making an angle with the horizontal. To
the right is the projection of this area upon a vertical plane. Let h be the variable depth to any point
and y be the corresponding distance from OX, the intersection of the plane produced and the free
surface.

22
Consider an element of area so chosen that the pressure is uniform over it. Such an element
is a horizontal strip. If x denotes the width of the area at any depth, then dA = xdy. As p = h and h
= ysin, the force dF on a horizontal strip is

dA y dA h dA p dF = = = sin (2.26)

The pressure distribution over the area forms a pressure prism, the volume of which is equal to the
total force acting on the area.
Integrating the preceding expression,

= = sin sin A y ydA F

c
(2.27)

where y
c
is the distance to the centroid of the wetted area of the wall A. If the vertical depth of the
centroid is denoted by h
c
, then h
c
= y
c

sin and

A h F
c
= (2.28)

Thus the total force acting on any plane area submerged in a liquid is found by multiplying
the specific weight of the liquid by the product of the area and the depth of its centroid. The value
of F is independent of the angle of inclination of the plane so long as the depth of its centroid is
unchanged (For a plane submerged as in Figure 2.8, it is obvious that this equation applies to one
side only. As the pressures are there identical on the two sides, the net force is zero. In most
practical cases where the thickness of the plane is not negligible, the pressures on the two sides are
not the same).
Since h is the pressure at the centroid, another statement is that the total force on any plane
area submerged in a liquid is the product of the area and the pressure at its centroid.
The point of application of the resultant force on an area is called the center of pressure.
Taking OX in Figure 2.8 as an axis of moments, the moment of an elementary force ysin dA is:

dA y dF y = sin
2
(2.29)

and if y
p
denotes the distance to the center of pressure,

o p
I dA y F y = =

sin sin
2
(2.30)

where I
o
is the moment of inertia of the plane area about an axis through O.
If the preceding expression is divided by the value of F given by A h F
c
= the result is

A y
I
A y
I
y
c
o
c
o
p
=

=

sin
sin
(2.31)

That is, the distance of the center of pressure from the axis where the plane or plane produced
intersects the liquid surface is obtained by dividing the moment of inertia of the area A about the
surface axis by its static moment about the same axis.
This may also be expressed in another form, by noting that

c c o
I A y I + =
2
(2.32)

where I
c
is the moment of inertia of an area about its centroidal axis. Thus

23

A y
I
y
A y
I y A
y
c
c
c
c
c c
p
+ =
+
=
2
(2.33)

From this equation it may be seen that the location of the center of pressure is independent
of the angle ; that is, the plane may be rotated about axis OX without affecting the location of the
center of pressure. Also, it may be seen that the center of pressure is always below the centroid and
that, as the depth of immersion is increased, the center of pressure approaches the centroid.
The lateral location of the center of pressure may be determined by considering the area to
be made up of a series of elemental horizontal strips. The center of pressure for each strip would be
at the midpoint of the strip. Since the moment of the resultant force F, must be equal to the moment
of the distributed force system about any axis, say, the y axis

= dA p x F X
p p
(2.34)

where X
p
is the lateral distance from the selected y axis to the center of pressure of the resultant
force F, and x
p
is the lateral distance to the center of any elemental horizontal strip of area dA on
which the pressure is p.

2.7. Force on curved surface
On any curved area such as MN in Figure 2.8, the forces upon the various elements are
different in direction and magnitude, and an algebraic summation is impossible. But for any
nonplanar area the resultant forces in certain directions can be found.

Figure 2.9. Hydrostatic forces on curved surfaces

2.7.1. Horizontal force on curved surface
Any irregular curved area (Figure 2.9) may be projected upon a vertical plane whose trace is MN.
The projecting elements, which are all horizontal, enclose a volume whose ends are the vertical
plane MN and the irregular surface MN. This volume of liquid is in static equilibrium. Acting on
the vertical projection MN is a force F, and the horizontal force on the irregular area is F
x
.
Gravity W is vertical, and the lateral forces on all the projecting elements are normal to these
elements and hence normal to F. Thus the only horizontal forces on MNMN are F and F
x
, and
therefore F
x
= F.
Hence the horizontal force in any given direction upon any area is equal to the force upon
the projection of that area upon a vertical plane normal to the given direction.


24
2.7.2. Vertical force on curved surface
The vertical force on a curved area, such as MN in Figure 2.9, can be found by considering the
volume of liquid enclosed by the area and vertical elements extending to the free surface. This
volume of liquid is in static equilibrium. Disregarding the pressure on the free surface, the only
vertical forces are gravity W and F
z
, the vertical force on the irregular area. The forces on the
vertical elements are normal to these and hence are horizontal. Therefore, F
z
= W.
Hence the vertical force upon any area is equal to the weight of the volume of fluid
extending above that area to the free surface. The line of action of F
z
must be the same as that of
W; that is, it must pass through the center of gravity of the volume.

2.7.3. A particular case: horizontal and vertical forces on a cylindrical surface
The horizontal force is equal to the force of pressure of the fluid on the vertical projection of the
cylindrical surface
vert c x
A h g F = where h
c
is the depth of submergence of the centre of mass
and A
vert
is the wetted area of the wall, while the vertical component of the pressure force is equal to
the force of gravity acting in the volume V of fluid bounded from above by the free surface of the
fluid, and from below by the curviliniar solid surface under consideration, and from the sides, by
the vertical surface drawn through the perimeter of the walls V g F
z
= . The total force of
pressure acting on a cylindrical surface is ( )
2 / 1
2 2
z x
F F F + = . The direction of the total force of
pressure is determined by the angle formed by the vector F with the horizontal plane
x z
F F = tan .
For the case illustrated in Figure 2.10 we have

Figure 2.10. Horizontal and vertical forces on a cylindrical surface

2.8. Buoyancy of submerged and floating bodies
The body DHCK immersed in the fluid in Figure 2.11 is acted upon by gravity and pressures
of the surrounding fluid. On its upper surface the vertical component of the force is F
z
and is equal
to the weight of the volume of fluid ABCHD. In similar manner the vertical component of force on
the undersurface is
z
F and is equal to the weight of the volume of fluid ABCKD. The difference
between these two volumes is the volume of the body DHCK.
The buoyant force of a fluid is denoted by F
B
, and it is vertically upward and equal to
z z
F F
, which is equal to the weight of the volume of fluid DHCK. That is, the buoyant force on
any body is equal to the weight of fluid displaced.
If the body is in equilibrium, W = F
B
, which means that the densities of body and
fluid are equal.
If W > F
B
, the body will sink.
Horizontal force RL R F
x
2 = ,
L is the length of the cylindrical surface

Vertical force:
( ) L R V V V F
OBAO AO A ABO A z
2
' '
2
1
= = =
(directed upwards)

25
If W < F
B
, the body will rise until its density and that of the fluid are equal, as in
the case of a balloon in the air or, in the case of a free surface, the body will rise
until the weight of the displaced liquid equals the weight of the body. Hence a
floating body displaces a volume of liquid equivalent to its weight. If the body is
less compressible than the fluid, there is a definite level at which it will reach
equilibrium. If it is more compressible than the fluid, it will rise indefinitely,
provided the fluid has no definite limit of height, as in the case of earths
atmosphere.

I llustrative example. Determine whether a gold nugget contains an impurity if its weight in air is
known to be G
0
= 9.65 N and in water G
w
= 9.15 N. The density of pure gold is
g
= 19.310
3

kg/m
3
.
The gold nugget contains no impurity if its density
g
is equal to the density of gold
g
. The
density of the nugget is

n
n
gV
G
0
= .

Using the weight of displaced water, we find G
0
G
w
= V
n
w
g and so,

( )
3 3
0
0
0
0
kg/m 10 3 . 19 =
=
w
w
w
w
n
G G
G
g
G G g
G

The nugget contains no impurity since
g n
= .

Figure 2.11. Forces acting on a submerged body

26
3. KINEMATICS OF FLUID FLOW

In this chapter we deal only with velocities and acceleration and their distribution in space
without consideration of any forces involved.

3.1. Laminar and turbulent flow
That there are two distinctly different types of fluid flow was demonstrated by Osborne
Reynolds in 1883. He injected a fine, threadlike stream of colored liquid having the same density as
water at the entrance to a large glass tube through which water was flowing from a tank. A valve at
the discharge end permitted him to vary the flow. When the velocity in the tube was small, this
colored liquid was visible as a straight line throughout the length of the tube, thus showing that the
particles of water moved in parallel straight lines. As the velocity of the water was gradually
increased by opening the valve further, there was a point at which the flow changed. The line would
first become wavy, and then at a short distance from the entrance it would break into numerous
vortices beyond which the color would be uniformly diffused so that no streamlines could be
distinguished. Later observations have shown that in this latter type of flow the velocities are
continuously subject to irregular fluctuations.
The first type is known as laminar, streamline, or viscous flow. The significance of these
terms is that the fluid appears to move by the sliding of laminations of infinitesimal thickness
relative to adjacent layers; that the particles move in definite and observable paths or streamlines, as
in Figure 3.1; and also that the flow is characteristic of a viscous fluid or is one in which viscosity
plays a significant part (Figure 1.6 of chapter 1).
The second type is known as turbulent flow and is illustrated in Figure 3.2, where (a)
represents the irregular motion of a large number of particles during a very brief time interval, while
(b) shows the erratic path followed by single particle during a longer time interval. A distinguished
characteristic of turbulence is its irregularity, there being no definite frequency, as in wave action,
and no observable pattern, as in the case of eddies.
Large eddies and swirls and irregular movements of large bodies of fluid, which can be
traced to obvious sources of disturbance, do not constitute turbulence, but may be described as
disturbed flow. By contrast, turbulence may be found in what appears to be a very smoothly flowing
stream and one in which there is no apparent source of disturbance. The fluctuations of velocity are
comparatively small and can often be detected only by special instrumentation.

Figure 3.2. Illustrative example of turbulent flow

At a certain instant a particle at O in Figure 3.2b may be moving with the velocity OD, but
in turbulent flow OD will vary continuously both in direction and magnitude. Fluctuations of
velocity are accompanied by fluctuations in pressure, which is the reason why manometers attached
to a pipe in which fluid is flowing usually show pulsations. In this type of flow an individual
particle will follow a very irregular and erratic path, and no two particles may have identical or even
similar motions. Thus a rigid mathematical treatment of turbulent flow is impossible, and instead
statistical means of evaluation must be employed.
Figure 3.1. Illustrative example of laminar flow

27

Figure 3.2c. A turbulent jet of water emerging from a circular orifice into a tank of still water. The fluid
from the orifice is made visible by mixing small amounts of a fluorescing dye and illuminating it with a thin
light sheet. The picture illustrates swirling structures of various sizes amidst an avalanche of complexity. The
boundary between the turbulent flow and the ambient is usually rather sharp and convoluted on many scales.
The object of study is often an ensemble average of many such realizations. Such averages obliterate most of
the interesting aspects seen here, and produce a smooth object that grows linearly with distance downstream.
Even in such smooth objects, the averages vary along the length and width of the flow, these variations being
a measure of the spatial inhomogeneity of turbulence. The inhomogeneity is typically stronger along the
smaller dimension (the width) of the flow. The fluid velocity measured at any point in the flow is an
irregular function of time. The degree of order is not as apparent in time traces as in spatial cuts, and a range
of intermediate scales behaves like fractional Brownian motion.

28

3.2. Steady flow and uniform flow
A steady flow is one in which all conditions at any point in a stream remain constant with
respect to time, but the conditions may be different at different points. A truly uniform flow is one in
which the velocity is the same in both magnitude and direction at a given instant at every point in
the fluid. Both of these definitions must be modified somewhat, because true steady flow is found
only in laminar flow. In turbulent flow there are no continual fluctuations in velocity and pressure at
every point, as has been explained. But if the values fluctuate equally on both sides of a constant
average value, the flow is called steady flow. However, a more exact definition for this case would
be mean steady flow.
Steady (or unsteady) and uniform (or nonuniform) flow can exist independently of each
other, so that any of four combinations is possible. Thus the flow of liquid at a constant rate in a
long straight pipe of constant diameter is steady uniform flow, the flow of liquid at a constant rate
through a conical pipe is steady nonuniform flow, while at a changing rate of flow these cases
become unsteady uniform and unsteady nonuniform flow, respectively.

3.3. Path lines, streamlines, and streak lines
A path line (see Figure 3.2b) is the trace made by a single particle over a period of time. If a
camera were to take a time exposure of a flow in which a fluid particle was colored so it would
register on the negative, the picture would show the course followed by the particle. This would be
its path line. The path line shows the direction of the velocity of the particle at successive instants of
time.
Streamlines show the mean direction of a number of particles at the same instant of time. If
a camera were to take a very short time exposure of a flow in which there were a large number of
particles, each particle would trace a short path, which would indicate its velocity during that brief
interval. A series of curves drawn tangent to the means of the velocity vectors are streamlines.
Path lines and streamlines are identical in the steady flow of a fluid in which there are no
fluctuating velocity components, in other words, for truly steady flow. This is because particles
always move along streamlines, since these lines show the direction of motion of every particle.
Truly steady flow may be either that of an ideal frictionless fluid or that of one so viscous and
moving so slowly that no eddies are formed. This latter is the laminar type of flow, wherein the
layers of fluid slide smoothly, one upon another. In turbulent flow, however, path lines and
streamlines are not coincident, the path lines being very irregular while the streamlines are
everywhere tangent to the local mean temporal velocity.
In experimental fluid mechanics, a dye or other tracer is frequently injected into the flow to
trace the motion of the fluid particles. If the flow is laminar, a ribbon of color results. This is called
a streak line, or filament line. It is an instantaneous picture of the positions of all particles in the
flow which have passed through a given point (namely, the point of injection).

3.4. Flow rate and flow mean
The quantity of fluid flowing per unit time is called the flow rate. It may be expressed in
terms of volume flow rate (in SI units cubic meters per second), or mass flow (in SI units kilograms
per second). In dealing with incompressible fluids, volume flow rate is commonly used, whereas
mass flow rate is more convenient with compressible fluids.
Figure 3.3 illustrates a streamline in steady flow lying in the xz plane. Element of area dA
lies in the yz plane. The mean velocity at point P is u. The volume flow rate through the element of
area dA is
( ) ( ) A d u dA u dA u d dQ = = = = cos cos A u (3.1)

where dA is the projection of dA on the plane normal to the direction of u. This indicates that the
volume flow rate is equal to the magnitude of the velocity multiplied by the flow area at right angles

29
to the direction of the velocity. The mass flow rate may be computed by multiplying the volume
flow rate by the density of the fluid.

If the flow is turbulent, the instantaneous velocity component u along the streamline will
fluctuate with time, even though the flow is nominally steady. A plot of u as a function of time is
shown in Figure 3.4. The average ordinate of u over a period of time determines the temporal
mean value of velocity u at point P.
The difference between u and u, which may be designated as u , is called the turbulent
fluctuation of this component. Thus at any instant

u u u + = (3.2)
and u may be evaluated for any finite time t as ( )
=
t
dt u t u
0
1 .

The flow rate may be expressed as

v A dA u Q
A
= =
(3.3)
or
v A dA u Q
A
m
= =

(3.4)

where u is the temporal mean velocity through an infinitesimal area dA, while v is the mean, or
average, velocity over the entire sectional area A (Note that the are A is defined by the surface at
right angles to the velocity vectors). Q is the volume flow rate (m
3
/s) and Q
m
is the mass flow rate
(kg/s). If u is known as a function of A, the foregoing may be integrated. If the flow rate has been
determined directly by some other method, the mean velocity may be found by

Figure 3.3. A streamline in a steady flow
Figure 3.4. Fluctuating velocity at a point

30

A
Q
A
Q
v
m
= =
(3.5)

3.5. Equation of continuity
Figure 3.5 illustrates a short length of a stream tube, which may be assumed, for practical
purposes, as a collection of streamlines. Since the stream tube is bounded on all sides by streamlines
and since there can be no net velocity normal to a streamline, no fluid can leave or enter the stream
tube except at the ends. The fixed volume between the two fixed sections of the stream tube is
known as the control volume and its magnitude will designated by According to
Newtonian physics (i.e., disregarding the possibility of converting mass to energy), mass must be
conserved. If the mass of the fluid contained in the control volume of volume () at time t is
mass, then the mass of fluid contained in at time t + dt would be

( ) ( )dt dA u dt dA u
t dt t 2 2 2 1 1 1
mass mass + =
+
(3.6)

But the mass contained in at t + dt can also be expressed as:

( ) vol dt
t
t dt t

+ =
+

mass mass (3.7)

where t is the time rate of change of the mean density of the fluid in . Equating these
two expressions for mass
t+dt
yields:

( ) ( ) ) (
2 2 2 1 1 1
vol dt
t
dt dA u dt dA u
=

(3.8)
and

=
vol A A
vol d
t
dA u dA u ) (
2 1
2 2 2 1 1 1

(3.9)

This is the general equation of continuity for flow through regions with fixed boundaries. It states
that the net rate of mass inflow to the control volume is equal to the rate of increase of mass within
the control volume. This equation can be reduced to more useful forms.

For steady flow, t = 0 and

Figure 3.5. Length of stream tube as control volume

31

=
2 1
2 2 2 1 1 1
A A
dA u dA u (3.10)
or
m
Q v A v A = =
2 2 2 1 1 1
(3.11)

These are the continuity equations that apply to steady, compressible or incompressible flow within
rigid boundaries.
If the fluid is incompressible, = constant and thus

=
2 1
2 2 1 1
A A
dA u dA u (3.12)
or
Q v A v A = =
2 2 1 1
(3.13)

This is the continuity equation that applies to incompressible fluids for both steady and unsteady
flow within rigid boundaries.
These equations are generally adequate for the analysis of flows in conduits with rigid
boundaries, but for the consideration of flow in space, as that of air around airplane, for example, it
is desirable to express the continuity equation in another, as indicated below.

3.6. Differential equation of continuity
Figure 3.6 shows three coordinate axes x, y, z mutually perpendicular and fixed in space. Let
the velocity components in these three directions be u, v, w, respectively. Consider now a small
parallelepiped, having sides x, y, z. In the x direction the rate of mass flow into this box through
the left-hand face is approximately z y u , this expression becoming exact in the limit as the box
is shrunk to a point. The corresponding rate of mass flow out of the box through the right-hand face
is ( ) [ ] { } z y x x u u + . Thus the net rate of mass flow into the box in the x direction is
( ) [ ] z y x x u . Similar expressions may be obtained for the y and z directions. The sum of
the rates of mass inflow in the three directions must equal the rate of change of the mass in the box,
or ( ) z y x t . Summing up, applying the limiting process, and dividing both sides of the
equation by the volume of the parallelepiped, which is common to all terms, we get

( ) ( ) ( )
t z
w
y
v
x
u

(3.14)

which is the equation of continuity in its most general form.

Figure 3.6. An elemental volume of fluid

32

This equation is, of course, valid regardless of whether the fluid is a real one or an ideal one. If the
flow is steady, does not vary with time, but it may vary in space. Since
( ) ( ) ( ) x u x u x u + = , it follows that for steady flow the equation may be written

0 =
z
w
y
v
x
u
z
w
y
v
x
u

(3.15)

In the case of an incompressible fluid, whether the flow is steady or not, the equation of continuity
becomes
0 =
z
w
y
v
x
u
(3.16)

Figure 3.7. Two-dimensional flow along a curved path. (a) Irrotational flow. (b) Rotational flow.

3.7. Rotational and irrotational flow
Irrotational flow may be briefly described as flow in which each element of the moving fluid
suffers no net rotation from one instant to the next, with respect to a given frame of reference. In
irrotational flow, however, a fluid element may deform as shown in Figure 3.7a, where the axes of
the element rotate equally toward or away from each other. As long as the algebraic average
rotation is zero, the motion is irrotational. In Figure 3.7b is depicted an example of rotational flow.
In this case there is a net rotation of the fluid element. Actually, the deformation of the element in
Figure 3.7b is less than that of Figure 3.7a.

Figure 3.8. Deformation of a fluid element

33

Let us now express the condition of irrotationality in mathematical terms. It will help to
restrict the discussion at first to two-dimensional motion in the xy plane. Consider a small element
moving as depicted in Figure 3.8a. During a short time interval t the element moves from one
position to another and in the process it deforms as indicated. Superimposing A on A, defining an x
axis along AB, and enlarging the diagram, we get Figure 3.8b. The angle between AB and AB
can be expressed as

( ) [ ]
t
x
v
x
t x x v
x
B B

= (3.17)

Hence the rate of rotation of the edge of the element that was originally aligned with AB is

x
v
t
(3.18)
Likewise
( ) [ ]
t
y
u
y
t y y u

=
/
(3.19)

and the rate of rotation of the edge of the element that was originally aligned with AC is

y
u
t
(3.20)

with the negative sign because +u is directed to the right. The rate of rotation of the element about
the z axis is now defined to be
z
, the average of
and
, thus

=
y
u
x
v
z
2
1
(3.21)

But the criterion we originally stipulated for irrotational flow was that the rate of rotation be zero.
Therefore the flow is irrotational in the xy plane if

y
u
x
v
(3.22)

In three-dimensional flow the corresponding expressions for the components of angular-
deformation rates about the x and y axes are

=
z
v
y
w
x
2
1

=
x
w
z
u
y
2
1
(3. 23a,b)

Thus v
r r
rot
2
1
= . Finally, for the general case, irrotational flow is defined to be that for which
0 = = =
z y x
(3.24)

The primary significance of irrotational flow is that ideal (frictionless) flow is irrotational.

34

3.8. Circulation
To get a better understanding of the character of a flow field, we should acquaint ourselves
with the concept of circulation. Let the streamlines of Figure 3.9 represent a two-dimensional flow
field, while L represents any closed path in this field. The circulation is defined mathematically as
a line integral of the velocity about a closed path. Thus

= =
L L
dL v d cos L v (3.25)

where v is the velocity in the flow field at the element dL of the path, and is the angle between v
and the tangent to the path (in the positive direction along the path) at that point. This equation is
analogous to the common equation in mechanics for work done as a body moves along a curved
path while the force makes some angle with the path. The only difference here is the substitution of
a velocity for a force.

In a two-dimensional flow field we can write

dy dx
y
u
x
v
d

= (3.26)

The vorticity is defined as the circulation per unit of enclosed area. Thus

y
u
x
v
dy dx
d
= (3.27)

and we note that an irrotational flow is one for which the vorticity = 0. Similarly, if the flow is
rotational, 0.

3.9. Velocity and acceleration in steady flow
In a typical three-dimensional flow field the velocities are everywhere different in
magnitude and direction. Also, the velocity at any point in the field may change with time. Let us
first consider the case where the flow is steady and thus independent of time. If the velocity of a
fluid particle has the components u, v, and w parallel to the x, y, and z axes, then for steady flow:

( ) z y x u u
st
, , = ( ) z y x v v
st
, , = ( ) z y x w w
st
, , =

Figure 3.9. Circulation around a closed path in a
two-dimensional velocity field

35
Applying the chain rule of partial differentiation, the acceleration of the fluid particle for steady
flow can be expressed as
( )
t d
z d
z t d
y d
y t d
x d
x
z y x
dt
d
st
= =
V V V
V , , a (3.28)
where
( )
2 / 1
2 2 2
w v u + + = V (3.29)

Noting that u t d x d = , v t d y d = , and w t d z d = , we have

z
w
y
v
x
u
st
=
V V V
a (3.30)

This equation can be written as three scalar components

( )
z
u
w
y
u
v
x
u
u a
st x
= (3.31a)
( )
z
v
w
y
v
v
x
v
u a
st
y
= (3.31b)
( )
z
w
w
y
w
v
x
w
u a
st z
= (3.31c)

These equation show that even though the flow is steady, the fluid may possess an
acceleration by virtue of a change in velocity with change in position. Thus type of acceleration is
commonly referred to as convective acceleration.

3.10. Velocity and acceleration in unsteady flow
If the flow is unsteady, then the velocity components take the form

( ) t z y x u u , , , = ( ) t z y x v v , , , = ( ) t z y x w w , , , =

Following a similar procedure to that of the preceding section results in the following set of scalar
equations
z
u
w
y
u
v
x
u
u
t
u
a
x
= (3.32a)
z
v
w
y
v
v
x
v
u
t
v
a
y
= (3.32b)
z
w
w
y
w
v
x
w
u
t
w
a
z
= (3.32c)

In the above set of equations the terms containing t represent the acceleration caused by
the unsteadiness of the flow. This type of acceleration is commonly referred to as the local
acceleration.
In the case of uniform flow (streamlines parallel to one another) the convective acceleration
is zero and
t
=
V
a (3.33)
while in the case of steady flow the local acceleration becomes zero.

36
4. BASIC HYDRODYNAMICS

In this chapter we discuss various mathematical methods of describing the flow of fluids.
The presentation here provides some notion of the possibilities of a rigorous mathematical approach
to flow problem.

4.1. The stream function
The stream function , based on the continuity principle, is a mathematical expression that
describes a flow field. In Figure 4.1 are shown two adjacent streamlines of a two-dimensional flow
field. Let ( ) y x, represent the streamline nearest the origin. Then d + is representative of the
second streamline. Since there is no flow across a streamline, we can let be indicative of the flow
carried through the area from the origin O to the first streamline. And thus d represents the flow
carried between the two streamlines of Figure 4.1.

From continuity referring to the triangular fluid element of Figure 4.1, we see that for an
incompressible fluid
y d u x d v d + = . (4.1)

The total derivative d may be also express as

y d
y
x d
x
d
=

. (4.2)

Comparing these last equations, we note that

y
u
=

and
x
v
=

. (4.3)

Thus, if can be expressed as a function of x and y, we can find the velocity components (u and v)
at any point of a two-dimensional flow field by application of the last equations. It should be noted
that since the derivation of is based on the principle of continuity, it is necessary that continuity
be satisfied for the stream function to exist. Also, since vorticity was not considered in the
derivation of , the flow need not be irrotational for the stream function to exist.
The equation of continuity
0 =
y
v
x
u
(4.4)

may be expressed in terms of by substituting the expressions for u and v, and doing so we get

Figure 4.1. Stream function

37
0 =
x y y x

or
x y y x
=

2 2
, (4.5)

which shows that, if = (x,y), the derivatives taken in either order give the same result and that a
flow described by a stream function automatically satisfies the continuity equation.

4.2. Velocity potential
For two-dimensional flow the velocity potential = (x,y) may be defined in cartesian
coordinates as
x
u
=

and
y
v
=

. (4.6)

If we substitute these expressions into the continuity equation, we get

0
2
2
2
2
=
y x

, (4.7)

which is known as the Laplace equation.
If the expressions of the velocity components are substituted into the equation of vorticity,
we get
0
2 2
=

=
x y y x x y y x y
u
x
v
. (4.8)

Since = 0, the flow is irrotational, and thus, if a velocity potential exists, the flow must be
irrotational. Conversely, if the flow is rotational, the velocity potential does not exist.
The rotation of fluid particles requires the application of torque, which in turn depends on
shearing forces. Such forces are possible only in a viscous fluid. In inviscid (or ideal) fluids there
can be no shears and hence no torques. Hence the flow of an ideal fluid is irrotational.

4.3. Orthogonality of streamlines and equipotential lines
Figure 4.2. Orthogonality of streamlines
and equipotential lines

38
It was found that
y d
y
x d
x
d
=

. (4.9)
Similarly,
y d
y
x d
x
d
=

. (4.10)

Using the expressions for the velocity components we can express these two equations as

y d u x d v d + = (4.11)
and
y d v x d u d = . (4.12)

Along a streamline, = constant, so d = 0, and from the first equation we get u v x d y d / / = .
Along an equipotential line, = constant, so d = 0, and from the second equation we get
v u x d y d / / = . Geometrically, this tells us that the streamlines and equipotential lines are
orthogonal, or everywhere perpendicular to each other.
The equipotential lines = C
i
and the streamlines = K
i
, where C
i
and K
i
have equal
increments between adjacent lines, form a network of intersecting perpendicular lines which is
called the flow net (Figure 4.2).

4.4. An illustrative example
A flow is defined by u = 2x and v = -2y. Find the stream function and potential function for
this flow.
Check continuity:
0 2 2 = =
y
v
x
u
. (4.13)

Hence continuity is satisfied and it is possible for a stream function to exist

xdy ydx dy u dx v d 2 2 + = + = , (4.14)

1
2 C xy + = . (4.15)

Check to see if the flow is irrotational:

0 0 0 = =
y
u
x
v
. (4.16)

Hence the flow is irrotational and a potential function exists

ydy xdx dy v dx u d 2 2 + = = , (4.17)

2
2 2
) ( C y x + + = . (4.18)

The location of lines of equal can be found by substituting values of into the expression =
2xy. Thus for = 60, x = 30/y. In a similar fashion lines of equal potential can be plotted. For
example, for = 60 we have x = (y
2
- 60)
0.5
. The flow net depicts flow in a corner.

39

4.5. Energy considerations in fluid flow
In this paragraph fluid flow is approached from the viewpoint of energy considerations. The
first law of thermodynamics tells us that energy can be neither created nor destroyed. Moreover, all
forms of energy are equivalent. In the following sections the various forms of energy present in
fluid flow are discussed.

4.5.1. Kinetic energy of a flowing fluid
A body of mass m when moving at a velocity V possesses a kinetic energy,
2
2
1
KE mV = .
Thus if a fluid were flowing with all particles moving at the same velocity, its kinetic energy would
be also
2
2
1
mV ; this can be written as:

( )
g
V
volume
V volume
volume
mV
2
2
1
2
1
weight
KE
2
2 2
=
.

In SI units the term g V 2
2
is expressed in meters.
In the flow of a real fluid the velocities of different particles will usually not be the same, so
it is necessary to integrate all portions of the stream to obtain the true value of the kinetic energy. It
is convenient to express the true value in terms of the mean velocity V and a factor . Hence,

g
V
2 weight
KE True
2
= .

4.5.2. Potential energy
The potential energy of a particle of fluid depends on its elevation above any arbitrary datum
plane. We are usually interested only in differences in elevation and therefore the location of the
datum plane is determined solely by considerations of convenience. A fluid particle of weight W
situated at a distance z above datum plane possesses a potential energy Wz. Thus, its potential
energy per unit weight is z.

4.5.3. Internal energy
Internal energy is the energy due to the motion of molecules and forces of attraction between
them. Internal energy is a function of temperature; it can expressed in terms of energy per unit mass
i or in terms of energy per unit of weight I. Note that i = gI.
The zero of internal energy may be taken at any arbitrary temperature, since we are usually
concerned only with differences. For a unit mass, i = c
v
T, where c
v
is the specific heat at
constant volume whose units are Nm/(kgK). Thus i is expressed in Nm/kg and the internal energy
I per unit of weight is expressed in meters.

4.5.4. General equation for steady flow of any fluid
The first law of thermodynamics states that for steady flow the external work done on any system
plus the thermal energy transferred into or out of the system is equal to the change of energy of the system.
Thus, for steady flow,
work + heat = energy.

Let us now apply the first law of thermodynamics to the fluid system defined by the fluid
mass contained at time t in the control volume between sections 1 and 2 of the stream tube in Figure
4.3. The control volume is fixed in position and does not move or change shape (Figure 4.3(b)). The

40
fluid system we are dealing with consists of the fluid that was contained between sections 1 and 2 at
time t. This fluid system moves to a new position during time interval t, as indicated in Figure 4.3.

Figure 4.3. Work, heat, and energy for a stream tube

During this short time interval we shall assume that the fluid moves a short distance ds
1
at section 1
and ds
2
at section 2. In this discussion we restrict ourselves to steady flow so that
2 2 2 1 1 1
ds A ds A = .
In moving these short distances, work is done on the fluid system by the pressure forces p
1
A
1
and
p
2
A
2
. This work is referred to as flow work. It may be expressed as

2 2 2 1 1 1
work flow ds A p ds A p = . (4.19)

The minus sign in the second term indicates that the force and displacement are in opposite
directions.
In addition to flow work, if there is a machine between section 1 and 2 there will be a shaft
work. During the short time interval dt we can write

( )
M M
h ds A dt h
dt
ds
A =
= =
1 1 1
1
1 1
time
weight
energy
time
weight
shaft work , (4.20)

where h
M
is the energy put into the flow by the machine per unit weight of flowing fluid. If the
machine is a pump, h
M
is positive; if the machine is a turbine, h
M
is negative.
The heat transferred from an external source into the fluid system over time interval dt is

( )
H H
Q ds A dt Q
dt
ds
A =
=
1 1 1
1
1 1
heat , (4.21)

where Q
H
is the energy put into the flow by the external heat source per unit weight of flowing
fluid. If the heat flow is out of the fluid, the value of Q
H
is negative.
In using the concept of the control volume, we consider a fluid system defined by the mass
of fluid contained in the control volume at time t. At time t + dt this same mass of fluid has moved
to a new position. At that instant the energy E
2
of the fluid system equals the energy E
1
that was
possessed by the fluid mass when it was coincident with the control volume at time t plus the
energy E
out
that flowed out of the control volume during time interval dt minus the energy E
in

that flowed into the control volume during time interval dt. Thus

in out
E E E E + =
1 2
. (4.22)

41

Hence the change in energy E of the fluid system under consideration during time interval dt is

in out
E E E E E = =
1 2
. (4.23)

During time interval dt the weight of fluid entering at section 1 is ( )
1 1 1
ds A and for steady flow
an equal weight must leave section 2 during the same time interval. Hence the energy E
in
which
enters at section 1 during time dt is ( ) ( ) ( )
1
2
1 1 1 1 1
2 I g V z ds A + + , while that which leaves, E
out
,
at section 2 is represented by a similar expression. Thus

( ) ( )

+ +
+ + = =
1
2
1
1 1 1 1 1 2
2
2
2 2 2 2 2
2 2
energy I
g
V
z ds A I
g
V
z ds A E . (4.24)

Applying the first law of thermodynamics, at the same time factoring out
( ) ( )
2 2 2 1 1 1
ds A ds A = for steady flow and rearranging, we get

+ +
+ + = + +
1
2
1
1 1 2
2
2
2 2
2
2
1
1
2 2
I
g
V
z I
g
V
z Q h
p p
M M

, (4.25)
or
+ + + = + +
+ + +
2 2
2
2
2
2
2 1 1
1
1
2
1
1
2 2
I z
p
g
V
Q h I z
p
g
V
M M
. (4.26)

This equation applies to liquids, gases, and vapours, and to ideal fluids as well as to real
fluids with friction. The only restriction is that it is for steady flow. In many cases this equation is
greatly shortened because certain quantities are equal and thus cancel each other, or are zero. Thus,
if two points are at the same elevation, z
1
z
2
= 0. If the conduit is well insulated or if the
temperature of the fluid and that of its surroundings are practically the same, Q
H
may be taken as
zero. If there is no machine between sections 1 and 2, then the term h
M
drops out.
In the above equation each term has the dimension of length. Thus p/, called pressure head,
represents the energy per unit weight stored in the fluid by virtue of the pressure under which the
fluid exists; z, called elevation head, represents the potential energy per unit weight of fluid; and
V
2
/(2g), called velocity head, represents the kinetic energy per unit weight of fluid. The sum of
these three terms is called the total head and is denoted by H, where

z
p
g
V
H + + =
2
2
. (4.27)

In this equation we have disregarded . In turbulent flow the value of is only a little more than
unity and thus it can be assumed equal to unity. If the flow is laminar, the velocity head is usually
very small compared to the other terms and thus little error is introduced if is set to 1 rather than
2, its true value.

4.5.5. Equation for unsteady flow along a streamline for ideal fluid
Let us consider now the frictionless unsteady flow of a fluid along a streamline. Referring to
Figure 4.3 we note p
1
p, p
2
(p + dp), z
2
z
1
dz and A
1
A
2
dA. We will use Newtons
second law, and express the acceleration of the fluid element as ( ) dt dV ds dV V + . Applying
= ma F to the fluid element we get for unsteady flow


42

L
h
dt
dV
ds
dV
V dsdA dAdz dpdA +
+ = . (4.28)

In this equation h
L
represents the head loss between sections 1 and 2. Dividing through by dA
gives
L
h
dt
dV
ds gdz VdV
dp
= + + +
. (4.29)

The term
dt
dV
ds accounts for the effect of acceleration caused by the unsteadiness of the flow. This
equation may also be expressed as

L
h
dt
dV
g
ds
dz
g
V
d
dp
= + + +
2
2
. (4.30)

This equation applies to unsteady flow of both compressible and incompressible real fluids.
However, an equation of state relating to p must be introduced before integration if we are
dealing with a compressible fluid. For an incompressible fluid ( = constant) we can integrate from
some section 1 to another section 2, where the distance between them is L, to obtain:

L
h
dt
dV
g
L
z z
g
V
g
V p p
= + + +
1 2
2
1
2
2 1 2
2 2
, (4.31)
or
dt
dV
g
L
z
g
V p
h z
g
V p
L
+
+ + = +
+ +
2
2
2 2
1
2
1 1
2 2
. (4.32)

It should be noted here that if a pipe, for example, consists of two or more pipes in series, an
dt
dV
g
L
(the accelerative head) term each pipe should appear in the equation just as there would be a
separate term for the head loss in each pipe.

4.6. Momentum and forces in fluid flow
Previously, two important fundamentals concepts of fluid mechanics were presented: the
continuity equations and the energy equation. In this paragraph a third basic concept, the impulse-
momentum principle will be developed.
Consider now a flow that may be incompressible or incompressible, real or ideal, steady or
unsteady. Newtons second law may be expressed as

( )
=
dt
m d V
F . (4.33)

Thus, the sum of the external forces on a body is equal to the rate of change of momentum of that
body. This equation can also be expressed as ( ) ( )
= V F m d dt , i.e., impulse equals change of

momentum, hence the terminology impulse-momentum principle used.
Let us apply this equation to a body defined by the mass of fluid contained at time t in the
control volume of Figure 4.4. Henceforth we shall refer to this mass of fluid as the fluid system. The
control volume is fixed in position; it does not move, nor does it change shape or size. At time t +

43
t the fluid mass we are dealing with (i.e., the fluid system) has moved to a new position indicated
by the shaded area of Figure 4.4. Let us now define some terms:

(mV)
t
Momentum at time t of the fluid system (coincident with the control volume at
time t)

(mV)
t+t
Momentum at time t + t of the fluid system (coincident with the shaded area of
Figure 4.4 at time t + t)

(mV)
t
Momentum of the fluid mass contained within the control volume at time t

(mV)
t+t
Momentum of the fluid mass contained within the control volume at time t + t

(mV)
out
Momentum of the fluid mass that leaves the control volume during time interval t

(mV)
in
Momentum of the fluid mass that enters the control volume during time interval t

Figure 4.4. Control volume for general case. (left) fluid mass acted on by certain forces. (right) location of
the fluid system at times t and t + t

At time t the momentum of the fluid system is equal to the momentum of the fluid mass
contained in the control volume at time t because the same fluid mass is involved in both cases.
Thus
(mV)
t
= (mV)
t
. (4.34)

At time t + t the momentum of the fluid system is equal to the momentum of the fluid mass in the
control volume at t + t plus the momentum of the mass that has flowed out of the control volume
during time interval t minus that which flowed into the control volume during time interval t.
Thus
( ) ( ) ( ) ( )
in out t t t t
mV mV V m mV + =
+ +
. (4.35)

The change of momentum of the fluid system is

( ) ( ) ( )
t t t
mV mV mV =
+
, (4.36)

44

and thus we get
( ) ( ) ( ) ( ) ( )
in out t t t
mV mV V m V m mV + =
+
. (4.37)

Applying Newtons second law, dividing through by t, rearranging, and noting that the
limit of ( ) ( ) dt mV d t mV = as t 0, we get

( ) ( ) ( ) ( ) ( ) ( )
dt
m m
dt
m d m d
dt
m d
t
m
t t t in out
t
V V V V V V
F

+
= =
=
+
0
lim . (4.38)

The above equation states that the force acting on a fluid mass is equal to the rate of change of the
momentum of the fluid mass which, in turn, is equal to the sum of the two terms on the right-hand
side of the equation. The first term on the right side of the equation represents the net rate of
outflow of momentum across the control surface while the second term represents the rate of
accumulation of momentum within the control volume. In the case of steady flow, the last term of
the equation is equal to zero and the equation is

( ) ( )
dt
m d m d
in out
V V
F

=
. (4.39)

Thus, for steady flow the force on the fluid mass is equal to the net rate of outflow of momentum
across the control surface.
It is advantageous to select a control volume such that the control surface is normal to the
velocity where it cuts the flow. Consider such a situation depicted in Figure 4.5. Also, let the
velocity be constant where it cuts across the control surface. In Figure 4.5 the fluid system we are
dealing with is contained between sections 1 and 2 at time t.

This fluid system moves to a new position during time interval dt, as indicated in Figure 4.5. During
this short interval we will assume the fluid moves a short distance ds
1
at section 1 and ds
2
at section
2. Also, we are restricting ourselves to steady flow. The momentum crossing the control surface at
section 1 during the interval dt is
1 1 1 1
V ds A while that crossing section 2 is
2 2 2 2
V ds A .
Noting that since the control surface cuts the velocity at right angles,

V = ds/dt and Q = AV, (4.40)

we get for steady flow along a stream tube

Figure 4.5. Control volume for steady flow with control
surface cutting a constant velocity stream at
right angles

45
1 1 1 2 2 2
V V F =
Q Q . (4.41)

From continuity, for steady flow,

2 2 1 1
Q Q Q = = ; (4.42)

thus we can write

( ) V V V F = =
Q Q
1 2
. (4.43)

The
F represents the vectorial summation of all forces acting on the fluid mass including
gravity forces, shear forces, and pressure forces including those exerted by fluid surrounding the
fluid mass under consideration as well as the pressure forces exerted by the solid boundaries in
contact with the fluid mass.
The corresponding scalar equations are:

( )
x x x x
V Q V V Q F = =

1 2
, (4.44)

( )
y y y y
V Q V V Q F = =

1 2
, (4.45)

( )
z z z x
V Q V V Q F = =

1 2
. (4.46)

If the velocity is not uniform over a section, the momentum transfer across that section is
greater than that compute by using the mean velocity. Hence a correction factor must be introduced
that multiply the term V Q :

=
A
dA u
AV
2
2
1
. (4.47)

For laminar flow in a circular pipe, 3 4 = , but for turbulent flow in circular pipes, it usually
ranges from 1.005 to 1.05.

In the succeeding sections these equations will be applied to several situations that are
commonly encountered in engineering practice. The big advantage of the impulse-momentum
principle is that one need not be concerned with the details of what is occurring within the flow,
only the conditions at the end sections of the control volume govern the analysis.

4.7. Force exerted on pressure conduits
Consider the case of horizontal flow to the right through the reducer of Figure 4.6(a). A
free-body diagram of the forces acting on the fluid mass contained in the reducer is shown in Figure
4.6(b). We shall apply the momentum-impulse equations to this fluid mass to examine the forces
that are acting in the x direction. The forces p
1
A
1
and p
2
A
2
represent pressure forces exerted by fluid
located just upstream and just downstream of the fluid mass under consideration. The force ( )
x F R
F
/

represents the force exerted by the reducer on the fluid in the x direction. Neglecting shear forces at
the boundary of the reducer, the force ( )
x F R
F
/
is the integrated effect of the normal pressure forces
that are exerted on the fluid by the wall of the reducer. The intensity of pressure at the wall will
decrease as the diameter decreases because of the increase in velocity head.


46

Figure 4.6. Forces on a reducer

Applying the momentum-impulse equations and assuming the fluid to be ideal with ( )
x F R
F
/

directed as shown, we get

( ) ( )
= =
1 2 / 2 2 1 1
v v Q F A p A p F
x F R x
. (4.48)

In this equation each term can be evaluated independently from given flow data, except ( )
x F R
F
/
,
which is the quantity we wish to find. Thus

( ) ( )
1 2 2 2 1 1 /
v v Q A p A p F
x F R
= . (4.49)

This gives the value of the total force exerted by the reducer on the fluid in the x direction. This
force acts to the left as assumed in Figure 4.6. The force of the fluid on the reducer is, of course,
equal and opposite to that of the reducer on the fluid. If friction were considered with the flow to the
right, ( )
x F R
F
/
would be somewhat larger.
Consideration of the weight of fluid between sections 1 and 2 in Figure 4.6 results in the
conclusion that pressure are larger on the bottom half of the pipe than on the upper half. It should be
noted that it is the conditions at the end sections of the control volume that govern the analysis.
What occurs within the flow between sections 1 and 2 is unimportant so far as the determination of
forces is concerned.
If the fluid undergoes a change in both direction and velocity, as in the reducing pipe bend
in Figure 4.7, the procedure is similar to that of the preceding case, except that it is convenient to
deal with components. Assuming the flow is in a horizontal plane so that the weight can be
neglected, applying the momentum-impulse equations by summing up forces acting on the fluid in
the x direction, and equating them to the change in fluid momentum in the x direction gives

( ) ( )
1 2 2 2 1 1 /
cos cos v v Q A p A p F
x F B
= . (4.50)

Similarly, in the y direction

( ) sin sin
2 2 2 /
v Q A p F
y F B
+ = . (4.51)

In a specific case, if the numerical values determined by these equations are positive, then the
assumed directions are correct. A negative value for either one merely indicates that that component
is in the direction opposite from that assumed.
Note that V F = Q is the resultant of all forces acting on the fluid, which includes the
pressure forces on the two ends, while
F B
F
/
is the force exerted by the bend on the fluid. The value
of
F B
F
/
is
( ) ( )
2
/
2
/ / y F B x F B F B
F F F + = , (4.52)

47

and it is represented by the closing line in the force diagram shown in Figure 4.7. The total force
exerted by the fluid on the bend is equal to
F B
F
/
, but it is opposite to the direction shown in the
figure.

Figure 4.7. Forces on a reducing bend

It may be seen that such a force tends to move the portion of the pipe considered. Hence,
where such changes in velocity or alignment occur, a large pipe will usually be anchored by
attaching it to a concrete block of sufficient weight to provide the necessary resistance.

I llustrative example. Determine the magnitude and direction of the force exerted by the liquid
(water) on this double nozzle. Both nozzle jets have a velocity of 12 m/s and D
1
= 15 cm, D
2
= 10
cm, and D
3
= 7.5 cm. The axes of the pipe and both nozzles lie in a horizontal plane. Neglect
friction.
Continuity:
3
2
3
2
2
2
1
2
1
4 4 4
v
D
v
D
v
D
+ = , 33 . 8
2
1
3
2
3 2
2
2
1
=
+
=
D
v D v D
v m/s

147 . 0
4
1
2
1
1
= = v
D
Q

m
3
/s, 094 . 0
2
= Q m
3
/s, 053 . 0
3
= Q m
3
/s

Energy equation:

2
2
2 1
2
1
2 2
p
g
v p
g
v
+ = + 0
81 . 9 2
12
81 . 9 2
33 . 8
2
1
2
+
= +

p
2 . 37
1
= p kN/m
2
p
1
A
1
= 0.656 kN

Momentum-impulse principle:

( ) ( )
x x x x L N x
v Q v Q v Q F A p F
1 1 3 3 2 2 / 1 1
+ = =
,
1 1
v v
x
= ,
O
2 2
15 cos v v
x
= ,
O
3 3
30 cos v v
x
=

( ) ( )
y y y y L N y
v Q v Q v Q F F
1 1 3 3 2 2 /
+ = =
,
0
1
=
y
v ,
O
2 2
15 sin v v
y
= ,
O
3 3
30 sin v v
y
=


48
( ) 241 . 0
/
=
x L N
F kN, ( ) 026 . 0
/
=
y L N
F kN

The minus sign indicates that the assumed direction of ( )
y L N
F
/
was wrong. Therefore, ( )
y L N
F
/
, acts
in the negative y direction. Equal and opposite to
L N
F
/
is
N L
F
/

( ) 241 . 0
/
=
x N L
F kN (in positive x direction)
( ) 026 . 0
/
=
y N L
F kN (in positive y direction)

Figure 4.8. Forces on a double nozzles

4.8. Reaction of a jet
In Figure 4.9 consider a jet issuing steadily from a tank which is large enough so that the
velocity within it may be neglected. Let the area of the jet be A
2
and its velocity v
2
, and assuming an
ideal fluid, gh v 2
2
= . Applying the momentum-impulse principle we get

( ) ( ) ( )
2 2
2
2 2 2 2 /
2 2 0 A h gh A v A v Q F F
x L T x
= = = = = . (4.53)

This is the net force of the tank on the liquid in the x direction; it acts to the right and causes the
change of velocity of the flowing liquid from zero to v
2
. Equal and opposite to this force is the force
of the liquid on the tank, often referred to as the jet reaction. If the tank were supported by

49
frictionless rollers, it would be moved to the left by this action. The net force
2 2
v Q is equal to
the difference in the magnitude of the pressure forces on the two ends of the tank.

Figure 4.9. A jet issuing steadily from a tank

Figure 4.10. Same of Figure 4.9 but with a jet discharged into the tank

Refer now to Figure 4.10 where a jet of liquid of cross-sectional area A
1
is discharged into
the tank with a velocity v
1
. In this case a force
1 1
v F = Q is exerted by the jet on the liquid
which, in turn, transmits the force to the tank. This is referred to as the jet action.
The resultant force on the tank caused by one jet entering the tank at section 1 and the other
jet leaving the tank at section 2 is the vector sum of
1 1
v Q and
2 2
v Q where the first vector
(jet action) acts in the direction of v
1
(downward to the right in Figure 4.10) and the second vector
(jet reaction) acts in the direction opposite to that of v
2
. Thus a jet entering a system acts on the
system in the direction in which the jet is traveling while a jet leaving a system acts on the system in
the direction opposite to that in which the jet is traveling.

4.9. J et propulsion
In section 4.8. an expression was derived for the reaction of a jet from a stationary tank.
Assume now that the tank in Figure 4.9 moves to the left with a velocity u. If the orifice is small
compared with the size of the tank, the relative velocity within the latter may be disregarded, as

50
may also any change in h for a short interval of time. Thus the absolute velocity of the fluid within
the tank is V
1
= u to the left. If the jet issues from the orifice with a relative velocity v
2
, taking
velocities to the right as positive, the absolute velocity of the jet will be

V
2
= v
2
u. (4.54)

Hence
( )
2 2 1 2
) ( v u u v V V V = = = . (4.55)

The same result is obtained by referring to Figure 4.9 for the case of a stationary tank (i.e. u = 0). In
this instance
2 2
0 v V V = = . Thus the force of reaction is independent of the velocity of the tank,
and this equation applies for either rest or motion.

Rocket
Both the fuel and the oxygen for combustion are contained within a rocket which is analogous to
the tank of Figure 4.9. The only difference is that the exit pressure p
0
of the gases leaving the orifice
or nozzle at section 2 may exceed the atmospheric pressure p
a
. If A
2
equals the area of the jet, the
rocket thrust is
( )
2 0
2
2 2
A p p v A F
a
+ = , (4.56)

where v
2
is the velocity at which the jet issues from the rocket. The thrust F is independent of the
speed of the rocket.

Jet engine
By jet engine is meant a device which carries only its fuel and takes in the air for combustion from
the atmosphere. It is analogous to the tank of Figure 4.10, including the intake of fluid at section 1,
except that the velocity of the air received is usually in the same straight line as the velocity of the
exit jet at section 2. There are three forms of jet engines, but the equation is the same for all three.
The ram jet must be brought up to a high speed by rockets or some other means, and then it scoops
in the air from in front and compresses it by virtue of the stagnation pressure due to its speed. The
turbojet can take off from the ground, for in it the air is compressed by a compressor driven by a gas
turbine, the exhaust from which supplies the jet propulsion. Then there is a pulsating machine,
which scoops in air in cycles. The inlet is then closed, the fuel-air mixture is exploded; a jet then
gives the device a spurt; and the process is repeated.
The thrust of a jet engine is

( )
g
u G v G G
F
a f a
+
=
2
, (4.57)

where:
G
a
= weight of air taken in per second
G
f
= weight of fuel consumed per second
v
2
= velocity of exhaust with respect to the engine
u = velocity of flight = velocity of air entry with respect to the engine

The thrust will vary with the speed of flight. Usually p
0
= p
a
, and so the second term of the equation
describing the thrust of a rocket is not included in this equation.

In addition to the cases that have already been discussed, there are numerous other fluid-
flow situations where the momentum principle is useful. It is used to develop an expression for the
head loss in an expansion, the concept of the shock wave, torque in rotating machines flow in

51
propellers and windmills, and the equations of the hydraulic jump. Another application is that of
finding the forces exerted on open-flow structures. The magnitudes of such forces may generally be
found by application of the momentum principle.

4.10. Forces on immersed bodies
A body which is wholly immersed in a fluid may be subjected to two kinds of forces arising
from the relative motion between the body and the fluid. These forces are termed the drag and the
lift, depending on whether the force is parallel to the motion or at right angles to it, respectively.
The drag forces can be viewed as having two components: a pressure drag F
p
and a friction
drag F
f
. The pressure drag is equal to the integration of the components in the direction of motion
of all the pressure forces exerted on the surface of the body. It may be expressed as the dynamic
component of the stagnation pressure acting on the projected area A of the body normal to the flow
times a coefficient c
p
which is dependent on the geometric form of the body and generally
determined by experiment. Thus
A
V
c F
p p
2
2
= (4.58)

The friction drag is equal to the integration of the components of the shear stress along the
boundary of the body in the direction of motion. For convenience, the friction drag is commonly
expressed in the same general from as (4.58). Thus

BL
V
c F
f f
2
2
= (4.59)

where c
f
is the friction-drag coefficient, dependent on viscosity, L is the length of surface parallel to
flow, and B is the transverse width, conveniently approximated for irregular shapes by dividing total
area by L.
The total drag on a body is the sum of the friction drag and the pressure drag

p f D
F F F + = (4.60)

It is customary to employ a single equation which gives the total drag

A
V
c F
D D
2
2
= (4.61)

in terms of an overall drag coefficient c
D
.
The foregoing principles are vividly illustrated in the case of the flow around a sphere. For
very low Reynolds numbers (DV/ < 1, in which D is the diameter of the sphere) the flow about the
sphere is completely viscous. Experimentally, it is found that a recirculating zone (or vortex ring)
develops close to the rear stagnation point at about Re = 30 (Figure 4.11). With further increase in
the Reynolds number this recirculating zone or wake expands. Defining locations on the surface by
the angle from the front stagnation point, the separation point moves forward from about 130
o
at Re
= 100 to about 115
o
at Re = 300. In the process the wake reaches a diameter comparable to that of
the sphere when Re 130. At this point the flow becomes unstable and the ring vortex that makes
up the wake begins to oscillate. However, it continues to be attached to the sphere until about Re =
500. At Reynolds numbers above about 500, vortices begin to be shed and then convected
downstream. Furthermore, as Re increases above 500 the flow develops a fairly steady near-wake
behind which vortex shedding forms an unsteady and increasingly turbulent far-wake. This process
continues until, at a value of Re of the order of 1000, the flow around the sphere and in the near-
wake again becomes quite steady. Transition to turbulence occurs on the free shear layer (which de

52
defines the boundary of the near-wake) and moves progressively forward as the Reynolds number
increases.
Since the Reynolds number range between 0.5 and several hundred can often pertain in
multiphase flows, one must resort to an empirical formula for the drag force in this regime. A
number of empirical results are available; for example, a widely used empirical result is

+ =
6
Re
1 6
3 / 2
VD F
D
(4.62)

which fits the data fairly well up to Re 1000. At Re = 1 the factor in the square brackets is 1.167.
On the other hand, at Re = 1000, the two factors are respectively 17.7 and 188.5. For Re < 1, the
term in the bracket can be neglected, and equating this expression to (4.51), where A is defined as
4
2
D , the frontal area of the projected sphere, gives the result that c
D
= 24/Re.

Figure 4.11. Streamlines of steady flow (from left to right)past a sphere at various Reynolds numbers


53
The lift force acts on an immersed body normal to the relative motion between the fluid and
the body. The most commonly observed example of lift is that of the airplane wing suspended in the
air by this force. The elementary explanation for such a lift force is that the air velocity over the top
of the wing is faster than the mean velocity, while that along the underside is slower than the mean.
The Bernoulli theorem then shows a lower pressure on the top and higher pressure on the bottom,
resulting in a net upward lift.
The lift force is commonly expressed in the same general from as (4.58). Thus

A
V
c F
L L
2
2
= (4.63)

where c
L
is the lift coefficient whose value depends principally on the angle of attack and the shape
of the airfoil or body, and A is the projected area of the airfoil or body normal to the lift vector.
A wealth of data on the lift and drag of various airfoils has been obtained from wind-tunnels
tests. The results of such test may be presented graphically as plots of the lift and drag coefficients
vs. the angle of attack. Since the efficiency of the airfoil is measured by the ratio of lift to drag, the
value of
D L
c c / is generally plotted also. These three curves can be combined neatly into a single
curve, suggested by Prandtl, known as a polar diagram (Figure 4.12).

Figure 4.12. An example of a polar diagram

The coordinates of the polar diagram are the lift and drag coefficients, while the angles of
attack are represented by different points along the curve. The ratio of lift to drag is the slope of the
line from the origin to the curve at any point. Evidently, the maximum value of the ratio occurs
when this line is tangent to the curve. The lift is seen to increase with the angle of attack up to the
point of stall. Beyond this point the flow along the upper surface of the foil separates an creates a
deep turbulent wake.
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 0.02 0.04 0.06 0.08 0.1 0.12
c
D
c
L
-9
o
-6
o
-3
o
0
o
3
o
6
o
9
o
12
o
15
o
18
o
of maximum lift/drag
of maximum lift
(stall point)
values of , angle of attack

54
5. POTENTIAL FLOWS

Potential flows are the flows of ideal incompressible fluids for which the local velocities can
be expressed as a function of the velocity potential

grad = V . (5.1)

Therefore, potential flows are irrotational flows because the vorticiy of the flow is zero.

5.1. Complex potential for plane flows
Plane flows differ from other two-dimensional flows because the two independent variables
x and y can be combined into one complex variable:

y x z i + = , 1 i = . (5.2)

A complex function F(z) is said to be analytic in an open set G, if it is complex
differentiable at every point z there, that is the limit

( ) ( )
dz
dF
z
z F z z F
z
=
+
0
lim (5.3)

exists and is independent of the path from z to z + z. If this requirement is not satisfied, the point is
a singular point.
First we note that along a path parallel to the x axis

x
F
dz
dF
= (5.4)

holds, and along a path parallel to the y axis

( ) y
F
dz
dF
i
= . (5.5)

Since every complex function F(z) is on the form

( ) ( ) ( ) y x y x z F , i , + = , (5.6)

we therefore have
y
F
i y y i x
i
x x
F
1 1
. (5.7)

Clearly for the derivative to exist it is necessary that

y x

=

and
x y

=

(5.8)

hold. The Cauchy-Riemman differential equations (5.8) are also sufficient for the existence of the
derivative of F(z). We can also show easily that both the real part ) , ( ) ( y x F = and the
imaginary part ) , ( ) ( y x F = satisfy Laplace equation (see sections 4.1 and 4.2). Both functions

55
can therefore serve as the velocity potential. We choose as the velocity potential and shall now
consider the physical meaning of . With

j v i u j
y
i
x
r r r r
+ =

= = grad V (5.9)

because of (5.8) we also have

j u i v j
y
i
x
r r r r
+ =

= grad . (5.10)

From 0 grad grad = we conclude that grad is perpendicular to the velocity vector V, and
therefore = constant are streamlines.
We now calculate the flow rate between the points A and B (Figure 5.1):

=
AB
n AB
ds V Q . (5.11)

We note that
( ) ( )
vdx udy ds V ds V
ds V ds V ds V
n
= =
= =

cos sin sin cos
cos sin cos sin sin
,

and therefore
=
AB
AB
vdx udy Q . (5.12)

If we write
y
u

= and
x
v

= , we get

A B
AB AB
AB
d dy
y
dx
x
Q = =

=

. (5.13)

We can also calculate the circulation along AB:

=
AB
t AB
ds V . (5.14)

We note that
Figure 5.1. Velocity potential and stream function

56
( ) ( )
vdy udx ds V ds V
ds V ds V ds V
t
+ = + =
+ = =

sin sin cos cos
sin sin cos cos cos
,

and therefore
+ =
AB
AB
vdy udx . (5.15)

If we write
x
u

= and
y
v

= , we get

A B
AB AB
AB
d dy
y
dx
x
= =

=

. (5.16)

These results confirm that the flow rate and circulation are independent of the path between A and
B. The velocity components can be most easily calculated using

v u
x x x
F
dz
dF
i i =
= , (5.17)

where the sign of v is to be noted: dF/dz furnishes the complex conjugate velocity

v u
dz
dF
i = = , (5.18)

that is, the reflection of the complex conjugate velocity v u i + = at the real axis.
We shall now look at some examples of complex potentials:

i) Translational flow

( ) ( )z V U z F

= i , (5.19)
or
( ) ( ) x V y U y V x U F

+ + = i , (5.20)
and thus we get
y V x U

+ = , (5.21)

x V y U

= . (5.22)

The streamlines follow from = constant to C U xV y + =

and the complex conjugate is

= V U
dz
dF
i . (5.23)

ii) Source flow

( ) z
E
z F ln
2
= (5.24)

or because
i
r z e =

57
( ) ( )
i ln
2
+ = r
E
z F . (5.25)

We obtain the velocity potential and the stream function as

r
E
ln
2
= , (5.26)

2
E
= . (5.27)

The streamlines = constant are straight lines through the origin.

iii) Potential vortex

( ) z z F ln
2
i
= , (5.28)

where the negative sign is needed because we count anticlockwise positive. In polar coordinates
we obtain
( ) ( )
i ln
2
i +
= r z F , (5.29)
therefore
= , (5.26)

r ln
2
= . (5.27)

The streamlines = constant are circles (r = constant).

iv) Dipole

( )
z
M
z F
1
2
= , (5.28)
or
( ) ( ) y x
r
M
r
M
F i
1
2
sin i cos
1
2
2
= =
, (5.29)

from which we read off directly

2
2
cos
2 r
x M
r
M
= = (5.30)
and
2
2
sin
2 r
y M
r
M
= = . (5.31)

From = constant we obtain with r y = sin


58
y
C
M
y x r = + =
2 2 2
, (5.32)

that is, the streamlines are circles which are tangent to the x axis in the origin (Figure 5.2).

Figure 5.2. Streamlines and equipotential lines of the plane dipole

v) Corner flow

n
z
n
a
z F = ) ( , (5.33)

with
i
r z e = it follows:

( ) n n r
n
a
F
n
sin i cos + = , (5.34)
and therefore
n r
n
a
n
cos = , (5.35)
and
n r
n
a
n
sin = . (5.36)

For the magnitude of the velocity we obtain

1 1
= = =
n n
r a az
dz
dF
u
r
. (5.37)

The streamlines are generally found from = constant. In particular for = 0, therefore 0 sin = n
or ,...) 2 , 1 , 0 ( = = k n k , these are straight lines through the origin which can represent walls in
the flow. Figure 5.3 shows the streamline plot for different values of the exponent n.


59

Figure 5.3. Corner flow for different values of the exponent n

vi) Flow past a circular cylinder (Figure 5.4)

|
|
.
|
\
|
+ =

z
r
z U z F
2
0
) ( (5.38)
or again with
i
r z e =
/n

60
sin i cos
2
0
2
0
|
|
.
|
\
|
+
|
|
.
|
\
|
+ =

r
r
r U
r
r
r U F , (5.39)

and therefore
cos
2
0
|
|
.
|
\
|
+ =

r
r
r U , (5.40)
and
sin
2
0
|
|
.
|
\
|
=

r
r
r U . (5.41)

We obtain 0 = for r = r
0
and = 0, . From the complex conjugate velocity

|
|
.
|
\
|
=
2
2
0
1
z
r
U
dz
dF
(5.42)

by dF/dz = 0 we find the location of the stagnation points at z = r
0
and deduce the velocity
components

|
|
.
|
\
|
=

2
2
0 2 i
e 1 i
r
r
U v u

(5.43)
or
|
|
.
|
\
|
=

2 cos 1
2
2
0
r
r
U u , (5.44)
and
2 sin
2
2
0
r
r
U v

= . (5.45)

The maximum velocity is reached for r = r
0
, i.e. on the body at = /2, 3/2:

= U U 2
max
. (5.46)

Figure 5.4. Flow past a circular cylinder without circulation


61
vii) Flow past a cylinder with potential vortex
This superposition is possible since a potential vortex at the axis satisfies the kinematic boundary
condition at the circular cylinder. The complex potential of this flow is

0
2
0
ln
2
i ) (
r
z
z
r
z U z F
|
|
.
|
\
|
+ =

(5.47)
from which we read off the velocity potential and the stream function

2
cos
2
0

+
|
|
.
|
\
|
+ =

r
r
r U , (5.48)
and
0
2
0
ln
2
sin
r
r
r
r
r U
|
|
.
|
\
|
=

. (5.49)

Figure 5.5. Flow past a circular cylinder with clockwise circulation k U r

=
0
4 .
a) k = 0; b) 0 > k < 1; c) k = 1; d) k > 1 (no stagnation point on the body)

Since F(z) represents the flow past a circular cylinder for all values of , it is not unique. We obtain
a survey of the different flows, by computing the stagnation points on the body contour. From


62
0
1
2
sin 2
1
0
r
U
r
u
r r

+ =

=

=
(5.50)

the equation for the stagnation points follows

0
1
4
sin
r U
. (5.51)

Figure 5.5 shows the flow forms for different values of the circulation .
The force on the cylinder in the x direction vanishes for symmetry reasons, and that in the y
direction is
=

U F
y
, 0 < . (5.52)

The flow field in Figure 5.5d can be experimentally realized if a rotating cylinder is exposed to a
cross-flow with undisturbed U
sufficiently small compared to the circumferential velocity r

0
,
corresponding to the condition
> U r
0
4 . As experiments show, the lift calculated from potential
theory is already reached at 4 /
0
>

U r . We call the phenomenon where rotating cylinder in a
cross-flow experiences a lift, the Magnus effect.

5.2. Blasius theorem
We shall restrict ourselves to steady flows and shall consider a simply connected domain,
say the cross-section of a cylinder in a flow (Figure 5.6). From

=
) (C
i i
ds pn F (5.53)

we compute the components of the force per unit depth

=
) (C
x
pdy F (5.54)
and
+ =
) (C
y
pdx F . (5.55)

In complex notation
y x z i + = , y x z i =

we combine the force components as

=
) (
) i ( i
C
y x
z d p F F . (5.56)

The moment on the cylinder about the origin only has a component in the z direction

+ =
) (
) (
C
ypdy xpdx M . (5.57)

We write the line integral using complex notation as

63

( )
=
) (C
z d z p M . (5.58)

In steady flow, from Bernoulli equation

( )
0
2 2
2
p v u p = + +

, (5.59)

and from the square of the magnitude of the complex conjugate velocity

2 2
2
v u
z d
F d
dz
dF
dz
dF
+ = = (5.60)

it follows that the pressure is
z d
F d
dz
dF
p p
2
0

= . (5.61)

Figure 5.6. Blasius theorem

Therefore, we write for the force

=
) (
2
i i
C
y x
F d
dz
dF
F F

, (5.62)

because the closed integral over the constant pressure p
0
vanishes. Since the contour of the body is a
curve = const, we have

dF d F d = = , (5.63)

and from (5.62) emerges the first Blasius theorem:

|
.
|
\
|
=
) (
2
2
i i
C
y x
z d
dz
dF
F F

. (5.64)

64

In an analogous manner we obtain from (5.58) the second Blasius theorem:

(
(
|
.
|
\
|
=

) (
2
2
C
dz z
dz
dF
M

. (5.65)

According to the derivation the integration is to be carried out alng the contour of the body. As a
consequence of Cauchy theorem

{
=
) (
by enclosed domain in the and on c holomorphi is if 0 ) (
C
C C f(z) dz z f (5.66)

the integration can also be carried out along any arbitrary closed curve enclosing the body, as long
as there are no singularities between the contour of the body and the integration curve. Using the
sense of integration in Figure 5.7 it follows from (5.66) that

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

C C
dz z f dz z f (5.67)

Figure 5.7. Application of Cauchy theorem

or, if C
1
and C
2
are followed in the same sense,

=
) ( ) (
2 1
) ( ) (
C C
dz z f dz z f . (5.68)

5.3. Kutta-J oukowski theorem
Using the first Blasius theorem we can calculate the force on a cylinder of arbitrary contour
in steady flow. Let the stream velocity at infinity be U
+ iV
, and let there be no singularities

outside the body, although there will be inside, in order to represent the body and to produce the lift.
Outside the singularities we can represent the velocity field by a Laurent series of the form

= =
0
i
n
n
n
z A v u
dz
dF
, (5.69)

which yields the complex potential


65
( )
( )
=

+
+ =
2
1
1 0
const
1
1
ln
n
n
n
z A
n
z A z A z F . (5.70)

From the condition at infinity

= V U
dz
dF
i (5.71)
it follows that

= V U A i
0
. (5.72)

To calculate the coefficient A
1
we form the integral of (u iv) around the contour of the body

( ) ( )

+ =
(C) (C)
i ) i ( i dy dx v u dz v u (5.73)
or
( )

+ =
(C) (C) ) (
i i
C
d d dz v u x u , (5.74)

where the second integral vanishes since d is zero along the contour of the body. With the
definition of the circulation, we can write

( )
=
(C)
i dz v u . (5.75)

Since the Laurent series has only one essential singularity (z = 0), then from the residue theorem we
have
( ) = =
(C)
1
i 2 i A dz v u . (5.76)

From this we obtain the complex conjugate velocity in the form

=
2
2
i i i
n
n
n
z A z V U v u
. (5.77)

We can now calculate the force on the cylinder using Blasius theorem. Because of

( ) ( ) ( ) ....... i
2
2
i i i
2
2
2
2
2
+ +
|
|
.
|
\
|

=
|
.
|
\
|

V U
z
A
z
V U
z
V U
dz
dF

(5.78)

and by applying the residue theorem we first obtain

( )
( )

=
|
.
|
\
|
V U
dz
dz
dF
c
i
i i 2
) (
2
(5.79)

and then from (5.64) the Kutta-Joukowski theorem

( )

= V U F F
y x
i i i . (5.80)


66
From this equation we firstly conclude that the lift is perpendicular to the undisturbed stream at
infinity, that is, the body experiences no drag, and secondly for a given circulation the lift is
independent of the contour of the body.
In a similar manner we obtain for the moment

(
|
|
.
|
\
|
=
U
V
A U M i 1 i 2
2
. (5.81)

It can be seen that the moment depends on the complex coefficient A
2
and therefore on the contour
of the body.

67
6. SIMILITUDE AND DIMENSIONAL ANALYSIS

It is usually impossible to determine all the essential facts for a given fluid flow by pure
theory, and hence dependence must often be placed upon experimental investigations. The number
of tests to be made can be greatly reduced by a systematic program based on dimensional analysis
and specifically on the laws of similitude or similarity, which permit the application of certain
relations by which test data can be applied to other cases.
Thus the similarity laws enable us to make experiments with a convenient fluid such as
water or air, for example, and then apply the results to a fluid which is less convenient to work with,
such as gas, steam or oil. Also, valuable results can be obtained as a minimum cost by tests made
with small-scale models of the full-sized apparatus. The laws of similitude make it possible to
determine the performance of the prototype, which means the full-size device, from tests made with
the model. It is not necessary that the same fluid be used for the model and its prototype. Neither the
model necessary smaller than its prototype.

6.1. Geometric similarity
One of the desirable features in model studies is that there be geometric similarity, which
means that the model and its prototype be identical in shape but differ only in size. The important
consideration is that the flow patterns be geometrically similar. If the scale ratio is denoted by L
r
,
which means the ratio of the linear dimensions of the prototype to corresponding dimensions in the
model, it follows that areas vary as
2
r
L and volumes as
3
r
L . Complete geometric similarity is not
always easy to attain. Thus the surface roughness of a small model my not be reduced in proportion
unless it is possible to make its surface very much smoother than that of the prototype.

6.2. Kinematic similarity
Kinematic similarity implies geometric similarity and in addition it implies that the ratio of
the velocities at all corresponding points in the flow is the same. If subscripts p and m denote
prototype and model, respectively, the velocity ratio v, is

m
p
r
v
v
v = (6.1)

and its value in terms of L
r
will be determined by dynamic considerations.
As time T is dimensionally L / v, the time scale is

r
r
r
v
L
T = (6.2)

and in a similar manner the acceleration scale is

r
r
r
r
r
L
v
T
L
a
2
2
= = (6.3)

6.3. Dynamic similarity
If two systems are dynamically similar, corresponding forces must be in the same ratio in
the two. Forces that may act on a fluid element include those due to gravity F
G
, pressure F
P
,
viscosity F
V
, and elasticity F
E
. Also, if the element of fluid is at a liquid-gas interface, there are
forces due to surface tension F
T
. If the summation of forces on a fluid element does not add up to
zero, the element will accelerate in accordance to Newtons second law. Such an unbalanced force

68
system can be transformed into a balanced system by adding an inertia force F
I
that is equal and
opposite to the resultant R of the acting forces. Thus generally,

R F F F F F F
T E V P G
= + + + + =
(6.4)

and
R F
I
= (6.5)

Thus
0 = + + + + +
I T E V P G
F F F F F F (6.6)

These forces may be expressed in simplest terms as follows:

Gravity
g L mg F
G
3
= =

Pressure
( ) ( )
2
L p A p F
P
= =

Viscosity
L v L
L
v
A
dy
du
F
V
=
=
2

Elasticity
2
L E A E F
v v E
= =

Surface tension L F
T
=

Inertia
2 2
2
4
2
3
L v
T
L
T
L
L ma F
I
= = = =

In many flow problems some of these forces are either not present or insignificant. Consider,
for example, the case where the forces acting on a fluid system are F
G
, F
P
, F
V
, and F
I
. Then
dynamic similarity will be achieved if

Im
F
F
F
F
F
F
F
F
Ip
Vm
Vp
Pm
Pp
Gm
Gp
= = = (6.7)

where subscripts p and m refer to prototype and model, respectively. These relations can be
expressed as

m
G
I
p
G
I
F
F
F
F
,
m
P
I
p
P
I
F
F
F
F
,
m
V
I
p
V
I
F
F
F
F
(6.8)

Each quantity is dimensionless and the significance of the dimensionless ratios is discussed below.

Reynolds number
In the flow of a fluid through a completely filled pipe, gravity does not affect the flow pattern. It is
also obvious that capillarity is of no practical importance, and hence the significant forces are
inertia and fluid friction due to viscosity.
Considering the ratio of inertia forces to viscous forces, the parameter obtained is called the
Reynolds number, or Re. The ratio of the two forces is

69

Lv Lv
Lv
v L
F
F
V
I
= = = =
2 2
Re (6.9)

For any consistent system of units, Re is a dimensionless number. The linear dimension L may be
any length that is significant in the flow pattern. Thus, for a pipe completely filled, it might be
either the diameter or the radius, and the numerical value of Re will vary accordingly. General
usage prescribes L as the pipe diameter.
Figure 6.1 depicts two simple illustrations. The first involve a fluid that flows with velocity
U
0
through a channel of width w that makes a sudden right turn. During the turn time
0 0
~ U w , a
fluid element rounding the corner loses momentum density
0
U by exerting an inertial centrifugal
density w U U f
i
2
0 0 0
~ = . The other involves a fluid flowing in a channel that contracts over a
length l. By mass conservation, the velocity increases as ( ) l z U u / 1 ~
0
+ , causing a fluid element to
gain momentum at a rate

l
U
dz
du
U
dt
du
f
i
2
0
0
~ ~

= . (6.10)

The force exerted on the rest of the fluid is equal and opposite to the force required to accelerate
each fluid element. Thus the force on the fluid due to a curved streamline points outwards
centrifugally, and the inertial force in an expansion or contraction points towards the wide end of
the channel, regardless of the flow direction. In both cases, the magnitude of inertial and viscous
force densities must be compared. Viscous force densities results from gradients in viscous stress,
and thus scale as
2
0 0
~ L U f
v
, where L
0
is a typical length scale. The ratio of these two force
densities is just the Reynolds number

Re
0 0
= =
L U
f
f
v
i
(6.11)

Figure 6.1. Inertial forces exerted by accelerating fluid elements. (a) A small fluid element roundng a
corner. (b) A fluid element flowing through a contraction.

Froude number
Considering inertia and gravity forces alone, a ratio is obtained called a Froude number, or Fr.
The ratio of inertia forces to gravity forces is


70
gL
v
gL
v L
2
3
2 2
=
(6.12)

Although this is sometimes defined as the Froude number, it is more common to use the square root
so as to have v in the first power, as in the Reynolds number. Thus a Froude number is

gL
v
= Fr (6.13)

System involving gravity and inertia forces are the flow of water in open channels, the flow
over a spillway, or the flow of a stream from an orifice.
For the computation of Fr, the length L must be some linear dimension that is significant in
the flow pattern. For an open channel, for example, it is taken as the depth of flow.
The velocity ratio is

1
r
m
p
r
L
v
v
v = = (for same Fr) (6.14)

while the ratio of time for prototype to model is

1
r
m
p
r
L
T
T
T = = (for same Fr) (6.15)

and a
r
= 1.
Since the velocity varies as and the cross section area as , it follows that

1
2 / 5
r
m
p
r
L
Q
Q
Q = = (for same Fr) (6.16)

Mach number
Where compressibility is important, it is necessary to consider the ratio of the fluid velocity to that
of a sound wave in the same medium. This ratio, called the Mach number, or Ma, is

c
v
= Ma (6.17)

where c is the acoustic velocity in the medium in question. If Ma is less than 1, the flow is called
subsonic; if it is equal to 1, the flow is sonic; if it is greater than 1, the flow is called supersonic;
and for extremely high values of Ma the flow is called hypersonic.
The ratio of inertia forces to elastic forces

v v
E
v
L E
L v
2
2
2 2

= (6.18)

is called the Cauchy number. The celerity c of an acoustic wave is given by


71
v
E
c = (6.19)

and, therefore, the Cauchy number is the square of the Mach number

2
Ma Ca = (6.20)

Weber number
In a few cases of flow, surface tension may be important, but normally it is negligible. The ratio of
inertia forces to surface tension forces is ( ) L L v
2 2
, the square root of which is known as the
Weber number:
L
v
= We (6.21)

Euler number
A dimensionless quantity related to the ratio of the inertia forces to the pressure forces is known as
the Euler number. It is expressed as

p g
v
p
v
=
2 2
Eu (6.22)

If only pressure and inertia influence the flow, the Euler number will remain constant.

Weissenberg and Deborah numbers
Thus far, we have considered the influence of inertia, pressure, viscosity, surface tension,
compressibility and gravity on fluid flows. Dissolved polymers, for example, add an elastic
component to the fluid that further enriches flow behavior. We provide here a simple picture of
polymer dynamics to illustrate the basic physics at hand. The simplest model system treats a
polymer as a dumbbell with two beads each of hydrodynamic resistance connected by an entropic
spring of stiffness k
H
. The spring constant k
H
can be obtained for a freely jointed chain with N steps
of length b as k
H
= 3k
B
T/Nb
2
. For simplicity, we confine the polymer to the one-dimensional line
along the center of a contracting channel, as in Figure 6.2. Due to the contraction, the local
(extensional) flow felt by the polymer increases as it moves down the channel, roughly like
z e u u
z
& + =
0
. The bead separation R
p
changes due to three physical effects: (i) Brownian motion
drives beads apart with average speed / ~ ~ T k D R R
B p p
&
[the diffusivity ) 6 /( ~ a T k D
B
for
spherical solute molecules of size a], (ii) the spring pulls the beads together with velocity
/ ~
p H p
R k R
&
, and (iii) the extensional flow advects the forward bead more quickly than the rear
bead, driving them apart with velocity
p p
R e R &
&
~ . Each effect has its own characteristic time scale:
D R
D
/
2
0
= for the beads to diffusively explore a length scale R
0
,
H p
k / = for the polymer spring
to relax, and
1
= e
e
& is a time scale associated with extensional flow.
While this example helps in developing physical insight and intuition, its applicability is
generally limited. It assumes polymers do not interact, which requires the solution to be dilute. The
solution stress is assumed to be dominant by the viscous solvent rather than the polymers,
constituting a so-called Boger fluid. Polymer deformations are also assumed to be small since we

72
have ignored the nonlinear effects for large deformations. Despite these simplifications, several
important phenomena are illustrated by the above simple example.

Figure 6.2. The effect of a non-uniform flow on a model polymer, represented as two beads separated by a
distance R on a spring. (a) The center of mass of the polymer moves with velocity U
CM
, whereas the two
beads experience relative motion / ~ ~
p H p R
R k R e U & due to external flow gradients and spring forces.
Brownian forces balance spring forces to give a steady-state polymer size. When Wi O(1), the extensional
flow overwhelms the spring, and the polymer unravels via a coil-stretch transition. (b) As the beads are
pulled toward the spring, each exerts a force back on the fluid, resulting in a force-dipole flow.

Weissenberg number
In equilibrium 0 = e& , spring forces balance Brownian forces to give a characteristic polymer
size ( ) ( )
2 / 1
2 2 / 1
0
~ ~ Nb k T k R
H B
. By contrast, an extensional flow acts to drive the beads apart (like
a negative spring), and alters the steady polymer size via

( )
2 / 1
0
2 / 1
Wi 1
~ ~ ) (
R
e k
T k
e R
H
B
p
&
& . (6.23)

Here we have introduced the Weissenberg number

& &
p p
e or Wi = (6.24)

which relates the polymer relaxation time to the flow deformation time, either inverse extension rate
e& or shear rate & . When Wi is small, the polymer relaxes before the flow deforms it significantly,
and perturbations to equilibrium are small. As Wi approaches 1, the polymer does not have time to
relax and is deformed significantly.

Deborah number
Another relevant time scale
flow
characteristic of the flow geometry may also exist. For example, a
channel that contracts over a length L
0
introduces a geometric time scale
0 0
/U L
flow
= required for
a polymer to transverse it. The flow time scale
flow
can be long or short compared with the
polymer relaxation time
p
, resulting in a dimensionless ratio known as the Deborah number

flow
p
= De . (6.25)

Note that the usage of De and Wi can vary. Some references use Wi exclusively to describe
shear flows and use De for the general case, whereas others use Wi for local flow time scales due to
a local shear and De for global flow time scales due to a residence time in flow.

73

Elasticity number
As the flow velocity U
0
increases, elastic effects become stronger and De and Wi increase.
However, the Reynolds number Re increases in the same way, so that inertial effects become more
important as well. The elasticity number

El = De / Re =
2
h
p

(6.26)

where h is the shortest dimension setting shear rate, expresses the relative importance of elastic to
inertial effects. Significantly, El depends only on the geometry and material properties of the fluid,
and is independent of flow rate.

Knudsen number
We have implicitly focused on liquids thus far, and have not explicitly discussed gas flows. The
molecular-level distinction between liquids and gases can have important ramifications for fluid
flows. While liquid molecules are in constant collision, gas molecules move ballistically and collide
only rarely. Using the kinetic theory, one can calculate the mean free path between collisions to be

2
1
~
na
f
, (6.27)

where n is the number density of molecules with radius a. For example, an ideal gas at 1 bar and
25
o
C has a mean free path 70 ~
f
nm that increases at lower pressures or higher temperatures.
As the flow geometry get smaller, the mean free path occupies an increasingly significant
portion of the flow, and thus plays an increasingly important role. The Knudsen number

L
f
= Kn (6.28)

expresses the ratio of the mean free path (the length scale on which molecules matter) to a
macroscopic length scale L. The latter is typically a length scale representative of the device, but
could also be given by the length scale for temperature, pressure, or density gradients.
Noncontinuum effects play an increasing role as Kn increases. Roughly speaking, molecules
located farther than
f
from a solid wall do not see the wall, whereas closer molecules can collide
with the wall rather than other molecules. This implies that the fluid behaves like a continuum up to
a distance
f
from the wall, and influences the boundary conditions obeye by the fluid. Maxwell
was the first who predicted the no-splip boundary condition, yielding instead the slip condition

0
0
dn
du
u = (6.29)

where is a slip length of order
f
.
On the other hand, the density of a gas typically depends much more strongly on
temperature and pressure than that of a liquid. Therefore, compressibility can play a much more
important role in gas flows, particularly when significant differences in T or p exists. Both fluid
density and mean free path
f
are affected. Three distinct Kn regimes have been measured in
pressure-driven flows: Kn << 1, where the gas behaves as a no-slip fluid, Kn ~ 1, where the gas

74
behaves as a continuum but slips at the boundaries, and Kn >> 1, where the continuum
approximation breaks down completely.
Liquid molecules are in constant contact, resulting in significantly higher incompressibility
and making the concept of
f
less meaningful. As such, noncontinuum effects appear to play a role
only when the fluids themselves are confined to molecular length scales.
A fundamental question that arises is whether fluid truly slips over a solid surface,
invalidating the no-slip boundary condition, or whether the experiments reflect an apparent slip that
arises from surface inhomogeneities or a complex interface with additional physics. Recent
molecular dynamics simulations, whose physical ingredients are known, have indeed shown fluid to
slip (albeit with much shorter slip length), particularly when the attraction between molecules is
stronger than the wall/molecule attraction. Various explanations have been proposed to account for
apparent slip effects. A perfectly slipping but rough surface obeys a macroscopic boundary
condition with slip length of order the roughness length scale. From this standpoint, any
macroscopic length (real or apparent) will be confined to the roughness scale. However, even
nonwetting, molecularly smooth surfaces have exhibited large apparent slip lengths. A gas layer at
the interface, which would lubricate and alter the fluid flow, has been invoked as an explanation.
The idea is that dissolved gas molecules are drawn out of solution to nucleate a gas layer at the
solid-liquid interface and lower the surface energy.
The no-slip boundary condition never enjoyed solid theoretical background, but rather was
accepted due to its apparent experimental success. An understanding of the apparent slip might
allow surfaces to be designed specifically to slip, e.g. reduce hydrodynamic resistance. In this case,
an analogous Knudsen number

L
= Kn (6.30)

will be significant in the description of such flows.

Bond number and capillary number
In some cases, surface tension and gravity may be important. The ratio of gravity forces to surface
tension forces is ( ) L L v
2 2
, which is known as the Bond number:

g L
2
Bo = (6.31)

Consider now a long droplet of length L in a channel of radius w with an inhomogeneous surface:
hydrophobic (with
L
sl
) for z < 0 and hydrophilic (with
L
sl
R
sl
> ) for z > 0. Energetically, the
droplet wants to move onto the hydrophilic surface, and moving with velocity U decreases the
stored interfacial energy at a rate ( )wU ~ , where
R
sl
L
sl
= . This energy is lost to viscous
dissipation, which consumes a power ( )
L U dV z u
2 2
~ when dissipation is dominated by
viscous shear in the bulk. Assuming the capillary energy released to be balanced by viscous
dissipation gives

L
w U
= =
Cp
(6.32)

The capillary number arises naturally, because capillary stresses are balanced by viscous stresses. In
this case, the droplet moves with velocity scale L w U ~ .


75

Figure 6.3. Droplet motion due to a gradient in solid-liquid interfacial energy

I llustrative example. A submerged body is to move horizontally through a liquid with = 8340
N/m
3
and = 1.510
-3
Ns/m
2
at a velocity of 2.5 m/s. To study the characteristics of this motion, an
enlarged model is tested in a liquid with = 9810 N/m
3
and = 10
-3
Ns/m
2
. The model ratio is 8:1.
Determine the velocity at which this enlarged model should be pulled through the water to achieve
dynamic similarity. If the drag force (F L
2
v
2
) on the model is 10000 N, predict the drag force on
the prototype. Reynolds criterion must be satisfied.
The parameters of the prototype are:

p
= 8340 N/m
3
,
p
= 1.510
-3
Ns/m
2
, v
p
= 2.5 m/s .

The known parameters of the model are:
m
= 9810 N/m
3
,
m
= 10
-3
Ns/m
2
, F
m
= 10000 N, and
8
1
=
m
p
L
L
.
m
m
m m
p
p
p p
L v
L v
= m/s 417 . 1 5 . 2 8
10
9810
8340
10 5 . 1
1
3
3
1
=
p
p
m
m
m
p
p
m
v
L
L
v

0043 . 0
5 . 2 8 9810
417 . 1 8340
2 2
2
2 2
2 2
=

=

=
m m m
p p p
m
p
v L
v L
F
F
F
p
= 0.0043 F
m
= 42.67 N

6.4. Comments on models
In the use of models it is essential that the fluid velocity should not be used so low that
laminar flow exists when the flow in the prototype is turbulent. Also, conditions in the model
should not be such that surface tension is important if such conditions do not exist in the prototype.
When modeling a subsonic flow of a body in a wind tunnel, it is commonly necessary to
conduct the test under high pressure in order to satisfy the Reynolds criterion

p m
Dv Dv
(6.33)

without introducing compressibility effects. For example, suppose L
r
= D
p
/ D
m
= 20. If the
viscosity and the density of the air were the same in the model and prototype, then to satisfy
Reynoldss criterion, v
m
= 20 v
p
. For a body operating at low speed this would make the model
Mach number much greater than one, and compressibility effects would invalidate the behaviour of
the model. If, however, the test were conducted under a pressure of 20 bar with identical model and
prototype temperatures
p m
= 20 and
p m
since the viscosity of air changes very little with
pressure. In this case the model should be operated at a velocity equal to that of the prototype in
order for the Reynolds number to be the same.

76

6.5. Dimensional analysis
Fluid mechanics problems may be approached by dimensional analysis, a mathematical
technique making use of the study of dimensions. In dimensional analysis, from a general
understanding of fluid phenomena, one first predicts the physical parameters that will influence the
flow, and then, by grouping these parameters in dimensionless combinations, a better understanding
of the flow phenomena is made possible.
To illustrate the steps in a dimensional-analysis problem, let us consider the drag force F
D

exerted on a sphere as it moves through a viscous liquid. We must visualize the physical problem to
consider what factors influence the drag force. Certainly, the size of the sphere must enter the
problem; also, the velocity of the sphere must be important. The fluid properties involved are
density and the viscosity . Thus we can write

( ) , , , v D f F
D
= (6.34)

Here D, the sphere diameter, is used to represent sphere size.
We want to determine how these variables are interrelated. Our approach is to satisfy
dimensional homogeneity. That is, we want the dimensions on one side of the equation to
correspond to those on the other. The preceding expression may be written as a power equation

d c b a
D
v CD F = (6.35)

where C is a dimensionless constant. Substituting the proper dimensions we get

d c b
a
LT
M
L
M
T
L
L
T
ML
=
3 2
(6.36)

To satisfy dimensional homogeneity the exponents of each dimension must be identical on both
sides of the equation. Thus

For M: 1 = c + d
For L: 1 = a + b 3c d
For T: 2 = b d

Since we have three equations with four unknowns, we must express three of the unknowns in
terms of the fourth. Solving for a, b, and c in terms of d, we get

a = 2 d , b = 2 d , c = 1 d

Thus
( ) ( )
d
D
vD
v D C F

2 2
(6.37)

It may be seen that the quantity vD is a Reynolds number. Thus the original power equation
can be expressed as
( )
2 2
Re v D f F
D
= (6.38)

The results indicates that the drag on a sphere is equal to some coefficient times
2 2
v D , where the
coefficient is a function of the Reynolds number.

77
The foregoing approach to dimensional analysis is commonly referred to as the Rayleigh
method. Another more generalized approach is through use of the Buckingham theorem. This
theorem states that if there are n dimensional variables in a dimensionally homogeneous equation,
described by m fundamental dimensions, they may be grouped in n-m dimensionless groups. Thus,
in the preceding example, n = 5 and m = 3 (M, L, and T) and n-m = 2; these dimensionless groups
were Re and ( )
2 2
/ v D F
D
.
Applying the theorem to the preceding example, one would proceed as follows:

( ) 0 , , , , = v D F f
D
(6.39)

where n = 5, m = 3, so n-m = 2. Thus we can write:

( ) 0 ,
2 1
= (6.40)

The problem now is to find the s by arranging the five parameters into two dimensionless groups.
Taking , D, and v as the primary variables, the terms are:

1 1 1 1
1
d c b a
v D = (6.41)

2 2 2 2
2
d
D
c b a
F v D = (6.42)

The values of the exponents are determined as before, noting that since the s are dimensionless,
they can be replaced with M
0
L
0
T
0
. Working with
1
,

1 1
1
1
3
0 0 0
d c
b
a
LT
M
T
L
L
L
M
T L M

= (6.43)

thus
for M: 0 = a
1
+ d
1

for L: 0 = 3a
1
+ b
1
+ c
1
d
1

for T: 0 = c
1
d
1

Solving for a
1
, b
1
, and c
1
in terms of d
1
,

a
1
= d
1
, b
1
= d
1
, c
1
= d
1

Thus,
Re
1
1 1 1 1
1
=
= =

d
d d d d
Dv
v D
(6.44)

Working in a similar fashion with
2
, one gets

2 2 2
v D
F
D
= (6.45)

Finally, ( ) 0 ,
2 1
= may be expressed as

( )
2 1
= or ( )
1 2
= (6.46)

78
So
( ) Re
2 2

=
v D
F
D
(6.47)
and
( )
2 2
Re v D F
D
= (6.48)

It should be emphasized that dimensional analysis does not provide a complete solution to
fluid problems. It provides a partial solution only. The success of dimensional analysis depends
entirely on the ability of the individual using it to define the parameters that are applicable. If one
omits an important variable, the results are incomplete and this may lead to incorrect conclusions.
On the other hand, if one includes a variable that is totally unrelated to the problem, an additional
insignificant dimensionless group will result.

I llustrative example. Derive an expression for the flow rate q over the spillway shown in the
accompanying figure. Assume that the sheet of water is relatively thick so that surface-tension
effects may be neglected. Assume also that gravity effects predominate so strongly over viscosity
that viscosity may be neglected.

Figure 6.4. Flow over a spillway

Under the assumed conditions the variables that effect q would be the head H, the
acceleration of gravity g, and possibly the spillway height P. Thus

( ) P g H f q , , =
or
( ) 0 , , , = P g H q f

In this case n = 4, and m = 2. Hence, according to the theorem, there are n-m = 2 dimensionless
groups, and
( ) 0 ,
2 1
=

Using q and H as the basic variables,
1 1 1
1
c b a
g H q =

2 2 2
2
c b a
g H q =

Working with
1

1
1
1
2
3
0 0
c
b
a
T
L
L
TL
L
T L

=

79

for L: 0 = 2a
1
+ b
1
+ c
1

for T: 0 = a
1
2c
1

Hence
1 1
2
1
a c = ,
1 1
2
3
a b =

1
1 1 1
2 / 3 2 / 1
2 / 1 2 / 3
1
a
a a a
H g
q
g H q

= =

Working with
2

2 2
2
3
0 0 c b
a
L L
TL
L
T L

=

for L: 0 = 2a
2
+ b
2
+ c
2

for T: 0 = a
2

Hence
0
2
= a ,
2 2
b c =

2
2 2
0
2
b
b b
P
H
P H q

= =

Finally, ( ) 0 ,
2 1
= can be written as
( )
2 1
=

=
P
H
H g
q
2 / 3

or
2 / 3
H g
P
H
q

=

Thus dimensional analysis indicates that the flow rate per unit length of spillway is proportional to
g and to
2 / 3
H . The flow rate is also affected by the H/P ratio.

80
7. EQUATIONS OF MOTION FOR PARTICULAR FLUIDS

We shall now specialize the universally equations, namely the Newtons second law and the
energy equation, to the most technically important cases: Newtonian fluids and inviscid (ideal)
fluids.

7.1. Laminar flow of Newtonian fluids
Consider the differential element of a moving fluid shown in Figure 7.1. The forces acting
on the differential element are the mass forces dF
m
and the surface forces dF
s
. If we denote by f
m

the mass force per unit mass then dF
m
= f
m
dm = f
m
dxdydz with the components:

dxdydz X dF
mx
= , dxdydz Y dF
my
= , dxdydz Z dF
mz
= (7.1)

Figure 7.1. A differential element of a moving fluid

The surface forces that are acting on each face of the element have three components, one normal to
the surface and two along the surface. Table 7.1 gives the stresses acting on each face of the
element.

Table 7.1. Stresses acting on each face of the elemental element of a moving fluid
Normal to Ox axis Normal to Oy axis Normal to Oz axis
Surface ABCD EFGH ABFE DCGH ADHE BCGF
Area dy dz dy dz dz dx dz dx dx dy dx dy
Along Ox axis
xx

dx
x
xx
xx
+

yx

dy
y
yx
yx

zx

dz
z
zx
zx
+

Along Oy axis
xy

dx
x
xy
xy
+

yy

dy
y
yy
yy

zy

dz
z
zy
zy
+

Along Oz axis
xz

dx
x
xz
xz
+

yz

dy
y
yz
yz

zz

dz
z
zz
zz
+

The surface force in the x direction is

dxdy dz
z
dzdx dy
y
dydz dx
x
dF
zx
yx
xx
sx

(7.2)

If we apply now the Newtons second law, dma = dF
m
+ dF
s
, to the elemental fluid element we
find in the x direction

81

dxdydz
z y x
dxdydz X dxdydz
dt
du
zx
yx
xx
|
|
.
|
\
|
+ =

(7.3)

Taking into account that

z
u
w
y
u
v
x
u
u
t
u
dt
du
= (7.4)

we finally obtain in the x, y, and z directions:

|
|
.
|
\
|
+ =
z y x
X
z
u
w
y
u
v
x
u
u
t
u
zx
yx
xx

1
(7.5)

|
|
.
|
\
|
+ =
z y x
Y
z
v
w
y
v
v
x
v
u
t
v
zy yy xy

1
(7.6)

|
|
.
|
\
|
+ =
z y x
Z
z
w
w
y
w
v
x
w
u
t
w
zz
yz
xz

1
(7.7)

also known as the Cauchy equations.
The above equations may also be written in terms of the velocity components u, v, and w.
By analogy with the Newtons equation of viscosity, the normal stress component in x, y, and z
direction may be written as
x
u
z
w
y
v
x
u
p
v xx
+
|
|
.
|
\
|
+ = 2 (7.8)

y
v
z
w
y
v
x
u
p
v yy
+
|
|
.
|
\
|
+ = 2 (7.9)

z
w
z
w
y
v
x
u
p
v zz
+
|
|
.
|
\
|
+ = 2 (7.10)

where is the second coefficient of viscosity,

3
2
=
v
(7.11)

The tangential stress components have the expressions

|
|
.
|
\
|
=
y
u
x
v
xy
,
|
|
.
|
\
|
=
z
v
y
w
yz
,
|
.
|
\
|
=
x
w
z
u
zx
(7.12)

Inserting all these expressions into the Cauchys equations, and taking into account that =
and
2
2
2
2
2
2
z
u
y
u
x
u
u
= (7.13)

82

we get
|
|
.
|
\
|
+ +
z
w
y
v
x
u
x
u
x
p
X
z
u
w
y
u
v
x
u
u
t
u
3
1
(7.14)

|
|
.
|
\
|
+ +
z
w
y
v
x
u
y
v
y
p
Y
z
v
w
y
v
v
x
v
u
t
v
3
1
(7.15)

|
|
.
|
\
|
+ +
z
w
y
v
x
u
z
w
z
p
Z
z
w
w
y
w
v
x
w
u
t
w
3
1
(7.16)

or
( ) v v f
v
div grad
3
grad
1
+ + = p
dt
d
m
(7.17)

also known as the Navier-Stokes equations. It can be seen that inertial acceleration terms appear on
the left and forces on the right. When inertial forces are small compared to viscous forces, the
nonlinear term can be neglected, leaving the Stokes equation.

v f
v
+ =
p
t
m
grad
1
(7.18)

In both cases, mass conservation requires

( ) ( ) ( )
0 =
z
w
y
v
x
u
t

(7.19)

resulting a system of four nonlinear differential equations with four unknown functions, the velocity
components u = u (x, y ,z, t), v = v (x, y ,z, t), w = w (x, y ,z, t), and pressure p = p (x, y ,z, t).
The initial conditions are

u (x, y ,z, t
0
) = u
0
, v (x, y ,z, t
0
) = v
0
, w (x, y ,z, t
0
) = w
0
, p (x, y ,z, t
0
) = p
0
. (7.20)

and the boundary conditions read as
0
boundary solid
= v . (7.21)

and
0
surface free
p p = (7.22)

Physically, the stress
t
consists of normal and tangential components. Pressure, which is an
isotropic force per unit area n p exerted normal to any surface, is the most familiar normal stress.
Viscous stresses u ~
t
contain both normal and shear components. Additional stresses, both
normal and shear, arise in complex fluids with deformable microstructure: polymer solutions
provide and example. Interfacial (capillary) stresses R
c
~ , where is surface tension and R is
surface curvature, are exerted normal to the free fluid surfaces, whereas surface tension gradients
give Maragoni stresses ~
m
that are exerted along the surface.

83
In some cases of practical interest it is convenient to work with cylindrical coordinates. The
corresponding equations in cylindrical coordinates are:
Continuity equation:
( )
( )
( )
0
1 1
=
z
u
u
r r
r u
r t
z r

(7.23)

Navier-Stokes equations:

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

u
u
r
u
r
p
k u
u
u
r z
u
u
r
u
u
t
u
r r r
r r
z
r
r
r
2
1 1 1
2
2
(7.24)

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

r
r z r
u
u
r
u
p
r
k u u
u
u
r z
u
u
r
u
u
t
u
2
1 1 1 1
2
(7.25)

z z
z z
z
z
r
z
u
z
p
k
u
u
r z
u
u
r
u
u
t
u
+
1 1
(7.26)

where u
r
, u
, and u
z
are the velocity components in r, , and z directions, respectively.

7.2. Laminar flow of inviscid fluids
In the case of inviscid (ideal) fluids, = 0, and the Navier-Stokes equations take a
simplified form which, in the case of an incompressible fluid, can be written as

x
p
X
z
u
w
y
u
v
x
u
u
t
u
1
(7.27)

y
p
Y
z
v
w
y
v
v
x
v
u
t
v
1
(7.28)

z
p
Z
z
w
w
y
w
v
x
w
u
t
w
1
(7.29)

together with the equation of continuity
0 =
z
w
y
v
x
u
(7.30)

Similar initial and boundary conditions apply.

7.3. Exact solution of Navier-Stokes equations for particular flows

7.3.1. Motion between two parallel solid plates
Consider the steady motion of an incompressible fluid of density and viscosity between two
parallel solid plates of infinite extent. The velocity components are u, v, and w in a Cartesian
coordinate system with axes x, y, and z. We consider, however, only the case where the distance
between plates, h, is small enough to induce an unidirectional flow. Thus the only non-vanishing
velocity component is u (see figure). We thus consider the velocity field


84
0 , ) ( = = = w v y f u . (7.31)

The x component of the Navier-Stokes equations then reduces to

2
2
y
u
x
p
, (7.32)

and the y component reads
y
p
1
0 . (7.33)

A consequence of the last equation is that p can only be a function of x. However, since by
assumption the right hand side of the x-component of the Navier-Stokes equations is not a function
of x, neither is the left hand side, i.e. x p is not a function of x. Therefore, x p is a constant
which will be denoted by k. Thus
k
dy
u d
=
2
2
(7.34)

Integrating this equation twice leads us to the general solution

2 1
2
2
) ( C y C y
k
y u + + =
(7.35)

We specialize the general solution to flow through a plane channel whose upper wall moves with
velocity U in the positive x-direction. The function we are looking for, u(y), must satisfy the two
boundary conditions
u(0) = 0, (7.36)
and
u(h) = U, (7.37)

so that we determine the constant of integration as

0 ,
2
2 1
= + = C h
k
h
U
C
. (7.38)

Thus the solution of the boundary value problem is

h
y
h
y
U
h k
h
y
U
y u
|
.
|
\
|
+ = 1
2
) (
2
. (7.39)

For k = 0 we get the simple shearing flow (Couette flow); for U = 0 and k 0 we obtain a parabolic
velocity distribution (two-dimensional Poiseuille flow), while the general case (U 0, k 0) yields
the Couette-Poiseuille flow.
The flow rate per unit depth is
=
h
dy y u Q
0
) ( (7.40)

so that the average velocity defined by the equation h Q U = for the Couette-Poiseuille flow is


85
12 2
2
h k U
U + = . (7.41)

The maximum velocity for pure pressure driven flow is then

U
h k
U
2
3
8
2
max
= =
. (7.42)

Figure 7.2. Motion between two parallel solid plates

7.3.2. Flow down an inclined plane
Closely related to Couette-Poiseuille flow is flow down an inclined plane, although in this case we
deal with a free surface (Figure 7.3).

Figure 7.3. Flow down an inclined plane

Here the volume body force plays the same role as the pressure gradient in Couette-Poiseuille flow.
The flow is not driven by a pressure gradient x p but by volume body force of gravity, whose
components are
sin = g X and cos = g Y . (7.43)

The Navier-Stokes equations read now as

2
2
sin
y
u
g
x
p
(7.44)

and
cos g
y
p
=
(7.45)

86

Therefore we obtain two differential equations for the unknown functions u and p. The no slip
condition

u(0) = 0

is to be satisfied at the bubble wall (y = 0), while the condition at the free surface reads as

h y
y
u
=
= 0

where the effect of the friction in the air was ignored.
Integrating the second Navier-Stokes equation we get

) ( cos x C y g p + =

and with the boundary condition p(y = h) = p
0
we obtain

( ) cos
0
h y g p p + = .

Therefore, p is not a function of x, and the first Navier-Stokes equation simplifies to

2
2
sin
y
u
g
=

This is the same differential equation as that corresponding to Couette-Poiseuille flow, if we replace
x p by sin g . Therefore we read the general solution off from the Couette-Poiseuille flow
(with k = sin g ):

2 1
2
2
sin
) ( C y C y
g
y u + + =

and determine the constants from the boundary condition at y = 0 and y = h as

0 ,
2
sin
2 1
= = C h
g
C

.

The solution of the boundary value problem is therefore

h
y
h
y
h
g
y u
|
.
|
\
|
= 2
2
sin
) (
2

.

7.3.3. Hagen-Poiseuille flow
The flow through a straight circular pipe or Hagen-Poiseuille flow is the most important of all
unidirectional flows and it is the rotationally symmetric counterpart to channel flow. Cylindrical
coordinates are suited to this problem where they describe the wall of the circular pipe by the
coordinate surface r = R (Figure 7.4).


87

Figure 7.4. Flow in a straight circular pipe

At the wall 0 = =

u u
r
, and we set
r
u and
u identically to zero in the whole flow field; moreover

the flow is rotationally symmetric ( ). The continuity equation in cylindrical coordinates then
gives
0 =
z
u
z
or ( ) r u u
z z
=

The r component of the Navier-Stokes equations leads us to

r
p
= 0 p = p(z)

All terms of the Navier-Stokes equation in the direction vanish identically, while the z component
equation becomes
(
=
r
u
r r
u
z
p
z z
1
0
2
2

A close inspection of this equation indicates that z p does not depends on z and therefore the
pressure p is a linear function of z. We set k z p = and write the last equation in the form

(
=
dr
du
r
dr
d
r
k
z
1

which, integrated twice, gives
2 1
2
ln
4
) ( C r C r
k
r u
z
+ + =
.

Since u
z
(0) is finite, C
1
= 0 follows immediately. The no slip condition implies

0 ) ( = R u
z

thus
4
2
2
kR
C = .
Dropping the index z, the solution reads as

( )
2 2
4
) ( r R
k
r u =
.

88

The maximum velocity is reached at r = 0, and therefore we write

(
(
|
.
|
\
|
=
2
max
1 ) (
R
r
U r u .

Figure 7.5. Generalized Hagen-Poiseuille flow

With the flow rate Q through the pipe we introduce the average velocity through the pipe

2
R
Q
A
Q
U
= = ,
and because
4
2 ) (
2
max
2
0 0
R
U drd r r u Q
R

= =

we also find that
max
2
1
U U =
and
8
2
kR
U = .

Since the pressure gradient is constant, we may write

l
p p
l
p
k
2 1

=
=

and mean by p the pressure drop in the pipe over the length l. The pressure drop is positive if the
pressure gradient z p is negative. It is appropriate to represent this pressure drop in a
dimensionless form
2
2
U
p

=

which is called the loss factor. It can be written in the form


89
U d d
l
R U
l

64
16
2 2
= =

where d = 2R and we have set the dimensional quantities into two dimensionless groups l/d
and Re = U d . In particular, in pipe flows the friction factor is often introduced as

l
d
= ,

so that the dimensionless form of the resistance law of a straight circular pipe

Re
64
= .

The Hagen-Poiseuille equation can now be obtained as

l
p R
Q

=
8
4
.

We are led to a generalized Hagen-Poiseuille flow if we subject the general solution u
z
(r) to the
boundary conditions (Figure 7.5)
( ) 0
0
= R u ,
and
( ) U R u
I
= .

The resulting flow is the Couette-Poiseuille flow in a ring gap, and is given by

( )
( )
)
`
|
.
|
\
|
=
0
0 2 2
0
2 2
0
/ ln
/ ln 4
4
) (
R R
R r
k
U
R R r R
k
r u
I
I

For R
0
R
I
= h and R
0
r = y, and in the limit h / R
0
0, the two-dimensional Couette-Poiseuille
flow results. For pure pressure driven flow (U = 0), we find the average velocity

( )
( )
)
`
+ =
0
2 2
0
2 2
0
/ ln
1
8 R R
R R R R
k
U
I
I I

which, for R
I
0, agrees with the Hagen-Poiseuille result.

7.3.4. Flow due to a wall which is suddenly set in motion
In this case the x component of the Navier-Stokes equations reads

2
2
y
u
t
u
.

The no slip condition at the bubble wall furnishes

( ) 0 for , 0 > = t U t u .


90
The second boundary condition is
( ) = y t y u for 0 ,

and in addition we have the initial condition

( ) 0 for 0 , = t t y u .

The x component of the Navier-Stokes equations is a linear equation and since U enters the problem
only linearly from the boundary condition at y = 0, the field u(y, t) must be proportional to U, so
that the solution has to be in the form
( ) , ,t y f
U
u
= .

Since the function on the lefthand side is dimensionless, f must also be dimensionless, which is only
possible if the argument of the function is dimensionless. However, the only linear independent
dimensionless quantity is the combination ( ) t y
2
. We set

t
y
2
1
=

and we are now dealing with a similarity variable , because the solution cannot change if y and t
are changed such that remains constant. We can now write

( ) f
U
u
= ,

and from the equation of motion we obtain an ordinary differential equation

f f = 2

with d df f = . Integrating this equation twice we obtain the general solution

2
0
1
2
C d e C f + =

For y = 0 we have = 0, and the boundary condition at y = 0 becomes f(0) = 1, from which it
follows that C
2
= 0. If we subject the general solution, with C
2
= 0, to the boundary condition at y
,
=
0 1
2
1
d e
C

must hold. The improper integral has the value
2
1
, and therefore

2
1
= C ,

thus the solution reads

91
d e
U
u

=
0
2
2
1 for 0 t .
The integral is
( )
d e

=
0
2
2
erf

is the error function. For t = 0 we have and u/U = 0; thus the initial condition is satisfied.

7.4. Turbulent flow of Newtonian fluids
In turbulent flow all quantities associated with the flow field are random quantities. The
flow may be considered to be the superposition of a basic or main flow with irregular stochastic
fluctuations in the velocity or in other mechanical quantities. The velocity is therefore represented
as follows:
u u u + = .

This decomposition is particularly appropriate if the fluctuation velocity u is much smaller than the
basic velocity u . The basic velocity corresponds to the mean value of velocity. In flows where this
mean value is independent of time, we have the time-mean value calculated from the formula

=
72
2 /
1
lim
T t
T t
T
dt u
T
u ,

which would require only one experimental realization. In what follows we shall restrict ourselves
to incompressible flows which are steady in the mean. We shall now insert the velocity and the
corresponding form for the pressure
p p p + =

in the equation of continuity as well as into the Navier-Stokes equations and shall subject the
resulting equation to the averaging according to the time-mean formula. We have the following
rules for the calculation of the mean values of two arbitrary random quantities a and b:

a a = , b a b a + = +
_______

b a ab =
____
,
s
a
s a
=
_________
/ ,

= ds a ds a
_________

where s is any of the independent variables x, y, z, and t. From the continuity equation we obtain

0 =
z
w
y
v
x
u

since 0 = = = w v u and 0
________ ________ ________
= + + z w y v x u . It thus follows that for the fluctuating
velocities we have

0 =

z
w
y
v
x
u
.


92
For the same reason, all terms linear in the fluctuating quantities vanish from the Navier-Stokes
equations

0 =

t
w
t
v
t
u
, 0
_____ _____ _____
=

z
p
y
p
x
p

0
_____
2
2
_____
2
2
_____
2
2
=

z
u
y
u
x
u
, 0
_______ _______ _______
=
z
u
w
y
u
v
x
u
u

and we obtain the equation

(
|
.
|
\
|

+ |
.
|
\
|

+ |
.
|
\
|

_____ _____ _____

1 1
u w
z
u v
y
u u
x
u
x
p
X
z
u
w
y
u
v
x
u
u
.

By analogy we can obtain the remaining two equations which read as:

(
|
.
|
\
|

+ |
.
|
\
|

+ |
.
|
\
|

_____ _____ _____

1 1
v w
z
v v
y
v u
x
v
y
p
Y
z
v
w
y
v
v
x
v
u
|
.
|
\
|

+ |
.
|
\
|

+ |
.
|
\
|

_____ _____ _____

1 1
w w
z
w v
y
w u
x
w
z
p
Z
z
w
w
y
w
v
x
w
u

together with the equation of continuity

0 =
z
w
y
v
x
u
.

The terms of the form
______
j i
u u are called the Reynolds stresses. It can be seen that these
equations are not enough to determine the basic flow because the Reynolds stresses introduced by
the averaging appear as unknowns.
All attempts so far to make this system of equations determinate have been partly based on
considerable simplifications and hypothesis. At the lowest level, the closure of the system of
equations is accomplished by using relationships between the Reynolds stresses and the mean
velocity field. These semi-empirical relationships represent turbulence models, which can take on
the form algebraic relationships or of differential equations. As the name semi-empirical implies,
they all contain quantities which have to be determined experimentally.
One such model is the Boussinesq formulation of the Reynolds stresses

y
u
A v u
=
______

where A is the turbulent transport coefficient, or / A = is the so-called eddy viscosity. Clearly
the Boussinesq formulation follows the formulation of the viscous shear stress ( ) y u = and
only shifts the problem of the unknown Reynolds stresses to the problem of the unknown eddy
viscosity. The most simple assumption we can make is that A is a constant; however, we cannot do
so in flows bounded by walls since the Reynolds stresses must vanish near solid surfaces. For

93
turbulent flows not bounded by walls, as typically encountered in jets and wakes, the assumption of
constant eddy viscosity can be useful.
Using the concept known as the mixing length, Prandtl found a relationship between the
eddy viscosity and the mean velocity field. The basic idea is that the turbulent stresses arise through
macroscopic momentum exchange in the same manner as the viscous stresses arise from molecular
exchange of momentum. Molecular momentum exchange occurs when a molecule at position y with
a velocity u in the x direction moves to position y l under thermal motion where its velocity is u
du. Therefore, the molecule carries over the velocity difference du = l(du/dy), where l is the
distance in the y direction between two molecular collisions. These motions proceed in both
directions, and therefore momentum is carried over from the faster layer to the slower one. The
number of molecules (per unit volume) moving parallel to the y axis in the y direction is one third
of the total number, and one third moves parallel to the x and z axes, respectively. The molecules
move with thermal velocity v and the mass flux per unit volume is thus v 3 1 . The molecular
momentum flux which manifests itself as the viscous shear stress = (du/dy) is therefore
) / ( 3 1 dy du lv . Although in this extremely simplified derivation, where all molecules have the
same thermal velocity v, they do not affect each other except during collisions and should only
move parallel to the coordinate axes. This formula leads to a very good value for the viscosity of
dilute gases, lv 3 1 = .
In carrying these ideas over to turbulent exchange motion it is assumed that turbulent fluid
parcels (fluid masses which move more or less as a whole) behave like molecules, thus moving over
the distance l unaffected by their surroundings, mixing with their new surroundings and so losing
their identity. The fluctuation u in the velocity is, from the above point of view, proportional to
) / ( dy u d l . The mixing of two turbulent fluid parcels goes along with a displacement of the fluid,
which gives rise to the transverse velocity v whose magnitude is therefore proportional to
) / ( dy u d l . Thus, if we absorb the proportionality factor into the unknown mixing length l, the
turbulent stress becomes
2
2
_____
|
|
.
|
\
|
= =
dy
u d
l v u
t
.

The change of sign in the above equation comes from the fact that a parcel of fluid coming from
above ( v negative) generally carries a positive u with it. If we also take into account the fact that
the sign of
t
is the same as that of dy u d / , we can write down Prandtls mixing length formula

dy
u d
dy
u d
l
t
2
=
and the eddy viscosity
dy
u d
l
2
=

which, from Prandtls mixing model, therefore depends on the velocity gradient.
Although there are typical experiments which clearly contradict the mixing length idea, it is
certainly a model which is very useful and simple to apply, and the predictions of this model
compare favorably to models which take the history of the velocity field into account.
Since the Reynolds stresses must vanish at the wall, we chose l to be proportional to y:

ky l = .
Since the shear stress is constant and thus equal to the wall shear stress (
w
=
2
*
u , we find the
following relation from the mixing length formula:

94
dy
u d
ky u =
,

whose integration leads to the universal velocity distribution, the so-called logarithmic law of the
wall:
C y
k u
u
+ =
ln
1
.

This velocity profile does not hold for y 0, but only to the edge of a layer near the wall which can
be divided into a viscous sublayer and an intermediate layer where the Reynolds stresses decrease
as the wall is approached, while the viscous stresses increase. The velocity at the edge of this layer
therefore depends on the viscosity. The constant in the above equation serves to fit the velocity in
the logarithmic part of the law of the wall to this velocity and so depends on the viscosity too. We
then set the constant C to
+ =
u
k
B C ln
1

and obtain the dimensionally homogeneous form of the logarithmic law of the wall

B
u
y
k u
u
+
|
.
|
\
|
=

ln
1
.

Figure 7.6. Universal velocity distribution

This velocity distribution is met in every turbulent flow near a smooth wall. This equation is valid
in a domain described by the inequality
<< <<
y
u
,

where stands for either the boundary layer thickness, or else half the pipe radius. The constants k
and B are independent of the viscosity and therefore also independent of the Reynolds number
( /
u ). They are absolute constants for flow bounded by a smooth wall and are found
experimentally. Different measurements show a certain amount of scatter in these values, which is
in part explained by the fact that fully turbulent flow was not realized in the experiments, or that the
shear stresses were not constant because of very large pressure gradients. In the region

1000 30

u
y
0
5
10
15
20
25
0 200 400 600 800 1000 1200
yu* /v
u
/
u
*
0
5
10
15
20
25
1 10 100 1000
yu* /v
u
/
u
*

95

we find reasonably good agreement for k 0.4 and B 5. The values given in the literature for k
vary between 0.36 and 0.41, and for B between 4.4 and 5.85.
From measurements, the entire law of the wall max be divided into three regions: viscous
sublayer ( 5 / 0 < <

yu ), intermediate layer ( 30 / 5 < <

yu ), and logarithmic layer ( 30 / >
yu ).
The velocity profile in the viscous sublayer and the logarithmic layer are sketched in figure 7.6.


96
8. BOUNDARY LAYER THEORY

The complete approximate solution to the Navier-Stokes equations may be built up from
two part solutions valid in different regions of the flow field. One of these is the solution of the
inviscid flow problem, the so-called outer solution, and the other is the inner solution close to the
wall. The inner solution describes the boundary layer flow and must such that the flow velocity
from its value zero at the wall passes asymptotically into the velocity predicted by the outer
(inviscid) solution directly at the wall. We will show that the thickness of the boundary layer, i.e.
the layer where the friction effects cannot be ignored, is proportional to Re
-1/2
. The inviscid
solution then represents an approximate solution of the Navier-Stokes equations for large Reynolds
number, with an error of order O(Re
-1/2
).

Figure 8.1. Boundary layer coordinates

Consider an incompressible and plane two-dimensional flow. We introduce the so-called
boundary layer coordinate system, in which x is measured along the surface of the body and y
perpendicular to it. If the boundary layer thickness is small compared to the radius of curvature R
of the wall contour ( / R << 1), the Navier-Stokes equations hold in the same form as in cartesian
coordinates. In the calculation of the inner solution, i.e. of the boundary layer flow, the curvature of
the wall then plays no role. The boundary layer develops as if the wall were flat. The wall
curvature only manifests itself indirectly through the pressure distribution given by the outer
solution.
Since the boundary layer is very thin for large Reynolds numbers, the following inequalities
hold:
y
u
x
u
<<
and
2
2
2
2
y
u
x
u
<<
.

A consequence of the last condition is that the x component of the Navier-Stokes equations reduces
to
2
2
1
y
u
x
p
y
u
v
x
u
u
t
u
.

In order to determine the order of magnitude of the term
x
u
u
in comparison to
y
u
v
, we begin
with the continuity equation for plane two-dimensional and incompressible flow

0 =
y
v
x
u

and together with (8.1) conclude that y u y v << , so that v << u holds. Therefore the second
and third terms on the left hand side in (8.2) are of the same order of magnitude.

97
While the viscous forces are completely ignored in the outer flow, they do play a role in the
boundary layer. The order of magnitude of the boundary layer thickness can be determined by
considering the thickness of the layer where the viscous forces are of the same order of magnitude
as the inertial forces, e.g. where
1 ~
2
2
y
u
x
u
u
.

In the x direction, let L be the typical length scale (Figure 8.1), and if U
is the incident flow

velocity, we have the order of magnitude equation

L
U
x
u
u
2
~

.

The typical length scale in the y direction is the average boundary layer thickness
0
, so that

2
0
2
2
~
U
y
u
.

We then have the estimate

1 ~
2
0
2
U
L
U
,

from which we obtain

2 / 1 0
Re ~

L
.

With this result, the individual terms in the equations of motion are reviewed in order to
systematically simplify the equations themselves. It follows from the continuity equation that

U
L
v
0
~

and therefore
2 / 1
Re ~

U v .

We now introduce the dimensionless quantities, chosen so that they are all of the same order of
magnitude

=
U
u
u ,
2 / 1
0
Re

= =
U
v L
U
v
v
,
2
=
U
p
p

and
L
x
x =
,
2 / 1
0
Re
L
y y
y = =
,
L
U
t t

= .

Using these variables the Navier-Stokes equations take on the form

98

2
2
2
2
Re
1
y
u
x
u
x
p
y
u
v
x
u
u
t
u

and
2
2
2
2
2
Re
1
Re
1
Re
1
y
u
x
u
x
p
y
u
v
x
u
u
t
u

in which all differential expressions have the same order of magnitude, and the order of magnitude
of the whole term is controlled by the prefactor.
Since we are looking for an approximate solution for large Reynolds numbers, we take the
limit Re and obtain the boundary layer equations in dimensionless form:

2
2
y
u
x
p
y
u
v
x
u
u
t
u
,
and
=
y
p
0 .

In addition we have the continuity equation which remained unaffected by taking the limit

0 =
y
v
x
u
.

The dynamic boundary condition at the wall reads

for
y ,
U
U
u .

In the dimensionless boundary layer equations and in the boundary conditions the viscosity does
not appear, and therefore the solution is valid for all Reynolds numbers, as long as they are large
enough.
We shall now rewrite the boundary layer equations in dimensional form and shall restrict
ourselves to steady flow. These were first stated in this form in 1904 by Prandtl:

2
2
1
y
u
x
p
y
u
v
x
u
u
,

y
p
= 0 ,

0 =
y
v
x
u
.

From the second equation of this system of partial differential equations of the parabolic type we
see that ) (x p p = . In the remaining equations u and v are the independent variables, while p is no
longer to be counted as an unknown because the pressure in the boundary layer p(x) has the same

99
value as outside it, where it is known from the outer solution. Therefore the pressure gradient is a
known function, and using Euler equation, can be replaced by

x
U
U
x
p
1
.

We also note that for y only one condition is placed, on the component u. The boundary
conditions reads as

for
0
x x = , ( ) y u u
0
=

Figure 8.2. Growth of boundary layer along a smooth plate

8.1. Friction drag of boundary layer
Figure 8.2 shows the growth of the boundary layer along one side of a smooth plate in
steady flow of an incompressible fluid. Let us consider the control volume shown in Figure 8.3
which extends a distance from the plate, where is the thickness of the boundary layer at a
distance x along the plate. Here we use to indicate the thickness of the boundary layer, usually
defined as the distance from the boundary to the point where the velocity u = 0.99U. In this
analysis, however, we will assume that u = U at the edge of the boundary layer. Along control
surface AB the undisturbed velocity U exists. The pressure forces around the periphery of the
control volume will cancel one another out since the undisturbed flow field pressure must exist
along AB and DA, and the distance BC (=) is so small that it will have a negligible effect on
pressure variations.
Applying the momentum-impulse theorem we get

DA
AB
BC F
x
through entering momentum
through leaving momentum
through leaving momentum drag
+
= =

Since Q
BC
< Q
DA
, there is flow out of the control volume across control surface AB and Q
AB
= Q
DA

Q
BC
.
If the width of the plate is B, neglecting edge effects, the flows and momentums across the
control surface can be expressed as follows:

100

Control surface Flow Momentum
DA UB ( )U UB
BC
0
udy B
0
2
dy u B
AB
0
udy B UB
U udy B UB

0

Figure 8.3. Control volume for flow over one side of a flat plate

Substituting these momentum values in momentum-impulse theorem equation gives

0
) ( dy U u u B F
x

where F
x
is the total friction rag of the plate on the fluid from the leading edge up to x directed to
the left, as shown in Figure 8.3. Equal and opposite to this is the drag of the fluid on the plate.
It will now be assumed that the velocity profiles at various distances along the plate are
similar to each other:
) (
f
y
f
U
u
=
= .

There is experimental evidence that this assumption is valid if there is no pressure gradient along
the surface and if the boundary layer does not change from laminar to turbulent within the region
considered. Then, substituting for u in the last equation and changing the variable y to the
dimensionless , d dy = , and the limits become = to 1, giving

( ) ( ) [ ] d f f BU F
x

=
1
0
2
1 ,

which, for convenience, may be written

2
BU F
x
=


101
where is a function of the boundary layer velocity distribution only and is given by the indicated
integral.
We next investigate the local wall shear stress
0
at distance x from the leading edge. From
the definition of surface resistance, Bdx dF
x 0
= , or

( )
2
0
1 1
BU
dx
d
B dx
dF
B
x
= =

and as all terms in the expression for F
x
are constant except ,

dx
d
U

2
0
= .

This expression for the shear stress is valid for either laminar or turbulent flow in the
boundary layer, but in this form it is not useful until the quantities and dx d are evaluated.

8.2. Laminar boundary layer for incompressible flow along a smooth flat plate
As in the case of laminar flow in pipes, we may examine the shear stress at the plate by aid
of velocity gradient and the definition of viscosity (here we introduce as the dynamic viscosity of
the fluid),
( )
0 0 0
0
= = =

d
df U
d
du
dy
du
y
,

which may be abbreviated to
U
=
0
,

where , like , is a dimensionless function of the velocity distribution curve and is given by the
expression in brackets.
Equating the two last expressions for
0
results in a simple differential equation

dx
U
d

=

with the solution
C x
U
+ =

2
2

where C = 0, since = 0 at x = 0. Therefore,

Re
2 2 x
U
x
= = ,

where xU = Re may be called the local Reynolds number. It should be noted that Re increases
linearly in the downstream direction. Examination of the first expresion of the above equation
shows that the thickness of the laminar boundary layer increases with distance from the leading
edge; thus the shear stress decreases as the layer grows along the plate.

102
To evaluate the last equation, we must know or assume the velocity profile in the laminar
boundary layer. The velocity distribution may be closely represented by a parabola, as shown in
Figure 8.4. In dimensionless terms we have

( )
2
2 = = f
U
u
.

The other velocity profile in Figure 8.4 was derived by Blasius from the fundamental equations of
viscous flow. This curved is based on the thickness being defined for which u = 0.99U.

Figure 8.4. Velocity distribution in laminar boundary layer on a flat plate

The parabolic distribution gives numerical values for and of 0.133 and 2, rspectively.
The Blasius curve yields = 0.135 and = 1.63, the principal difference lying in the milder slope
of the velocity gradient at the wall. With the Blasius values substituted in the expression of we
obtain

Re
91 . 4
Re
1
135 . 0
63 . 1 2
=
=
x
.

Instead of the geometric boundary layer thickness , the uniquely defined displacement thickness
1

is often preferred:
dy
U
u
=
0
1
1 ,

which is a measure of the displacing action of the boundary layer. We obtain

Re
7208 . 1
1
=
x
.

A measure for the loss of momentum in the boundary layer is the momentum thickness
2
:

dy
U
u
U
u
=
0
2
1 ,

for which we obtain the value
Re
664 . 0
1
=
x
.
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
u / U
Blasius
Parabola

103

If the value of from Re 91 . 4 = x is substituted in the expression U =
0
with = 1.63,
there results for the shear stress
Re 322 . 0
0
x
U
= .

If the boundary layer remains laminar over the length of L of the plate, the total friction drag on
one side of the plate is given by integrating this expression:

3
0
2 / 1 3
0
0
664 . 0 322 . 0 LU B dx x U B dx B F
L L
f
= = =

,

and the coefficient of friction c
f
may be obtained as

Re
328 . 1
2
2
= =
L U
F
c
f
f
,

a result which is called Blasius friction law.
The laminar boundary layer will remain laminar if undisturbed, up to a value of Re of about
500,000. In this region the layer becomes turbulent, increasing noticeably in thickness and
displaying a marked change in velocity distribution (Figure 8.5).

Figure 8.5. Laminar and turbulent boundary layers along a smooth flat plate

The thickness of the turbulent boundary layer (which also contains a viscous sublayer of thickness
1
) is
5 / 1
Re
377 . 0
=
x
,

while the coefficient of friction have the expressions

5 / 1
Re
0735 . 0
=
f
c for Re < 10
7
,
and
( )
58 . 2
Re log
455 . 0
=
f
c for Re > 10
7
.


104
8.3. Boundary-layer separation
The motion of a thin stratum of fluid lying wholly inside the boundary layer is determined
by three forces:

1. The forward pull of the outer free-moving fluid transmitted through the laminar boundary
layer by viscous shear and through the turbulent boundary layer by momentum transfer.
2. The viscous retarding effect of the solid boundary which must, by definition, hold the fluid
stratum immediately adjacent to it at rest.
3. The pressure gradient along the boundary. The stratum is accelerated by a pressure gradient
whose pressure decreases in the direction of flow and is retarded by an adverse gradient.

The treatment of fluid resistance in the foregoing sections has been restricted to the drag of the
boundary layer along a smooth flat plate located in an unconfined fluid, that is to say, in the
absence of a pressure gradient.

Figure 8.6. Growth and separation of boundary layer owing to increasing pressure gradient. Note that U has
its maximum value at B and then gets smaller.

In the presence of a favorable pressure gradient the boundary layer is held in place. This is what
occurs in the accelerated flow around the forebody, or upstream portion, of a cylinder, sphere, or
other object. If a particle enters the boundary layer near the forward stagnation point with a low
velocity and high pressure, its velocity will increase as it flows into the lower pressure region along
the side of the body. But there will be some retardation from wall friction so that its total energy
will be reduced by a corresponding conversion into thermal energy.
Consider the flow around a body depicted in Figure 8.6. Let A represent a point in the
region of accelerated flow, with a normal velocity distribution in the boundary layer, while B is the
point where the velocity outside the boundary layer reaches a maximum. Then C, D, and E are
points downstream where the velocity outside the boundary layer decreases, resulting in an
increase in pressure. Thus the velocity of the layer close to the wall is reduced at C and finally
brought up to a stop at D. Now the increasing pressure calls for further retardation. However, this is
impossible, and so the boundary layer actually separates from the wall. At E there is a backflow
next to the wall, driven in the direction of decreasing pressure and feeding fluid into the boundary
layer which has left the wall at D.
Downstream from the point of separation the flow is characterized by irregular turbulent
eddies, formed as the separated boundary layer becomes rolled up in the reversed flow. This
condition generally extends for some distance downstream until the eddies are worn away by
viscous attrition.

105
Because the eddies cannot convert their kinetic energy of rotation into an increase pressure,
the pressure within the wake remains close to that at the separation point. Since this is always less
than the pressure at the forward stagnation point, there results a net pressure difference tending o
move the body with the flow, and this force is the pressure drag.


106
9. FLOW OF COMPRESSIBLE FLUIDS

A compressible flow for which the density change is not more than a few percent may be
treated as incompressible by using an average density for best results. However, if / > 0, the
effects of compressibility must be considered. The solution of a compressible-flow problem is
similar to that of an incompressible one except that the equation of state of the compressible fluid
must be introduced into the problem.

9.1. Definitions
The enthalpy h per unit mass of a fluid is defined by p i h + = , where i, the internal
energy per unit mass due to the kinetic energy of molecular motion and the forces between
molecules, is a function of temperature. The enthalpy per unit weight is p I h + =
.
The specific heat c
p
at constant pressure is defined as the increase in enthalpy per unit mass
when the temperature of a fluid is increased one degree with its pressure held constant. Thus,

p
p
T
h
c
|
.
|
\
|
= (9.1)

where h is the enthalpy per unit mass.
The specific heat c
v
at constant volume is defined as the increase in internal energy per unit
of mass when the temperature is increased one degree with its volume held constant. Thus,

v
v
T
i
c
|
.
|
\
|
= (9.2)

where i is the internal energy.
For perfect gases these equations can be written as dT c dh
p
= and dT c di
v
= . Now since
RT i p i h + = + = , with R the gas constant, RdT di dh + = . Combining these relationships leads
to
R c c
v p
+ = . (9.3)

Introducing the specific heat ratio
v p
c c k = and combining with (9.3) gives

R
k
k
c
p
1
= and
1
=
k
R
c
v
(9.4)

9.2. General equations for compressible fluid flow
Equation of continuity
The expression for continuity equation for one-dimensional flow of a compressible fluid is

constant = = AV G (9.5a)
or
constant = = AV Q (9.5b)

where G and Q are the weight flow rate and mass flow rate, respectively.

Energy equation
For one dimensional flow of a compressible fluid is there is no machine between sections 1 and 2
the energy equation is expressible as

107

g
V
h Q
g
V
h
H
2
2
2
2
2
1
1
+ = + + (9.6)

where the enthalpy per unit weight p i h + =
.

Impulse-momentum equation
The impulse-momentum equation for one-dimensional flow of a compressible fluid is

1 1 1 2 2 2
V V F Q Q = . (9.7)

Euler equation
For one-dimensional compressible flow in a circular pipe the Euler equation may be expressed as

L
h VdV
dp
= +
(9.8)

In this equation and in (9.6) the z terms were dropped, for in the flow of compressible fluids they
are almost always negligible compared with the other terms in the energy equation.

Acoustic velocity
The acoustic velocity kRT E c
v
= = represents the celerity at which a pressure wave will
travel through a compressible fluid.

If heat transfer Q
H
is zero, the flow is adiabatic. Hence (9.6) may be written as

g
V
h
g
V
h
2
2
2
2
2
1
1
+ = + . (9.9)

Since ( ) T g c h
p
=
, we get for adiabatic flow

( ) ( )
2 1 2 1
2
1
2
2
2

2 T T c h h g V V
p
= = . (9.10)

From (9.4), c
p
= kR / (k 1) and for a perfect gas p
v
= RT. Substituting these into (9.10) gives for
adiabatic flow

( ) ( )
2 1 2 2 1 1
2
1
2
2
1
2
1
2
T T R
k
k
v p v p
k
k
V V
= (9.11)

Equation (9.9) can be written as

s
p p p
T
g
c
g
V
T
g
c
g
V
T
g
c
= + = +
2 2
2
2
2
2
1
1
(9.12)

where T
s
is the stagnation temperature (where V is zero). Thus, in adiabatic flow, the stagnation
temperature is constant along a streamline regardless of whether of not the flow is frictionless.


108
Frictionless adiabatic (Q
H
= 0) flow is referred to as isentropic flow. Flow through a nozzle
or flow in a free stream of fluid over a reasonably short distance may be considered isentropic
because there is little heat transfer and fluid-friction effects are small. Equations for isentropic flow
can be derived by substituting
k
pv = constant in (9.11). The resulting equations are

( ) ( )
(
(
|
|
.
|
\
|
=
(
(
|
|
.
|
\
|

1
1
1
1 2
1
2
1
2
2
1
1
2
1
1
2
1
2
2
k k k k
p
p
k
k p
p
p
k
k p
g
V V

(9.13)

9.3. Effect of area variation on one-dimensional compressible flow
In steady flow the velocity of an incompressible fluid varies inversely with the area. This is
not the case with a compressible fluid because variations in density will also influence the velocity.
We shall now examine this phenomenon.

Figure 9.1. Effect of area variation on compressible flow. a) subsonic flow, b) supersonic flow

The continuity equation (9.5b) may be written in differential form as

0 = + +
V
dV d
A
dA
. (9.14)

Noting that d dp c =
2
, the Euler equation for an ideal fluid may be expressed as

0
2
= + = + VdV
d
c VdV
d
d
dp
. (9.15)

Combining the two preceding equations, replacing V/c with Ma, and rearranging we get

109

V
dV
A
dA
) 1 Ma (
2
= . (9.16)

From this equation we arrive at the following conclusions:

1. For subsonic flow (Ma < 1)
If dV/V > 0, dA/A < 0 (area must decrease for increase in velocity)
If dV/V < 0, dA/A > 0 (area must increase for a decrease in velocity)

2. For supersonic flow (Ma > 1)
If dV/V > 0, dA/A > 0 (area must crease if velocity is to increase)
If dV/V < 0, dA/A < 0 (area must decrease if velocity is to decrease)

3. For sonic flow (Ma = 1)
0 =
A
dA
.

Thus it can be seen that subsonic and supersonic flow behave oppositely if there is an area variation.
To accelerate a flow at subsonic velocity, a converging passage is required just as in the case of an
incompressible flow. To accelerated a flow at supersonic velocity, however, a diverging passage is
required. This is so because the decrease in fluid density exceeds the increase in flow velocity;
hence, to satisfy continuity, the passage must diverge.
For sonic velocity it is noted that dA/A = 0. This condition occurs at the throat of a
converging passage. The occurrence of sonic velocity in the throat of the passage requires a high
pressure differential to accelerate the flow the necessary amount to reach sonic velocity. The
velocity at the throat will be a maximum, but will not necessarily be s great as sonic velocity. If
sonic velocity is attained in the throat, the flow will become supersonic if the converging passage is
followed by a diverging one. On the other hand, if the flow in the throat were not sonic, there would
be a decrease in velocity in the following diverging passage. In Figure 9.1 is shown the behaviour
of subsonic and supersonic flow through converging and diverging passages.

9.4. Flow through converging nozzle discharging from a tank
We shall now consider the one-dimensional flow of a compressible fluid through the
converging nozzle of Figure 9.2. We shall assume isentropic conditions. If the velocity of approach
is negligible, equation (9.13) can be expressed as

Figure 9.2. Flow through a converging nozzle


110

( )
(
(
|
|
.
|
\
|
=

1
1 2
1
2
1
2
2
2
2
k k
p
p
k
k p
g
V
. (9.17)

Noting that
2 2 2
p g k c = , the above equation can be rearranged to give

( )
(
(
|
|
.
|
\
|
= =

1
1
2
Ma
1
2
1 2
2
2
2
2
k k
p
p
k c
V
. (9.18)

Thus it is seen that the velocity of flow at the throat depends on the p
1
/p
2
ratio. If there is a large
enough pressure differential between the inside and outside of the tank, sonic velocity will occur at
section 2. Furthermore, it is worth noting here that supersonic flow is impossible in this situation.
We shall now focus on the condition of sonic flow (Ma = 1) at the throat. Substituting Ma =
1 into (9.18) we get the critical pressure ratio

( ) k k
c
k p
p
1
1
2
1
2

|
.
|
\
|
+
=
|
|
.
|
\
|
. (9.19)

This critical pressure ratio exists whenever the flow through the throat is at sonic velocity. If the
flow through the throat is subsonic, then p
2
/p
1
must be larger than this ratio.
The rate of flow through the nozzle may be found by substituting V
2
from (9.17) into
2 2 2
V A G = . Making use of the isentropic relation between pressure and specific weight and
rearranging we get

( )
(
(
|
|
.
|
\
|
|
|
.
|
\
|
=
+ k k k
p
p
p
p
p
k
k
g A G
1
1
2
2
1
2
1 1 2
1
2 . (9.20)

This expression is applicable so long as p
2
/p
1
> (p
2
/p
1
)
c
. To determine the flow through the nozzle
when p
2
/p
1
< (p
2
/p
1
)
c
we substitute ( ) ( ) | |
( ) 1 /
1 2 1 2
1 / 2

+ = =
k k
c
k p p p p into (9.20). The result is

( )
1
1
1
2
1
2
1 2
1 1 2
+
|
.
|
\
|
+
=

k
k
k
p
k
k
g A G
k
. (9.21)

For a perfect gas, this equation may be expressed as

( ) ( ) 1 1
1
1 2
1
2
+
|
.
|
\
|
+
=
k k
k R
k
T
p A
g G . (9.22)

If the flow through is subsonic, the pressure at the throat is identical with that outside the
tank (
2 2
p p = ). If the flow through the throat is sonic, the pressure at the throat may be equal to, but
is generally greater than, the outside the tank (
2 2
p p > ). For air ( 4 . 1 = ) the value
( ) 528 . 0
1 2
=
c
p p .
All these equations are valid for the flow of an ideal fluid. The flows for real fluids through
converging nozzles are only slightly less than those given by the above equations.

111
9.5. One-dimensional shock wave
In Figure 9.3 is shown a one-dimensional shock wave where the approaching supersonic
flow changes to subsonic flow. This phenomenon is accompanied by a sudden rise in pressure,
density, and temperature.

Figure 9.3. A one-dimensional shock wave

Applying the impulse-momentum principle to the fluid in the shock wave we get

( )
= =
1 2 2 2 1 1
V V
g
G
A p A p F
x
. (9.23)

Substituting the continuity equation (
2 2 2 1 1 1
V A V A G = = ) and noting that A
1
= A
2
, we get

( )
2
2 2
2
1 1 1 2
1
V V
g
p p = (9.24)

which is the pressure jump across the wave.
The flow across the shock wave may be considered adiabatic and can be expressed as

( )
2 2 1 1
2
1
2
2
1
2
v p v p
k
k
V V
= . (9.25)

The last two equations may be solved simultaneously and rearranged algebraically to give
some significant relationships. Several such relations are as follows:

( )
1
1 Ma 2
2
1
1
2
+

=
k
k k
p
p
, (9.26)

( )
( )
2
1
2
1
1
2
Ma 1
2 Ma 1
+
+
=
k
k
V
V
, (9.27)
and
( )
( ) 1 2kMa
Ma 1 2
Ma
2
1
2
1 2
1

+
=
k
k
. (9.28)

These equations are applicable only if Ma > 1 (the incoming flow must be supersonic).


112
9.6. Collapse of spherical bubbles in a compressible liquid
We now consider the case of a spherical motion (induced by a bubble collapsing in a liquid
of infinite extent) of a compressible liquid to get a better understanding of some particularities of
the compressible flows of liquids.
Under the assumption of spherical symmetry and in the absence of heat conduction and
mass diffusion, the motion of the liquid is governed by the equation of continuity

0 2 = |
.
|
\
|
+
+
r
u
r
u
Dt
D
, (9.29)
and momentum
r r r
p
Dt
Du
rr
rr

=
2
, (9.30)

where u is the radial component of the velocity field, , the liquid density, p is the pressure in
the liquid,

and ,
rr
are the normal stress components in the and , r directions, and
r u t Dt D + = / / / is the material derivative. When thermal effects in the liquid are
unimportant, the liquid state is completely defined by a simple thermodynamic variable, which
permits the introduction of a speed of sound c and enthalpy h through the definitions:

d
dp
c =
2
,
=
p
p
dp
h
, (9.31a, b)

where
p is the pressure in the undisturbed liquid. For the liquid we take a pressure-density
relationship of the modified Tait form:
n
B p
B p
|
|
.
|
\
|
=
+
+

, (9.32)

where B and n are constants and
denotes the undisturbed value of the density. With (9.32) we

readily find the explicit expressions:

( )
( )h n c
B p n
c 1
2 2
+ =
+
=

, (9.33)

( )
(
(
|
|
.
|
\
|
+
+

1
1 1
/ 1
2 2 2
n n
B p
B p
n
c
n
c c
h , (9.34)

where ( )

+ = /
2
B p n c is the squared of the undisturbed speed of sound in the liquid. Furthermore,
the kinematic boundary condition with no mass transfer at the bubble boundary, ( ) t R is

( ) ( ) R dt dR t t R r u
&
= = = / , , (9.35)

while the condition on the normal stresses stipulates that

( ) ( )
R r rr i
R t p P
=
= / 2 , (9.36)


113
where P is the pressure on the liquid side of the interface,
i
p the bubble internal pressure and the
surface tension.
We now assume that the quantity

(
|
|
.
|
\
|
+
r
rr
rr
dr
r r
u h r

1
2
1
2
(9.37)

is propagated unaltered along the outgoing characteristic ( )dt c u dr + = . The argument leading to
this assumption may be formulated as follows. Since a purely radial motion is evidently irrotational,
we may introduce a velocity potential such that r u = / . The velocity potential of linear
outgoing waves has the form ( ) c r t f r / = and hence the quantity ( ) c r t f t r / / =
propagates along the linear characteristic c dt dr = / . Consider now the nonlinear case. With the
definition of the velocity potential, the momentum equation (9.30) may be integrated once to give

|
|
.
|
\
|
+
+ =
r
rr
rr
dr
r r
u h
t

1
2
1
2
, (9.38)

assuming that 0 / t at infinity. The outgoing characteristic has the form ( )dt c u dr + = , so
that as an extension to the nonlinear case we obtain

0
1
2
1
2
=
|
|
.
|
\
|
+
+
(
r
rr
rr
dr
r r
u h r
r
c
Dt
D

. (9.39)

Using the continuity equation (9.29)
r
u
Dt
Dh
c r
u
2
1
2
=
,

and the momentum equation (9.30) in the form

|
|
.
|
\
|
+
r r Dt
Du
r
h
rr
rr

1

to eliminate the derivatives with respect to r , equation (9.39) may be rewritten as:

0
1 1
1
1
2
3
1 1 1
2
=
|
|
.
|
\
|
+
|
|
.
|
\
|
+
|
.
|
\
|
+
|
.
|
\
|

|
.
|
\
|

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

r
rr
rr
r
rr
rr
dr
r r Dt
D
c
r
dr
r r c
u
c
u
u
c
u
Dt
Du
r
c
u
h
c
u
Dt
Dh
c
r

. (9.40)

When the pressure in the liquid does not deviate too strongly from its undisturbed value
p we may
write, using the definition (9.31) of c ,
|
|
.
|
\
|
+
... 1
1 1
2
c
p p

,


114
and, therefore, for small compressibility effects one can approximately set
in equation
(9.40). At the bubble wall, the material derivative coincides with the ordinary time derivative and
equation (9.40) becomes

|
.
|
\
|
+
|
.
|
\
|
+
|
|
.
|
\
|
+
|
|
.
|
\
|
+
|
|
.
|
\
|
+ =
|
|
.
|
\
|
+
|
|
.
|
\
|
R R
rr rr rr rr
dr
r r dt
d
C
R
dr
r r C
R
H
C
R
C
R
C
R
H
C
R
R
C
R
R R

3 1 3
1
1
1 1 1
2
3
1
2
&
&
& & &
&
&
& &
, (9.41)

whereby H and C denote the values of the quantities h and c at the bubble wall:

( ) | |
2 / 1
2
1 H n c C + =

, (9.42)

( )
( )
( )
(
(
|
|
.
|
\
|
+
+
+
=

1
1
/ 1 n n
B p
B P
n
B p n
H
. (9.43)

In the last step we have used the fact that the incompressible formulation applies at the bubble wall
so that, for a linear viscoelastic liquid, 0 = + +

rr
.
Equation (9.41) represents the Gilmore equation of a spherical bubble situated in a linear
viscoelastic liquid. The effect of liquid rheology is described by the last two terms in (9.41). For a
Newtonian liquid, 0 / 3 / = + r r
rr rr
, while for C the incompressible formulation is
recovered, namely:
dr
r r
H R R R
R
rr rr
|
.
|
\
|
+
= +

3 1
2
3
2
& & &
. (9.44)

By combining (9.41) and (9.44) it is possible to derive the equation:

( ) dr
r r
H R R R R R R
c
R R R
R
rr rr

|
.
|
\
|
+
= + + +

3 1
2 6
1
2
3
3 2 2
& & & & & & & & & &
, (9.45)

if c C and the term in
2
C is dropped. Furthermore, if one writes
( ) ( ) ( )( )
R R R R R R
& & &
2 2 2
1 and uses the incompressible formulation in the form

( ) ( )
dr
r r
H
R
R R
R
R R
R
rr rr
|
.
|
\
|
+
3 1
2
1
4
2
2 2
& &
(9.46)

to evaluate the first term and (9.44) to express the third derivative of the radius which appears on
expanding the second term, one finds

|
.
|
\
|
+
|
.
|
\
|
+
|
|
.
|
\
|
+
+
|
|
.
|
\
|
+ =
|
|
.
|
\
| +
+
|
|
.
|
\
| +
R R
rr rr rr rr
dr
r r dt
d
c
R
dr
r r
R
c
H
c
R
R
c
H R
c
R R
c
R R

3 1 3 1
1
1
1
1
3
1 3
1
2
3 1
1
2
&
& & & & & & &
, (9.47)


115
which represents the general Keller-Herring equation of bubble motion in a linear viscoelastic
liquid. It will be noted that, by dropping terms in
1
c , equation (9.47) reduces to equation (9.44),

which is therefore seen to have an error of the order
1
c . The arbitrary parameter (which does not

seem to have any physical meaning) must, of course, be of order 1 so as not to destroy the order of
accuracy of the approximate equation (9.47). To close the mathematical model, equations (9.44),
(9.45), and (9.47) must be coupled to a constitutive equation for the liquid which in turn relates the
radial stress component,
rr
, to the spatial deformations in the liquid.
Because of the presence of the third time derivative of the radius, the form (9.45) of the
radial equation is hardly more attractive than (9.44) or (9.47), if for nothing else than for the need to
prescribe an initial condition for R
& &
. Actually, this is a minor difficulty since, to the same order of
accuracy in the bubble wall Mach number, an initial condition for R
& &
can be obtained by
substituting the given initial conditions for R and R
&
in the incompressible formulation (16).
However, in view of its uniqueness, it is proper to consider equation (9.45) the fundamental form of
the motion equation of a spherical bubble in a compressible linear viscoelastic liquid.
As an example we consider the three-parameter, linear Oldroyd model as the constitutive
equation of the liquid surrounding the bubble. The rheological equation of this model can be
expressed as:
|
.
|
\
|
+ = +
Dt
De
e
Dt
D
rr
rr
rr
rr 2 1
2
, (9.48)

where
1
is the relaxation time,
2
is the retardation time, , the viscosity, and r u e
rr
= is the
strain rate component. The stress component may be determined as:

( ) ( ) ( ) ( ) ( ) ( ) | |
( )

d
R R r
R R R R R R t
t
rr
)
`
+
+ +
|
|
.
|
\
|
=
3 3 3
2 2
2
2
0 1 1
2
exp
4
& & & &
. (9.49)

We shall make use of dimensionless variables indicated by an asterisk and defined as
follows:
*
2
* *
2
* * 0
, , , ,
rr rr
U C c C H U H Tt t R R R

= = = = = (9.50)

where ( )
2 / 1
/

= p U is of the order of the bubble-wall velocity and U R T /
0
= . With these
definitions (9.41), (9.42), and (9.43) become

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

* *
*
*
*
*
*
* *
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
2
*
*
*
* *
3 3
1
1 1 1
2
3
1
R R
rr rr rr rr
dr
r r dt
d
C
R
dr
r r C
R
H
C
R
C
R
C
R
H
C
R
R
C
R
R R

, (9.51)

( ) | |
2 / 1
*
2
*
1 1 H n C + = , (9.52)

( )
( )
(
(
|
|
.
|
\
|
+
+
1
1
/ 1
2
*
n n
B p
B P
n
H

. (9.53)

where
= c U / is of the order of bubble-wall Mach number. The non-dimensional form of

equation (9.45) reads as:

116

( )
*
*
*
*
*
*
3
* * * * *
2
*
2
* * *
*
3
2 6
2
3
dr
r r
H R R R R R R R R R
R
rr rr
|
|
.
|
\
|
+
= + + +

, (9.54)

while that of equation (9.47) is:

( ) | | ( ) ( ) | |
( ) | |

|
|
.
|
\
|
+
|
|
.
|
\
|
+
+
+ + =
(
+ + +
* *
*
*
*
*
*
*
* *
*
*
*
*
*
* * * * *
2
* * * *
3 3
1 1
1 1 1 3
3
1
2
3
1 1
R R
rr rr rr rr
dr
r r dt
d
R dr
r r
R
H R R H R R R R R

, (9.55)
where
( ) ( ) ( ) ( )
( )
*
*
3
*
3
*
3
*
* *
2
* * * *
2
*
0
* *
*
*
exp
Re
4

d
R R r
R R De R R
De
t
De
t
rr
+
|
.
|
\
|
=

. (9.56)

In the above equations, the following dimensionless parameters have been introduced:
U R

=
0
Re , U R De
0 1
= , and
1 2
= . The first dimensionless parameter is the Reynolds
number and it represents the ratio of inertia and viscous forces. The second one is the Deborah
number defined as the ratio between the relaxation time of the liquid and the characteristic
timescale for the motion of the bubble boundary.

9.7. I mpact of a liquid jet on a rigid boundary
We consider here the impact of a plane-ended liquid mass on a plane solid surface. The
direction of motion of the liquid mass is at right angle to the surface, the plane end being parallel to
the surface. When the liquid strikes this surface a normal shock wave is propagated against the
liquid stream, and behind the shock the liquid velocity is reduced and the pressure increased.
Assuming that the liquid mass is infinitely wide, the problem can be analyzed in one-dimensional
terms and is most simply done in a reference frame moving with the shock (Figure 9.4).

Figure 9.4. Normal-shock configuration for a) unsteady and b) steady reference frame


117
If the velocity of the liquid is initially v, the velocity behind the shock is u and the velocity of the
shock wave is c, which is not necessarily equal to the sound speed, then, in the steady reference
frame, the equation of continuity is

( ) ( ) c u c v + = +
0
, (9.57)

where
0
is the ambient density of the liquid and that behind the shock. If the pressure behind
the shock is p, then by conservation of momentum it can be shown that

( )( ) u v c v p p + =
0 0
, (9.58)

where p
0
is the ambient pressure. Equations (9.57) and (9.58) can be solved provided that the
relation between pressure and density is known. It is usual to assume that the relation for water has
the form

n
B p
B p
|
|
.
|
\
|
=
+
+
0 0

, (9.59)

where n 7 and B 300 MPa. For small values of the change in velocity, v u, the quantity
( ) c v +
0
in (9.58) reduces to the acoustic form
0 0
c , where c
0
is the ambient sound speed. Since
for water and are approximately 1000 kg/m
3
and 1450 m/s respectively, for all but smallest
velocities, i.e. greater than about 10 m/s, the ambient pressure can be neglected in both (9.58) and
(9.59). With this approximation we get

2 / 1
/ 1
0
1 1
(
(
|
|
.
|
\
|
|
.
|
\
|
+ =
n
B
p p
u v
. (9.60)

If the impact is with a rigid surface, then u is zero, and (9.60) gives the impact pressure in terms of
the impact velocity. It can be seen that the pressure generated increases considerably above that
given by the acoustic approximation, i.e. v c
0 0
. We assume that the surface responds as a rigid
solid until a certain compressive stress is reached and then behaves as a perfectly plastic solid, for
which the stress will remain constant, equal to p
Y
, say. In this case (9.60) can be rewritten to give
the velocity deformation u in terms of the impact velocity v and the stress to produce plastic flow,
p
Y
, i.e.
0
v v u = , (9.61a)

where
2 / 1
/ 1
0
1 1
(
(
|
|
.
|
\
|
|
.
|
\
|
+ =
n
Y Y
B
p p
v
. (9.61b)

It is apparent that the impact velocity must exceed a certain critical value v
0
before any plastic
deformation can occur.
Hitherto we have assumed that the liquid mass is infinite in width; but in reality this will not
be the case, and, as soon as the liquid strikes the surface, a release wave will propagate from the
edge of the liquid mass towards the impact centre at the ambient sound speed. Assuming the liquid
mass to be a cylinder of radius a, the pressure at the centre will decrease after a time
0
/ c a from the

118
value given by (9.60) to the stagnation pressure 2
2
0
v , which will be an order of magnitude
smaller unless exceptionally high impact velocities are encountered.

Figure 9.5. Impact of plane-ended liquid mass on a rigid surface

To obtain the amount of deformation induced by jet impact we consider only the centre of
impact, where the only motion will be normal to the surface and the depth of penetration will be a
maximum. If i is assumed that the plastic flow is established immediately, and that it ceases as soon
as the release wave reaches the centre, the time available for deformation will be simply equal to the
time taken for the wave to travel across the radius of the liquid cylinder; i.e.
0
/ c a assuming that it
travels at the ambient liquid sound speed. Since the deformation velocity given by (9.61) is
constant, the depth d of penetration at the center is given simply by the product of the velocity u
from (9.61) and the time available, i.e. by

0
0
c
v v
a
d
= . (9.62)

We shall now consider the normal impact at velocity v of a wedge-shaped liquid mass upon
a plane surface, in which the front face of the wedge is plane but makes an angle with the surface
(Figure 9.6). If the angle is sufficiently small the contact point moves supersonically and an
oblique shock wave forms at the contact point, making an angle to the original surface. For the
wedge-shaped configuration both the velocity of the contact point and the shock and the contact
angles, and respectively, remain constant and therefore the problem can be analyzed as a steady
flow in a reference frame in which the contact point is at rest. In this reference frame which is
moving parallel to the surface at velocity cot v , the liquid approaches the surface at an angle
and with a velocity cosec v . Behind the oblique shock wave, the liquid changes direction and has a
velocity u, which can conveniently be resolved into components normal and parallel to the original
surface,
n
u and
t
u respectively.
The shock problem is analyzed by resolving the velocity upstream of the shock into
components normal and parallel to the shock,
1
n
v and
t
v , i.e.

( ) + = sin cosec
1
v v
n
, ( ) + = cos cosec v v
t
. (9.63)

On passing through the shock, the normal component is changed to
2
n
v , but the transverse
component
t
v is unchanged. The components normal and parallel to the original surface are
therefore given by

119

sin cos
2
t n n
v v u = , cos sin
2
t n t
v v u + = . (9.64)

Figure 9.6. Oblique shock wave configuration for (a) unsteady and (b) steady reference frames

The equations of continuity and conservation of momentum are used to relate the normal velocity
components to the pressure and density behind the shock, i.e.

2 1
0 n n
v v = , (9.65)

( )
2 1 1
0 0 n n n
v v v p p = . (9.66)

Making use of the equation of state (9.59) and the relations (9.63), and also neglecting the ambient
pressure as before, it can be shown that the pressure is given implicitly by

( )
(
(
|
.
|
\
|
+ + =
n
B
p
v p
/ 1
2 2 2
0
1 1 sin cosec . (9.67)


120

Figure 9.7. Pressure generated by impact of wedge-shaped liquid mass at a certain velocity on a rigid surface

Using (9.64), the normal velocity is given by

( )
( )
(
(
|
.
|
\
|
+ + =

tan
tan
1 cos sin cosec
/ 1 n
n
B
p
v u . (9.68)

These equations will in general give either two solutions, corresponding to the weak and strong
shock cases, or no solution. The transition between the two regions gives the locus where the
oblique shock just becomes detached from the contact point. In this case, since the front face of the
liquid wedge is a free surface, a release wave will propagate into the liquid, thus allowing the liquid
to flow laterally and the pressure to decrease to ambient. The condition for shock detachment is
found in general by simultaneously equating the differentials dv and d to zero. Then the boundary
condition of rigid surface, i.e. u
n
= 0, allows an expression to be deduced.
In the case of a rigid surface, the normal velocity component is zero, and (9.68) becomes

( )
n
B
p
(
+
= +

tan
tan
1 . (9.69)

On incorporating this into (9.67), it can be shown that this equation reduces to

|
|
.
|
\
|
+ =
tan
tan
1
2
0
v p . (9.70)

For given values of impact velocity v and wedge angle it is possible to solve (9.69) and (9.70) for
the impact pressure p and the shock angle . Since is of secondary interest, the solution is shown
graphically in Figure 9.7 as the variation of impact pressure with impact velocity v and angle .

121
When is zero these equations are singular, but it can be shown that, if the shock angle tends to
zero, (9.69) and (9.70) reduces to (9.60), i.e. the plane-ended case.
The curves are truncated at shock detachment, showing only the weak-shock solution; this is
taken to be the appropriate one because it agrees with the plane-ended case as tends to zero. The
shock-detachment condition is found by differentiating (9.69) and (9.70) and putting dv and d
equal to zero, from which it may be deduced that

( )
( )
(

+
(
+
= 1
2 sin
2 sin
tan
tan
tan
tan
2
0

n
nB
v
. (9.71)

When the left-hand side of (9.71), which is effectively the square of the impact Mach number,
becomes less than the right-hand side, the shock will become detached.
The locus of shock detachment is approximately linear, and it is found that the value of
) /(
0 0
v c p is always greater than the minimum value of approximately 2.82, which occurs at the
impact velocity of about 150 m/s, corresponding to a Mach number of 0.1. This result indicates that
the impact pressure is given to a good approximation by

v c p
0 0
9 . 2 = , (9.72)

with no more than a 3% error between the impact velocities of 70 and 340 m/s.

Fundamentals of Fluids Mechanics - Brujan

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Fundamentals of Fluids Mechanics - Brujan

Hochgeladen von

Copyright:

Verfügbare Formate

E.A.

Brujan Fundamentals of Fluid Mechanics

= = sin sin A y ydA F

= ma F to the fluid element we get for unsteady flow

= V F m d dt , i.e., impulse equals change of

sufficiently small compared to the circumferential velocity r

, and let there be no singularities

u identically to zero in the whole flow field; moreover

___ _ ___

___ _ ___

___ _ ___

is the incident flow

, we get for adiabatic flow

denotes the undisturbed value of the density. With (9.32) we

c , equation (9.47) reduces to equation (9.44),

c . The arbitrary parameter (which does not

= c U / is of the order of bubble-wall Mach number. The non-dimensional form of

Das könnte Ihnen auch gefallen